Opus Growth

Reporting

Ready-made prompts for Claude, ChatGPT and other assistants connected to your ad accounts. Copy one, fill in the brackets and send.

Reporting

Monthly paid channel efficiency ranking with a budget split proposal

A ranking of every paid channel by cost per conversion, with spend share vs conversion share and a proposed budget split for next month.

Compare last calendar month across Google Ads, Meta, TikTok and Microsoft Ads.

Build one table with these columns for each channel:
- spend
- conversions
- cost per conversion
- conversion value and ROAS, where value is tracked
- share of total spend (%)
- share of total conversions (%)

Rank the channels by cost per conversion. My target cost per conversion is [target CPA].

Decision rules:
- If a channel's share of conversions is at least 10 points below its share of spend, mark it as a candidate to cut.
- If a channel's share of conversions is at least 10 points above its share of spend, mark it as a candidate to scale. For Google Search, only do this if impression share analysis shows the campaigns are losing impression share to budget.
- If a channel has fewer than 10 conversions, mark it "not enough data" and leave its budget unchanged.

Then propose next month's split in a table with current monthly spend, proposed spend and a one-line reason for each channel. Keep the total within [monthly budget].

Do not apply anything. If I approve, show a preview of each budget change and wait for my confirmation.

Reporting

Month-to-date budget pacing across Google, Meta and TikTok

A pacing table showing which channels will overspend or underspend by month-end, with previewed daily budget corrections.

Check my budget pacing for the current month. My monthly plan is Google Ads [Google monthly budget], Meta [Meta monthly budget] and TikTok [TikTok monthly budget].

1. For Google Ads, use budget pacing to project each campaign's month-end spend against its daily budget.
2. For Meta and TikTok:
   - Take the month-to-date spend and the average daily spend of the last 7 days.
   - Project spend to month-end from those two numbers.
3. Deliver one table with these columns: channel, campaign, planned monthly spend, month-to-date spend, projected month-end spend, variance (%).

Decision rules:
- Flag any channel projected more than 10% over or under plan.
- For each flagged channel, propose the new daily budget that lands it on plan.
- Take the cut from the campaigns with the highest cost per conversion over the last 14 days.
- Add the extra budget to the campaigns with the lowest cost per conversion.

Do not change anything yet. Show a preview of every proposed budget change, with old and new daily amounts, and apply only the ones I confirm.

Reporting

Monthly Google Ads client report with a shareable white-label link

Builds a one-page, jargon-free monthly report for a client with exact calendar-month figures versus the previous month, explains every big KPI move from campaign data and creates the white-label link only after you approve the draft.

Prepare the monthly Google Ads report for [client account] for [month, e.g. 2026-09], compared with the previous calendar month.

Data:
- Use list_accounts to find the customer ID of [client account].
- Pull exact calendar-month figures with get_report (platform google_ads) using start_date and end_date for [month], then again for the previous month. Get both the account totals and the campaign-level rows.
- If you need more campaign detail, use performance_report at campaign level.
- KPIs: cost, clicks, conversions, conversion value, CPA, ROAS and CTR, each with the % change. If conversion value is not tracked in this account, leave ROAS out and say why in one line.

Write it for a client who is not a marketer:
1. A 3-sentence executive summary covering the result, the main driver and what comes next.
2. A KPI table with this month, last month and % change.
3. The top 3 campaigns by conversions. Also name the campaign with the highest CPA among those that took more than 10% of the month's total spend.
4. Three concrete next steps we will take next month, each tied to something in the data.

Flag every KPI that moved more than 15%. Explain each one with evidence from the campaign-level rows, such as a spend shift between campaigns, a CPA change in a specific campaign or a conversion drop. If the data does not explain the move, say so plainly and do not guess.

Keep it to one page with no jargon. Show me the draft first and stop.

After I approve the draft, create the shareable white-label link with client_report_link using brand_name [agency or client brand name]. Note that this tool only takes a rolling days window ending now, not a named month. So before creating it, tell me which exact dates the link will cover with the days value you plan to use, and whether that matches [month]. If it does not match (for example we are well past month end), tell me that clearly and let me choose: create the link anyway with its rolling window, or skip the link and I will send the approved draft myself. Then give me the URL.

Reporting

Agency portfolio triage: which client accounts need attention this week

Compares every connected Google Ads client account over the last 30 days versus the previous 30, adds month-to-date budget pacing, ranks accounts by risk and gives account managers one data-backed first check for the top 5. Read-only.

Triage all connected Google Ads client accounts. Compare the last 30 days with the previous 30 days.

How to get the numbers:
- Run portfolio_report with days=30, then again with days=60. For each account, derive the previous 30 days by subtracting the 30-day figures from the 60-day figures (cost, clicks, conversions, conversion value), then calculate CPA and ROAS for both periods yourself.
- Use list_accounts to get each account's customer ID, then run budget_pacing_alerts with threshold_pct=10 for each account. Pacing is a month-to-date projection to month-end, not the 30-day window, so label that column "pacing (this month)".
- If an account does not track conversion value, mark ROAS as n/a instead of treating it as below target.

Mark an account as at risk if it meets any of these conditions:
- CPA up more than 20%
- Conversions down more than 20%
- ROAS below [target ROAS, or list per-account targets]
- Any campaign flagged by budget pacing as over or under by more than 10% by month-end

Deliver one table with these columns: account, spend, conversions, CPA, ROAS, % change in CPA and conversions, pacing (this month), risk reasons. Sort it by the number of risk reasons, then by spend.

Below the table, list the 5 accounts our account managers should open first this week. For those 5 only, run conversion_tracking_health (days=30) and search_terms_report (days=30, sort by cost) to ground the advice. Give one concrete first check for each, for example "search terms in campaign X are spending without conversions" or "conversion action Y recorded nothing in 30 days", based on what the data shows. If neither report explains the account's problem, say what you would look at next instead of guessing.

Finish with the accounts that improved the most. These are good news for client calls.

This is read-only. Do not change anything in any account.

Reporting

Full-funnel client report: paid, organic and on-site results in one view

A monthly funnel report joining Google Ads, Search Console and GA4, showing which channel actually drove key events.

Reporting

Monthly SEO client report: queries, rankings and page speed

A client-ready SEO report with organic trends, tracked rank movement, page-2 opportunities and Core Web Vitals for key pages.

Prepare the monthly SEO report for Search Console property [property URL]. Compare the last 28 days with the previous 28 days.

Include these sections:
1. Headline numbers: organic clicks, impressions, CTR and average position, each with the % change.
2. The top 10 pages by clicks, with click change and position change.
3. Page-2 opportunities: queries at average position 8 to 20 with more than [impression threshold] impressions. Show query, page, position and impressions. Rank them by impressions.
4. Tracked keywords from the stored rank history: list the movers up or down by 3 or more positions this month. Do not run new paid rank checks.
5. Core Web Vitals for the homepage and the 3 landing pages with the most clicks. Mark each as pass or fail on LCP, INP and CLS.
6. Sitemap status, mentioning only errors or warnings.

Write a 4-bullet summary for the client in plain language. Then list next month's 3 priorities, choosing from the page-2 list and any failing Core Web Vitals.

Read-only. Do not submit or change anything.

Reporting

E-commerce checkout funnel drop-off by step and device

Builds a six-step GA4 checkout funnel for the last 28 days vs the previous 28, split by device, flags weak steps and ends with testable hypotheses for the weakest one.

In GA4 property [GA4 property], build my checkout funnel for these events:
- view_item
- add_to_cart
- begin_checkout
- add_shipping_info
- add_payment_info
- purchase

How to get the numbers:
1. Run ga4_list_conversions so I know which of these events are key events. All six can still be counted as events.
2. Run ga4_report with dimension=date, dimension2=event, days=56 and a high limit (for example 2000). Keep only the six events above, then split the rows into two windows: the last 28 days and the 28 days before that. This gives clean event counts for both periods.
3. Run ga4_report with dimension=event, dimension2=device, once with days=28 and once with days=56. Keep only the six events. The previous-period device counts are the 56-day figure minus the 28-day figure. Say in the output that the device prior period is derived this way.
4. Run ga4_overview with days=28 for the current engagement rate. A clean engagement rate for the previous 28 days is not available, so show the 56-day engagement rate only as a blended reference, clearly labelled, and do not present it as the prior period.

Build a funnel table: step, event count, step-to-step rate, overall rate from view_item, and the change vs the previous period. Add one block per device (desktop, mobile, tablet) with the same columns.

Flag a step in either of these cases:
- its step-to-step rate fell more than 10% (relative) vs the previous period
- the mobile rate at that step is below 60% of the desktop rate

If any of these events is missing or reports zero in either period, say so clearly and do not estimate it. If a step count is higher than the step before it (for example add_payment_info above add_shipping_info), call it out as a likely tracking issue rather than a real funnel result.

End with the single weakest step and three concrete hypotheses I can check on the site, such as shipping cost reveal, payment options or a forced account creation. This is read-only.

Reporting

Audit GA4 key events and their import into Google Ads bidding

A findings table that catches double-counted purchases, micro-conversions used as primary and broken GA4 to Ads links.

Reporting

Find high-traffic landing pages that fail to convert

Ranks landing pages with real traffic but weak purchase rates by lost purchases, adds organic data from Search Console and Core Web Vitals for the worst offenders, and names the three pages to fix first.

Find the landing pages that get traffic but fail to convert. GA4 property: [GA4 property]. Search Console property: [Search Console property]. Site: [https://www.example.com]. This is read-only.

1. List the key events in the GA4 property so we know what counts as a conversion.
2. Pull sessions and key events by landing page for the last 30 days. To isolate purchases, use landing page as the first dimension and event as the second. If that combination is not accepted, use the key events column instead, and if purchase is not the only key event, label the column "all key events" rather than "purchases" for the rest of the report.
3. Keep only pages with at least 300 sessions. Calculate each page's session-to-purchase rate and the site-wide average (total purchases / total sessions for the whole property over the same 30 days). Flag any page whose rate is below 50% of the site average.
4. Search Console: pull clicks, impressions and average position by page for the same 30 days in one call with a high row limit. Match the rows to the GA4 landing pages by path, ignoring the domain, query strings and trailing slashes. If a flagged page has no Search Console row, write "no organic data" rather than zero. Add organic clicks as a share of GA4 sessions, and note that the two tools count differently, so this is a proxy.
5. Core Web Vitals (mobile) for the top 5 flagged pages by sessions: LCP, INP and CLS. Use real-user field data where it exists. If a URL has no field data, show the lab values and mark them as lab.
6. Deliver a table ranked by lost purchases, where lost purchases = sessions × (site average rate − page rate). Columns: URL, sessions, purchases, rate, lost purchases, organic clicks, average position, LCP, INP, CLS, likely issue. Base the likely issue on the data you have: slow vitals, an intent mismatch suggested by position and traffic mix, or a rate gap with healthy vitals that points to the page content or offer.

End with the three pages to fix first and one concrete fix for each.

Reporting

Device purchase rates in GA4 and matching Google Ads bid adjustments

Compare GA4 purchase rates and Google Ads cost per conversion by device, then preview device bid adjustments that match the evidence, with a clear note on how Smart Bidding treats them.

I want to know whether device bid adjustments on Google Ads campaign [campaign] in account [Google Ads account ID] reflect how each device actually buys on the site.

1. GA4 side, property [GA4 property], last 30 days:
- List the key events and confirm the exact event name used for purchases (usually "purchase").
- Pull sessions by device category (ga4_report, dimension "device").
- Pull key events by device split by event name (dimension "device", dimension2 "event") and keep only the purchase event.
- Purchase rate per device = purchase events / sessions. Note that these are site-wide numbers across all channels, not only Google Ads traffic.

2. Google Ads side, same 30 days, campaign [campaign]:
- Pull cost, conversions and conversion value by device with performance_report segmented by device. If that segment is not returned, run a read-only GAQL query on the campaign resource with segments.device and metrics.cost_micros, metrics.conversions, metrics.conversions_value.
- Calculate cost per conversion and ROAS (conversion value / cost) yourself.

3. Current setup:
- Read the campaign's device criteria with get_campaign_targeting. If the bid modifier values are not in that response, read them with GAQL from campaign_criterion (campaign_criterion.device.type, campaign_criterion.bid_modifier) filtered to this campaign.
- Read the campaign's bidding strategy with list_entities (entity campaigns).

4. Build one table with these columns: device, GA4 sessions, GA4 purchase rate, Ads cost, Ads conversions, cost per conversion, ROAS, current bid adjustment. Map GA4 desktop/mobile/tablet to the Ads DESKTOP/MOBILE/TABLET rows, and list CONNECTED_TV separately if it has spend.

5. Decision rules:
- Skip any device with fewer than 20 Ads conversions in the period and say so.
- GA4 purchase rate below 70% of the all-device average AND Ads cost per conversion above 130% of [target CPA]: propose a -20% bid adjustment.
- GA4 purchase rate above 120% of the average AND cost per conversion below [target CPA]: propose +10%.
- If CONNECTED_TV has spend with no conversions, flag it separately and tell me whether excluding it (-100%) makes sense.
- If the campaign uses Smart Bidding (Maximize Conversions, Maximize Conversion Value, Target CPA, Target ROAS), tell me plainly that Google ignores device bid adjustments under these strategies except -100% exclusions, so a -20% or +10% would have no real effect. In that case, recommend what would actually move results instead and do not push the percentage changes.

6. Show every proposed adjustment as a preview first (manage_device_targeting with confirm=false). Do not apply anything until I approve each one.

Reporting

Google Ads service-area leakage report: where clicks really come from

A location report showing spend from people outside your service area, with the targeting setting that causes it and a previewed fix.

Check whether Google Ads account [account name] is paying for clicks outside our service area of [service area]. Use the last 90 days.

1. For each active campaign, read the current location and radius targeting and the location option, either presence or presence-or-interest.
2. Query performance by the user's physical location (user_location_view) and, separately, by targeted location (geographic_view). For each, show clicks, cost, conversions and cost per conversion at city or district level.
3. Resolve any location names you need with the geo target search.

Flag these cases:
- Locations outside the service area with cost above 3% of the campaign total. Mark them "zero conversions" or "converting" accordingly.
- Campaigns set to presence-or-interest. Quantify how much of their spend came from users physically outside the targeted area.

Deliverables:
1. A table per campaign showing in-area vs out-of-area cost, conversions and cost per conversion.
2. The top 15 out-of-area locations sorted by cost.

If a change is justified, such as switching to presence or excluding locations, show a preview of the exact setting change per campaign and wait for my approval.

Reporting

YouTube top videos ranked and classified by audience retention

A ranked list of the top 15 videos of the last 90 days, each tagged as a carrier, a clickbait risk or a hidden gem, with a content recommendation.

For the YouTube channel [channel name], analyze the last 90 days.

1. Get the channel's total watch time and average view duration for the period from the overview.
2. List the top 15 videos by watch time. Use these columns: title, views, watch time (min), average view duration, share of total channel watch time %.
3. Tag each video using these rules:
   - **Carrier:** in the top 5 by watch time, with average view duration above the channel average.
   - **Clickbait risk:** views above the median of the 15, but average view duration below 60% of the channel average.
   - **Hidden gem:** views below the median, but average view duration above 120% of the channel average.
   - **Standard:** everything else.
4. Below the table, write a 3-sentence summary for management:
   - which topics or formats hold attention,
   - which ones attract clicks but lose viewers,
   - what to produce more of next quarter.

Base every statement on the numbers in the table. Do not guess at causes the data cannot show.

Reporting

How viewers find our YouTube videos: traffic source shift report

A breakdown of views and watch time by traffic source for 28 and 90 days, showing share shifts and whether the channel depends on its own promotion.

Show management how people find our videos on the YouTube channel [channel name].

1. Pull traffic sources for the last 28 days and for the last 90 days.
2. Build one table with one row per source type (YouTube search, suggested, browse, Shorts, external, playlists, subscriptions, other). Use these columns:
   - views and watch time (min) for 28 days,
   - share of views % for 28 days,
   - share of views % for 90 days,
   - difference in percentage points.
3. Apply these rules and state the result clearly:
   - **Weak discovery:** YouTube search plus suggested is below 40% of views in the last 28 days.
   - **Promotion-dependent channel:** external sources are above 25% of views. The channel relies on our own promotion, which is not sustainable without spend or social effort.
   - **Format shift:** Shorts gained more than 10 points of share versus the 90-day baseline.
4. Finish with two sentences:
   - the main change,
   - what it means for content planning.

Report only. Do not change anything.

Reporting

What people search before finding our Business Profile: query report

The top Search and Maps queries for a location, split into branded and discovery demand and grouped into themes, with gaps against the listed services.

For our Business Profile location [location name], report which searches led people to the listing over the last 3 months.

1. Pull the search terms people used on Google Search and Maps. List the top 25 by impressions.
2. Split the terms into:
   - **Branded:** contains [brand name] or a common misspelling of it.
   - **Discovery:** everything else.
   Show the share of impressions each group takes. A discovery share below 50% means the profile is mostly found by people who already know us.
3. Group the discovery terms into 4 to 6 themes, such as service type, "near me" intent or product. Give the total impressions for each theme.
4. Read the location details: categories and the services list. For each theme, state whether a matching category or service exists.
5. List themes with real demand but no matching service as **recommendations**. Do not add or edit anything on the profile. I will decide and ask separately.

Output: one table with these columns: theme, impressions, example queries, covered yes or no. Then a 3-line summary for management.

Reporting

Google reviews reputation scorecard for the last 90 days

A management scorecard per location from Google Business Profile reviews: volume, period vs overall rating, star mix, unanswered share and median reply time, plus praise and complaint themes and a reply queue for low-star reviews.

Build a reputation scorecard from our Google reviews for management, covering all locations and the last 90 days. This is a reporting task only.

1. List the locations. For each one, pull the overall average rating and total review count, then page through the reviews with offset and limit until you have every review created in the last 90 days. The list is ordered by last update and a reply counts as an update, so do not stop at the first older review; keep paging until the update dates are older than 90 days as well. If a location's list comes back truncated, say so on its row.
2. For each location, report:
   - number of new reviews in the period,
   - average rating of the period versus the overall average,
   - star distribution (1 to 5) for the period,
   - unanswered reviews in the period, as a count and a share,
   - median days between a review and our reply, using the review date and the reply date, for the period's answered reviews.
3. Mark a location as **At risk** if any of these applies:
   - the period rating is at least 0.3 below its overall average,
   - more than 20% of the period's reviews are unanswered,
   - the median reply time is over 3 days.
   If a location has fewer than 5 reviews in the period, show the figures but note that the sample is small.
4. Across all locations, identify the top 3 praise themes and the top 3 complaint themes from the review text. Back each theme with one short quote and the number of reviews that mention it.
5. Pull the unanswered reviews per location and list every one rated 1 to 3 stars, oldest first, with location, date and text, so the team can respond.

Do not post or edit any replies.

Reporting

Explain a sudden CPA spike in one Google Ads client account

A campaign-by-campaign root-cause table for a CPA jump, with evidence, confidence level and a short client-ready explanation.

Cost per conversion in client account [account name] jumped this week. Find out why.

1. Compare the last 7 days with the previous 7 days by campaign. Cover cost, clicks, CTR, avg. CPC, conversions, conversion rate and CPA. Flag every campaign where CPA rose more than 20% and cost was above [minimum weekly spend].
2. For each flagged campaign, check the likely causes in this order:
   - change history for the last 14 days (bids, budgets, bid strategy, keywords, ads, targeting)
   - conversion tracking health (paused actions, days with zero conversions)
   - impression share lost to budget vs rank
   - new ad disapprovals
   - new search terms with spend and no conversions
3. Classify each problem as either a volume cause (fewer conversions at similar cost) or a cost cause (higher CPC at a similar conversion rate).

Deliverable: one table per flagged campaign. Columns: metric change, most likely cause, evidence (exact numbers or change records with date and user) and confidence (high/medium/low). End with a 3-sentence explanation I can send to the client.

This is read-only. Do not change anything in the account. Any fix you suggest is a proposal only.

Reporting

Monday anomaly scan across all client Google Ads accounts

Compares the last 7 days with the previous 7 across every connected Google Ads account, flags cost, conversion, CPA and ROAS swings, checks pacing and names the campaign behind each issue. Read-only.

Run my weekly anomaly scan across all connected Google Ads client accounts.

1. Run the portfolio report twice: once for the last 7 days and once for the last 14 days. Get the previous 7 days for each account by subtracting the 7-day figures from the 14-day figures (cost, clicks, conversions, conversion value), then recalculate CPA and ROAS for both periods. Skip accounts that spent less than [minimum weekly spend] in both periods, so small accounts do not create noise. Note any account the report skipped as inaccessible.
2. Flag an account if any of these is true:
   - cost changed by more than ±25%
   - conversions fell more than 25%
   - CPA rose more than 25%
   - ROAS fell more than 20% (only for accounts that record conversion value)
   - the account spent money but recorded zero conversions in the last 7 days
3. For every flagged account, run the budget pacing check and say whether it will overspend or underspend this month. If more than 15 accounts are flagged, do this for the 15 with the most spend at risk first and tell me how many are left.
4. For the 5 most severe accounts, take an account snapshot (recent_days=7) and pull a campaign-level performance report for the previous 7 days, so you can name the specific campaign behind the change, not just account totals. Give the single most likely reason in one line.

Deliverable: a ranked table with these columns: account, metric that triggered the flag, change %, pacing status, likely reason, and suggested next check. Rank by spend at risk. Below the table, list the accounts that are stable in one line.

Read-only scan. Do not change anything.

Reporting

Diagnose a Meta CPA rise: auction, creative fatigue or conversion issue

A breakdown of a Meta campaign's rising cost per result into auction, creative, conversion or self-inflicted causes, with ad-level fatigue flags.

Cost per result in campaign [campaign name] of Meta ad account [ad account] has been rising. Explain why.

1. Compare the last 14 days with the previous 14 days, using daily data. Cover spend, CPM, link CTR, CPC, frequency, results, conversion rate and cost per result.
2. Break the same metrics down by placement and by age/gender. Show which segment drove most of the increase.
3. At ad level, compare each ad's CTR in its first 7 days with its last 7 days. Flag fatigue when frequency is above 3 and CTR has dropped more than 30%.
4. Pull the change history for the last 21 days. List edits that can reset learning: budget changes over 20%, new or paused ads, targeting or optimization changes.
5. Classify the main cause as one of these:
   - auction: CPM up, CTR stable
   - creative: CTR down, frequency up
   - conversion: CTR stable, conversion rate down, so check the landing page or pixel
   - self-inflicted: a recent edit lines up with the drop

Deliverable: a short verdict, a table of evidence per cause, and a fatigued-ads list with each ad's creative summary.

Read-only. Do not pause or edit anything.

Reporting

Monthly product-level ROAS report: Shopping and PMax vs store stock

Pulls last month's product-level results for Shopping and Performance Max, joins them with WooCommerce SKU, price and stock, and sorts products into stars, zombies, bleeders and out-of-stock spenders. Report only.

Build last month's product-level ROAS report for our Shopping and Performance Max campaigns in Google Ads account [customer ID], with stock from our WooCommerce store [site URL].

1. Run list_accounts to confirm the account. Then run google_ads_query on shopping_performance_view with segments.date DURING LAST_MONTH, selecting segments.product_item_id, segments.product_title, campaign.name, campaign.advertising_channel_type, metrics.impressions, metrics.clicks, metrics.cost_micros, metrics.conversions and metrics.conversions_value. Calculate ROAS yourself as conversion value / cost. Raise the limit so no products are cut off, and tell me if the result was still truncated.
2. As a sanity check, compare the total cost from step 1 with campaign-level cost from get_report for the same month. If the gap is more than 5%, say so before going further.
3. Run list_wordpress_sites and wordpress_site_info to confirm the site and that WooCommerce is active. Then page through wordpress_woo_products (per_page=100, page 1, 2, 3...) until you have the full catalog with name, SKU, price and stock.
4. Join the two lists by matching item ID to SKU. Item IDs are often prefixed or use the WooCommerce product ID instead of the SKU (for example "woocommerce_gpf_123" or "gla_123"). If fewer than 80% match, show me 5 unmatched examples from each side and ask how to map them before continuing. Treat a product as out of stock if its stock field shows out of stock or a quantity of 0. Note that stock is today's value, not last month's.

Sort products into buckets:
- Stars: ROAS at or above [target ROAS] with cost above [minimum product spend].
- Zombies: more than 1,000 impressions but fewer than 10 clicks.
- Bleeders: cost above the minimum spend with zero conversions, or ROAS below half of target.
- Out of stock but spending: any cost while WooCommerce currently shows the product as out of stock.

Show the top 20 of each bucket by cost, the share of total Shopping/PMax spend going to bleeders and out-of-stock items, and a separate list of products with spend that could not be matched to a SKU. Finish with recommendations such as feed fixes, price checks, separate campaigns for stars or listing group exclusions. This is report only. Change nothing.

Reporting

Data-confidence check before sending the client lead report

Audits Google Ads lead conversion actions for broken tracking, wrong counting, short lookback windows and double counting, then reconciles them with GA4 key events from google / cpc traffic. It returns a reliability table and previews any fix before applying it.

Before I send the monthly report for [account name], check whether the lead numbers can be trusted. Cover the last 30 days.

1. Run conversion_tracking_health with days=30 and list every lead action that is paused, removed or has recorded zero conversions.
2. Use list_conversion_actions and get_conversion_settings to check every action that is included in the Conversions column. Use included_in_conversions as the primary flag, and use list_conversion_goals to see which categories are biddable. Form and call leads should count one per click, not every. Flag lead actions counted as every, lookback windows under 30 days and default values that do not match [average lead value].
3. Look for double counting. Check list_conversion_actions for the same form tracked both by a Google tag (webpage) action and by a GA4-imported action, with both included in conversions. Confirm it with the tag setup where possible:
   - Use get_conversion_tags to get the conversion ID and label of each Google tag action.
   - If Tag Manager is connected, use list_gtm_tags and list_gtm_triggers to see which Google Ads and GA4 tags fire on the same form trigger.
   - If Tag Manager is not connected, base the finding on action names, types and categories, and mark it "inferred, not confirmed".
4. Find the GA4 property with ga4_list_properties and the key events with ga4_list_conversions. Run ga4_report with dimension='event' and dimension2='source_medium' for 30 days, and keep only the google / cpc rows for the matching lead events. Compare them with the Google Ads lead conversions and flag any gap larger than 25%. Note that GA4 and Google Ads attribute and date conversions differently, so a small gap is expected.
5. Deliver a table with these columns: metric, source, status (reliable / check / unreliable), reason, and the fix.

Do not change any settings. If a fix is needed, run update_conversion_action (primary or secondary) or set_conversion_settings (counting, lookback window, default value) with confirm=false, show me the preview and wait for my approval. Pausing or removing a conversion action cannot be done through the API, so list those as manual steps in the Google Ads interface.

Reporting

Quarterly paid + organic funnel report for local service pages

Combines Google Ads, Search Console, GA4 and Business Profile data into a 90-day vs previous 90-day funnel. It shows the lead share by source, local query quick wins, service pages losing organic clicks and an owner summary with 3 SEO actions.

Build a quarterly funnel report for the client. Compare the last 90 days with the previous 90 days.

1. Confirm the Google Ads account with list_accounts. Run marketing_overview with days=90 and check its scope block to make sure the right Ads account, Search Console property and GA4 property are covered. For the previous 90 days, use get_report for google_ads, search_console and ga4 with start_date and end_date. Call describe if you are unsure of a platform's levels. Show one table with Google Ads spend, clicks and conversions, Search Console clicks, impressions and average position, and GA4 sessions and key events, for both periods with the % change.
2. Run ga4_report with dimension='source_medium' for days=90 and for days=180. Derive the previous 90 days by subtracting the first from the second. Tell me what share of key events came from paid search (google / cpc), organic search, Maps / Business Profile referrals and direct traffic. Business Profile traffic is only separable if its links carry UTM tags, for example utm_source=gbp. If they do not, it is counted in google / organic, so say that and recommend tagging the profile links. If Business Profile is connected, add website clicks and calls from gbp_insights as a separate line.
3. Run gsc_performance with dimension='query', sort='impressions', days=90 and a high enough limit. From the results, take the 10 service or location queries with the most impressions, such as "[service] [city]" and "near me" variants. For each, show clicks, CTR and position. Mark queries at positions 4 to 15 with more than 200 impressions as quick wins.
4. Run gsc_performance with dimension='page' for days=90 and days=180, and derive the previous 90 days by subtraction. Rank the service pages by organic clicks in the last 90 days and flag pages whose clicks fell compared with the previous 90 days.
5. Write a one-paragraph owner summary that explains how organic and paid work together, where we depend too much on paid, and the 3 SEO actions for next quarter.

This is a read-only report. Do not change anything.

Reporting

"Work done this month" log for Google Ads and Meta, with before/after impact

Turns the last 30 days of Google Ads and Meta change history into a client-ready log grouped by category. It measures 14-day before/after impact for the 5 most important changes and flags edits made by people outside the team.

Clients ask what they are paying for. Build the "work done this month" section of the report for Google Ads [Google Ads account] and Meta [Meta ad account].

1. Pull the last 30 days of changes with get_change_history (days=30) and meta_change_history (since and until covering the same 30 days). For each change, show the date, the user, the entity, the field and the old and new value. Meta does not return old and new values for every event, so write "not provided" where they are missing.
2. Group the changes into these categories: budgets, bidding, keywords and negative keywords, ads and creatives, targeting, conversion tracking, and campaign status. Merge repetitive edits, so 40 negative keywords become one line.
3. Pick the 5 most important changes, judged by the spend of the affected campaign and how much the change can move results. Explain briefly why you picked each one. For each, compare leads, cost per lead and conversion rate for the 14 days before and the 14 days after the change, at the level of the affected campaign or ad set. Use get_report with start_date and end_date for Google, and meta_campaign_insights or meta_insights_query with since/until for Meta.
   - If fewer than 14 days have passed, label the result "too early to judge".
   - If another important change hit the same campaign within the same window, say the impact cannot be attributed to one change.
   - Treat the result as an indication, not proof, because seasonality and auction conditions also move these numbers.
4. Flag any change made by someone who is not on this list: [team emails and Meta user names]. Google shows the user's email, while Meta shows the person's name, so match each platform on its own field.
5. Write each line in client language, for example "Blocked 38 irrelevant searches such as 'free' and 'jobs' to stop wasted spend."

This is read-only. Do not modify anything.

Reporting

Click fraud section for the monthly client report

Scans the last 30 days of ad-click traffic for suspicious IPs, checks which are already excluded, estimates the exposure as a labelled upper bound, writes a measured client summary and previews an approval-gated blocking list.

Local service ads often get clicked repeatedly by competitors and bots. Build the click-fraud section of this month's report for [account name].

Known IPs that belong to the client or to us (office, VPN, agency): [IP list, or "none known"]

0. Use list_accounts to find the customer id for [account name].

1. Run fraud_scan for the last 30 days with block_threshold=70.
   - If it reports that the fraud tag is not installed or has captured no ad-click traffic, stop and tell me. Without the tag there is nothing to report, so do not present zeros as a clean result.
   - Otherwise, list every IP with a score of 70 or more, with its score and reasons.
   - If the scan output includes the number of ad clicks and the campaign for each IP, show them. If it does not, mark those columns "not available from the tag" instead of guessing.

2. Run list_ip_exclusions for the account. Mark which of these IPs are already blocked, and on which campaign.

3. Pull cost, clicks and average CPC per campaign for the same 30 days with get_report (platform google_ads, level campaign). Estimate the exposure:
   - Formula: suspicious clicks from IPs not yet blocked x the average CPC of the campaign they clicked. Use the account average CPC where the campaign is unknown.
   - Show it as an amount and as a % of total spend.
   - Label it as a rough upper bound, not a measured loss. Google already filters and refunds part of invalid traffic, and average CPC is not what each individual click cost.

4. For the client, write 3 sentences covering how much suspicious activity we found, what it may have cost at most and what we are doing about it. Do not overstate. A high score is a signal, not proof.

5. Prepare a blocking list of IPs scored 85 or more that are not yet excluded, grouped by campaign. Get the campaign ids with list_entities.
   - Leave out every IP from my known list above.
   - Move to a separate "needs your check" list any IP whose reasons are only volume or repeat clicks, with no datacenter, Tor, known-malicious or bot user-agent signal. These are often real people on shared mobile carrier or office connections.
   - The tools cannot confirm who owns an IP range, so say plainly that this split is a judgement, not a verification.

6. For each campaign, show block_ip_addresses with confirm=false as a preview (at most 200 IPs per call, 500 per campaign). Do not run it with confirm=true until I approve.

Reporting

Next-month budget proposal: lost leads on Google and the Bing option

Combines pacing, impression share lost to budget and rank, and cost per lead to propose next month's Google Ads budgets with a capped, labelled lead estimate, and compares Bing if a Microsoft Ads account is connected.

Write the "next month's budget" section of the client report for Google Ads [account name]. If a Microsoft Ads account is connected, include it.

0. Use list_accounts to find the customer id. Use account_snapshot (platform google_ads) to get each campaign's current daily budget and status.

1. Run budget_pacing_alerts with threshold_pct=10. List the campaigns projected to overspend or underspend this month by more than 10%.

2. Run impression_share_analysis for the last 30 days at campaign level. For each Search campaign, show impression share, the share lost to budget and the share lost to rank.

3. Pull cost and conversions per campaign for the same 30 days with get_report (platform google_ads, level campaign) and calculate cost per lead.

4. For campaigns losing more than 15% to budget with a cost per lead below [target cost per lead], estimate the extra leads a budget increase could bring:
   - Assume cost per lead stays flat.
   - Cap the estimate at the volume the lost-budget share represents: current leads x (share lost to budget / impression share). More budget cannot buy impressions that do not exist.
   - Label the result as an estimate. In practice cost per lead usually rises as budget grows, so treat it as optimistic.

5. For campaigns losing mostly to rank, say clearly that more budget will not help. Run impression_share_analysis at keyword level and keyword_report for those campaigns, then name the likely cause (low Quality Score, a weak component, or low bids) and label it as a hypothesis based on that data.

6. If list_microsoft_accounts returns an account, run microsoft_performance_report for the same 30 days. Compare Bing's cost per lead and lead volume with Google's. If Bing shows no conversions, check whether that means no conversion tracking rather than no leads, and say which.

7. Deliver a table with these columns: campaign, current daily budget, proposed daily budget, expected extra leads (estimate) and reasoning. Keep each proposed increase within +50% of the current budget, since larger single changes are blocked. Add a one-line total monthly figure for the client.

Do not change any budget. If I approve a line, show the set_campaign_budget preview with confirm=false first and wait for my confirmation.

Reporting

Device and country conversion gaps in GA4

Finds the devices and countries that convert well below the site average over 90 days, sizes the missed key events and checks whether the gap survives within the same channel before calling it a real problem.

Using GA4 property [GA4 property], analyse the last 90 days to find where conversion performance breaks down by device and by country.

0. Find the property with ga4_list_properties. Check ga4_list_conversions. If there are no key events, stop and tell me, since rates would be meaningless.

1. With ga4_report and days=90, pull sessions, users and key events by device, then the same metrics by country (raise limit so smaller countries are included). For countries, keep only those with at least 300 sessions.

2. Calculate the key event rate for each row. Take the site-wide average from ga4_overview with days=90.

3. Flag every row whose key event rate is at least 30% below the site average.

4. For each flagged row, estimate the missed key events if that row converted at the site average. Use this formula: sessions x (site rate - row rate).

5. Check the traffic mix before drawing conclusions. Run ga4_report with days=90, dimension device and dimension2 channel, and the same for country. For each flagged row:
   - If the gap holds within the same channel (for example mobile paid search vs desktop paid search), call it a likely device or country issue.
   - If the gap mostly disappears within channels, say it comes from a weaker traffic mix in that segment, not from the segment itself.

Deliverable:
- a table sorted by missed key events, with these columns: segment, sessions, key events, key event rate, gap vs average, missed key events, gap holds within channel (yes/no)
- three short conclusions a manager can act on, for example "mobile checkout needs a UX review"

Do not suggest causes the data cannot support. If mobile and desktop rates are within 10% of each other, say the device gap is not material.

Reporting

Paid vs organic Google report: where am I paying for clicks I already get?

A 28-day report that matches Search Console queries to Google Ads search terms and shows where paid and organic overlap.

Build a paid vs organic report for the last 28 days and compare it with the previous 28 days.

1. From GA4, give me sessions and key events for paid search vs organic search, with the % change for each.
2. Match Search Console queries to Google Ads search terms. List every query where both of these are true:
   - our average organic position is 3 or better
   - we also paid for clicks on it
3. For each match, show this table: query, organic position, organic clicks, paid clicks, paid cost and paid conversions.
4. Split the table into brand queries (containing [brand name]) and non-brand queries.

Decision rules:
- Mark a brand query as a "test candidate" when its organic position is 2 or better and paid conversions are low relative to cost.
- Mark a non-brand query as a "page-2 opportunity" when it ranks 8 to 20 organically and converts well in paid. These are queries worth supporting with content.

Close with the estimated monthly spend tied to the test candidates.

This is read-only: no keyword or budget changes.

Reporting

Monthly local business report: Business Profile, reviews and Google Ads

A 30-day local report that puts Business Profile actions and Google Ads results for your city side by side, flags local searches your ads miss, and drafts replies to unhappy reviews for your approval.

Prepare my monthly local report for the last 30 days, compared with the 30 days before that. Use the Business Profile location for [business location].

1. Business Profile actions:
   - Business Profile insights only return totals for the last N days, so pull them twice: once for 30 days and once for 60 days. The previous period is the 60-day total minus the 30-day total.
   - Show search and Maps impressions, phone calls, direction requests and website clicks for both periods, with the % change for each.
   - Ignore the messages metric, it is no longer a live feature. If bookings is 0, mention it as a real gap.
2. Google Ads in the same two windows (use exact start and end dates that match the profile windows as closely as possible):
   - Find the campaigns that target [city] by checking each campaign's location targeting. If none target it explicitly, tell me which campaigns you used and why.
   - Show clicks, conversions and cost for those campaigns. For calls from ads, run a read-only query on campaign phone call metrics (phone calls, phone impressions). If the account has no call reporting or the numbers come back empty, say so instead of showing 0 as a result.
   - Put the ad numbers next to the profile numbers in one "customer actions" table so I see organic and paid side by side, with the % change for each row.
3. Local searches:
   - List the top 15 searches people used to find the profile. Note that this data lags about 2 months and that low-volume terms are only reported as "fewer than N".
   - Compare them with the active keywords in my Google Ads account and mark any search with no close matching keyword.
4. Reviews:
   - From the review list, count the reviews created in the last 30 days, and report the average rating and how many are still unanswered.
   - For unanswered reviews of 3 stars or below, draft short, specific replies. Show each one as a preview only (confirm=false) and do not publish anything until I approve each reply.

Finish with three actions for next month, ranked by likely impact on calls.

Reporting

Paid social client report: Meta and TikTok side by side

A combined Meta and TikTok monthly report with comparable KPIs, creative fatigue flags and a budget split recommendation.

Reporting

"What we did this month" work log for the client, with measured impact

A plain-language log of the agency's Google Ads and Meta changes, grouped by theme and linked to before/after results.

Reporting

Meta and TikTok click-to-session loss and UTM gaps in GA4

Compares Meta and TikTok link clicks with the sessions GA4 actually records, flags where traffic and attribution leak, audits destination URLs for missing or inconsistent UTMs, and ends with a paid social UTM standard.

Check how much Meta and TikTok traffic reaches GA4 and how well it is tagged. Meta ad account: [Meta ad account]. TikTok ad account: [TikTok ad account]. GA4 property: [GA4 property]. Period: the last 14 days. This is read-only.

1. Ad platforms: pull spend and link clicks per campaign from Meta and spend and clicks per campaign from TikTok.
2. GA4: list the key events first. Then pull sessions and key events by source / medium for the same 14 days, and separately by landing page with source / medium as the second dimension. If purchase is not the only key event, label the column "all key events" instead of "purchases".
3. Platform-level view: group the Meta sources (facebook, instagram, fb, ig, meta and similar) and the TikTok sources. Build a table: platform, link clicks, GA4 sessions, click-to-session ratio, GA4 purchases, spend per GA4 purchase. Flag a platform below 60%.
4. Campaign-level view: GA4 here gives source / medium and landing page, not session campaign. Match each ad campaign to GA4 through the landing page its ads point to. Where several campaigns share a landing page, report them together and say so. If I paste a GA4 export by session campaign, use that for an exact campaign match instead. Mark every match as exact, via landing page, or unmatched. Apply the same 60% flag, and flag any campaign with spend and no GA4 match.
5. UTM audit:
   - Meta: for each flagged campaign, list its active ads and read their creatives. Check the destination URL and any URL parameters set on the creative. List every ad with missing utm_source, utm_medium or utm_campaign, or with values that do not follow one consistent pattern.
   - TikTok: list the ads in the flagged campaigns. If the ad data includes the landing page URL, audit it the same way. If it does not, tell me and ask me to paste the TikTok destination URLs.
6. Unattributed traffic: report GA4 sessions arriving as (not set), (direct) / (none) or unknown sources during the period, and estimate how much paid social spend likely sits behind them, based on the click gap in step 3. Label it as an estimate.

End with a UTM naming standard I can apply to all paid social ads (source, medium, campaign, content values and casing rules). Do not edit any ad.

Reporting

Google Ads calls vs forms: lead conversion deep-dive by campaign

A tracking-health check plus a campaign table that separates phone calls from form leads and flags campaigns over target CPA.

Build a lead deep-dive for Google Ads account [account name]. Compare the last 30 days with the previous 30 days. Do not change anything.

Step 1, tracking check: list all conversion actions with category, status, primary or secondary, counting type and click lookback window. Flag any action that is paused, recorded zero conversions in the last 30 days, or counts "every" on a call or lead form action, because leads should count "one".

Step 2, campaign report: for each campaign, show cost, clicks, conversions split by conversion action (calls from ads, website calls, form submissions, lead form), total conversions, cost per conversion, conversion rate and the % change from the prior period.

Decision rule: mark a campaign RED if cost per conversion is more than 30% above [target CPA] and it has at least 50 clicks. Mark it AMBER if it spent more than 2x [target CPA] with zero conversions.

Deliverables:
1. A tracking-health table.
2. A campaign table sorted by cost, with a RED/AMBER/OK column.
3. Three short findings explaining where calls and form leads actually come from.

Reporting

When do calls actually come in? Google Ads hour, day and device report

A day-by-hour table of cost and calls compared with your current ad schedule, plus a previewed schedule proposal.

Reporting

Meta lead form quality report: cost per qualified lead by ad set

Ad-set level cost per lead combined with a quality check of the actual instant form leads, to find cheap but useless sources.

Analyse lead quality in Meta ad account [account name] for the last 30 days. This is read-only.

1. List the instant forms on our Page and the questions each one asks.
2. Pull spend, reach, frequency, CTR, leads and cost per lead per campaign, ad set and ad.
3. Download the leads for the same period, with the campaign, ad set and ad each lead came from.

Score each lead as qualified only if all three are true:
- The phone number looks valid for our country (correct length and prefix).
- It is not a duplicate of another lead by phone or email.
- The qualifying answer matches [qualifying answer, e.g. "within our service area"].

Deliverable: a table per ad set showing spend, leads, cost per lead, qualified rate % and cost per qualified lead, sorted by cost per qualified lead.

Flag these cases:
- Ad sets whose cost per lead is below the account average but whose qualified rate is under 40%.
- Ad sets with frequency above 3.

Show counts and percentages only. Do not print names, phone numbers or emails in the answer.

Reporting

TikTok lead gen deep-dive: creative fatigue and lead completeness

Week-by-week TikTok ad performance with view rates, cost per lead and a check of how complete the collected leads are.

Report on TikTok ad account [account name] for the last 28 days, split into 4 weekly buckets. Read-only.

1. Per ad, pull spend, impressions, CTR, 2-second and 6-second video view rates, leads (form submissions) and cost per lead, for each week.
2. Break down cost per lead by age group and placement.
3. List our lead forms and their field schema. Download the leads for the same period and map each answer to its field label.

Flag these cases:
- Creative fatigue: any ad whose CTR fell 30% or more from its best week to the latest week, while spend stayed flat or rose.
- Weak hook: any ad with a 6-second view rate under 15%.
- Incomplete leads: the share of leads missing a phone number or with an empty required question, per form.

Deliverables:
1. A weekly table per ad with a "fatigued" column.
2. An age and placement table sorted by cost per lead.
3. A lead completeness summary.
4. A short list of ads to refresh, ranked by spend.

Do not pause or edit anything. If you recommend pausing an ad, show it as a proposal for my approval.

Reporting

Microsoft Ads local deep-dive: locations, devices and out-of-area searches

A Bing performance breakdown by location, device and keyword, plus search terms that leak spend outside your service area.

Deep-dive Microsoft Advertising account [account name] for the last 60 days. We only serve [service area, e.g. city and districts].

1. Break down performance by location, by device and by keyword. For each, show impressions, clicks, CTR, spend, conversions and cost per conversion.
2. Pull the search terms report with the matched keyword and match type.
3. Read the current negative keywords at campaign and ad group level so you do not suggest duplicates.

Flag these cases:
- Locations outside our service area that received clicks, with their total spend.
- Search terms containing other city names, DIY or job-seeking intent ("how to", "salary", "course"), or competitor brands, with spend and zero conversions.
- Devices whose cost per conversion is 40%+ above the account average, with at least 30 clicks.

Deliverables:
1. A location table.
2. A device table.
3. A top-20 keyword table sorted by spend.
4. A proposed negative keyword list grouped by theme, each with match type and the spend it would have saved.

Show the negatives as a preview and wait for my approval before adding them.

Reporting

ChatGPT Ads delivery and lead tracking report for a local service

Daily ChatGPT Ads delivery and conversions, a pixel health check and a cost per lead comparison against your Google Ads benchmark.

Report on our ChatGPT Ads account for the last 30 days. Read-only.

1. Tracking: show our pixels, the conversion event settings and which campaigns each one is attached to, and whether any events arrived in the last 15 minutes. Flag any campaign that has no lead or call event attached.
2. Delivery: pull daily impressions, clicks, CTR, spend, conversions and cost per conversion per campaign and ad group. Include zero-impression rows, and add device and platform breakdowns.
3. Limits: show the account spending limits and spend so far, and say whether a limit is capping delivery.
4. If any campaign or ad group had 3 or more consecutive days with zero impressions, run the delivery diagnosis and list the ranked findings with evidence.

Benchmark: our Google Ads cost per lead is [Google Ads CPL]. Mark each campaign as:
- KEEP if cost per conversion is within 20% of that figure.
- TEST MORE if it has fewer than 10 conversions.
- REVIEW if it is more than 50% above the benchmark.

Deliverable: a tracking table, a campaign table with these labels, and three action recommendations. Do not change any campaign, bid or limit.

Reporting

Monthly YouTube channel report for management (28 days vs previous 28)

A paste-ready table comparing the last 28 days with the previous 28 on views, watch time and subscribers, plus three plain-language takeaways.

Reporting

Google Ads device breakdown: where mobile, desktop and tablet waste budget

A campaign-by-device table showing which device drives conversions, with flagged waste and previewed bid adjustments where they actually apply.

In my Google Ads account [account name], break down the last 30 days by device (mobile, desktop, tablet) for every active Search and Performance Max campaign.

Build one table with one row per campaign and device. Include these columns: cost, share of campaign spend, clicks, CTR, conversions, conversion rate and cost per conversion.

Flag two cases:
- A device whose cost per conversion is more than 30% above that campaign's average, and which has spent at least [minimum spend amount].
- A device that has spent money but recorded zero conversions.

Next, read each Search campaign's current device bid adjustments and its bidding strategy. Only campaigns on Manual CPC can use the device bid adjustments you suggest. Keep each suggestion between -50% and +30%. Smart Bidding campaigns (Maximize Conversions, Target CPA, Target ROAS) ignore device adjustments except -100%. For those, report the numbers only and tell me whether excluding a device entirely would be justified.

End with a 3-line summary of the biggest device-level opportunity. Show any proposed adjustment as a preview and wait for my approval before applying it.

Reporting

Meta age and gender breakdown: who actually buys from your ads

An age-by-gender table for one Meta campaign showing spend, results and cost per result, with the best and worst segments called out.

For my Meta campaign [campaign name], break down the last 30 days by age and gender.

1. List the campaign's ad sets with their age range, gender and Advantage+ audience settings. This tells me whether Meta was allowed to go beyond my targeting.

2. Pull insights with the age and gender breakdown at campaign level. Show these columns:
- spend and share of spend
- reach and frequency
- link click-through rate
- results, using the campaign's optimisation event (purchase or lead)
- cost per result

3. Flag two kinds of segments:
- Segments that took at least 10% of spend but whose cost per result is more than 40% above the campaign average.
- Segments whose cost per result is below the average and that received less than 10% of spend.

4. Note any segment with fewer than 5 results, where the numbers are too small to act on.

Deliver the table sorted by spend, followed by 3 plain-language takeaways. For example: "women 35-44 bring half of your leads at the lowest cost". This is a report only. Do not edit any ad set.

Reporting

Meta placement report: Feed, Reels, Stories or Audience Network?

A placement-level table of cost and results that exposes cheap-click placements that do not convert, with a previewed placement change.

Show me where my Meta ads actually run and which placements earn their money. Use account [ad account name] and the last 30 days.

1. Pull insights broken down by publisher platform and platform position. Examples: Facebook Feed, Instagram Feed, Instagram Stories, Instagram Reels, Facebook Reels, Messenger and Audience Network.

2. For each placement, show these metrics:
- spend and share of spend
- impressions and CPM
- link clicks and link click-through rate
- results and cost per result

3. Flag two warning signs:
- Placements with a link click-through rate above the account average but a conversion rate less than half of the average. This pattern often means accidental clicks, typical of Audience Network.
- Placements that took more than 10% of spend with zero results.

4. List which ad sets use Advantage+ placements and which use manual placements.

Recommend specific placement exclusions only where the evidence is clear. Mention the trade-off: fewer placements usually means a higher CPM.

Show any change to ad set placements as a preview and wait for my approval before updating anything.

Reporting

Website device check: is mobile traffic converting as well as desktop?

A GA4 and Search Console device comparison with mobile Core Web Vitals, ending in the three most likely causes of a mobile conversion gap.

Most of my ad and organic traffic is mobile. Check whether my website converts mobile visitors as well as desktop visitors.

1. In GA4, break down the last 28 days by device category (mobile, desktop, tablet). For each device, show sessions, engagement rate, key events and key event rate per session. Add the same figures for the previous 28 days and show the change.

2. In Search Console, for property [site URL], break down the last 28 days by device. Show clicks, impressions, CTR and average position.

3. Run a Core Web Vitals check on [main landing page URL] and report the mobile LCP, INP and CLS, each marked as good, needs improvement or poor.

Decision rule: if the mobile key event rate is below half of the desktop rate, treat it as a mobile problem rather than an audience difference. If the mobile engagement rate dropped by more than 10 points versus the previous period, call it out.

Deliver one comparison table, then the three most likely causes of any gap, ranked by evidence. Typical causes are a slow page, a form that is hard to use on a phone, or a ranking gap on mobile. This is a report only. Do not change anything on the site.

Reporting

Monthly cross-channel budget reallocation plan based on ROAS

A 30-day ROAS-based plan that shifts budget between Google Ads, Meta and TikTok at a fixed total, with scale and cut rules, month-to-date pacing and previewed budget changes only after approval.

Prepare next month's budget reallocation plan from the last 30 days of data across Google Ads, Meta and TikTok. Keep the total at [total monthly budget]. Only the split changes.

First find where each budget actually lives: on Meta, a campaign with its own daily budget (Advantage+ campaign budget) is changed at campaign level, otherwise at ad set level. On TikTok, check whether the budget sits on the campaign or the ad group. Meta list budgets come back in cents, so convert them to the main currency unit before comparing.

Scale candidates:
- Google Search campaigns with ROAS above [target ROAS] and more than 10% search impression share lost to budget.
- Meta and TikTok campaigns or ad sets / ad groups above target ROAS with at least 20 conversions in the period. For Meta, also require frequency under 3. If TikTok returns no purchase value for this account, judge TikTok on CPA against [target CPA] instead and say so.

Cut candidates: anything 30% or more below target ROAS with spend above [minimum spend].

Pacing for the current month:
- Google Ads: run budget pacing and note campaigns set to over- or underspend.
- Meta and TikTok: there is no pacing report, so calculate it yourself: month-to-date spend divided by (current daily budget × days elapsed). Flag anything below 80% or above 110%.

Output a table: channel, campaign or ad set / ad group, current daily budget, proposed daily budget, change %, ROAS, reason. Cap any single change at 20% so learning phases are not reset. Show the net change per channel and confirm the proposed total still equals [total monthly budget] divided by the days in the month.

Do not apply anything. When I approve specific lines, show a preview of each budget change (dry run, nothing written) and wait for my explicit confirmation before writing it.

Reporting

Weekly social spend vs branded search demand: halo effect check

Lines up 12 Monday-to-Sunday weeks of Meta and TikTok spend against branded Search Console demand and the Google Ads brand campaign to see whether social spend moves branded search, without claiming causation.

I want to see whether our Meta and TikTok spend drives branded search. Use the last 12 full weeks, bucketed Monday to Sunday, and state the exact date range at the top.

Per week, pull:
- Meta spend, impressions and reach at account level, requested in 7-day increments starting on a Monday so each row is one Monday-to-Sunday week (weekly reach cannot be built by adding days).
- TikTok spend and impressions by day, summed into the same weeks.
- Search Console clicks and impressions for queries containing [brand name] on [Search Console property]. First list the queries over the period and pick out the common misspellings and variants of the brand that actually appear, then pull a daily series for the brand term and each variant and sum them into weeks. Search Console data lags 2 to 3 days, so mark the latest week as possibly incomplete.
- Google Ads impressions, clicks, conversions and conversion value for the campaign [brand campaign name], by week (Google Ads weeks start on Monday). Also check its search terms for brand misspellings worth adding to the Search Console list.
- Non-brand Google Ads conversion value (all other campaigns), for context.

Output:
1. One weekly table with all of the above.
2. The weeks where combined social spend rose more than 20% week over week and whether branded impressions rose in the same week or the following one.
3. A simple directional read: do branded impressions move with social spend, lag it by one week, or show no visible link?

Note confounders you can see in the data, such as a jump in non-brand Google Ads or a seasonal pattern, and use my list of promotion weeks: [promotion or sale weeks, if any]. Do not claim causation. If the pattern is clear, tell me what a proper test would look like, for example a geo holdout on Meta. Report only. Change nothing.

Reporting

Monthly GA4 summary for management with channel breakdown

Compares the last 30 days with the previous 30 in GA4, breaks sessions and key events down by channel, flags big moves with hypotheses and explains what counts as a conversion, in plain language for a director.

Prepare the monthly GA4 summary for management for property [GA4 property]. Use the last 30 days and compare them with the previous 30 days.

0. Find the property with ga4_list_properties.

1. Run ga4_list_conversions first. If the property has no key events configured, say so at the top of the report instead of reporting zeros, and skip the key event columns below.

2. KPIs for the current period: run ga4_overview with days=30 for sessions, users, new users, engagement rate and key events.

3. KPIs for the previous period: run get_report with platform ga4, the property as account, and start_date / end_date covering the 30 days before the current window.
   - If it does not return a metric, fall back to ga4_overview with days=60 and subtract the 30-day figures. This only works for additive metrics (sessions, new users, key events).
   - Users and engagement rate cannot be derived by subtraction. If you cannot get them for the previous window, show the current value and mark the change "not comparable" rather than estimating it.
   - Show each metric's absolute and % change.

4. Break sessions and key events down by channel with ga4_report (dimension channel), once with days=30 and once with days=60 so you can derive the previous 30 days by subtraction. For each channel, show:
   - its share of sessions
   - its share of key events
   - its key event rate (key events / sessions)
   Note that key events here are the sum of all key events in the property.

5. Flag every metric that moved more than 15% in either direction. Name the most likely driver from the channel breakdown and label it as a hypothesis.

6. List the events currently marked as key events, with their counting method, so readers know what "conversion" means in this report.

Deliverable:
- one KPI table
- one channel table
- at most 5 bullet points written for a non-technical director, with no jargon

Reporting

E-commerce funnel drop-off from GA4 event counts

Builds a six-step purchase funnel from GA4 users per event, splits it by device, finds the biggest leak with a clear decision rule and sizes the purchases a 10% improvement there would add.

Build a purchase funnel for GA4 property [GA4 property] for the last 30 days, using users and event counts by event name.

0. Find the property with ga4_list_properties.

1. Run ga4_report with days=30 and dimension event, with a limit high enough to return every event name. From the rows, keep only these funnel steps, in order:
   1. session_start
   2. view_item
   3. add_to_cart
   4. begin_checkout
   5. add_payment_info
   6. purchase
   If an event is missing or has zero counts, say so and skip that step. Do not invent a value for it. Use ga4_overview with days=30 to sanity-check the session_start users against total users.

2. Deliverable 1: a table with these columns:
   - step
   - users
   - event count
   - step-to-step rate (based on users)
   - cumulative rate from step 1

3. Deliverable 2: the same funnel split by device. Run ga4_report with dimension event and dimension2 device. If the tool does not return that combination, say so and skip the device split rather than estimating it.

Decision rule:
- The biggest leak is the step with the lowest step-to-step rate that also loses at least 500 users.
- Highlight that step and state how many extra purchases a 10% relative improvement there would produce. Carry the extra users through the remaining steps at their current rates.

Note clearly that this is an event-count funnel, not a sequential path analysis, so step counts can include users who skipped earlier steps.

Reporting

Audit which GA4 events count as key events before reporting

A check of GA4 key event definitions against real event data and GTM tags, with any key event change previewed for approval.

Before I send conversion numbers to management, verify that GA4 property [GA4 property] counts the right things.

1. List every event marked as a key event, with its counting method.
2. Pull event counts by event name for the last 30 days. Flag each of the following:
   - key events with zero events in the period
   - key events that look like low-value actions, such as page_view, scroll or click
   - events named like real business outcomes (purchase, generate_lead, form_submit, phone_click) that are not marked as key events
3. In GTM container [GTM container], list the GA4 event tags and their triggers.
4. Flag any business event that is sent by more than one tag, or that fires on a trigger broader than its purpose. These are likely double counting.

Deliverable:
- a table with these columns: event, key event yes/no, 30-day count, GTM tag(s), verdict (keep / remove from key events / add as key event / check tag)

If you recommend marking a new event as a key event, show me exactly what will change and wait for my approval. Do not edit anything in GA4 or GTM yourself.

Reporting

Does YouTube drive website visits and conversions? GA4 check

Connects organic YouTube views to GA4 sessions and key events from YouTube referrals, compares the key event rate to the site average and labels YouTube as a qualified traffic source or an awareness channel.

Management wants to know whether our organic YouTube channel sends people to the website. Use the last 30 days and the GA4 property [GA4 property name].

1. List the YouTube channels and pick ours. If there are several, ask me which one. Get total views and watch time for the period.
2. List the GA4 key events so we know what "key events" includes. If there is more than one, note that the key event count is the sum of all of them.
3. Pull a GA4 report by source / medium for the last 30 days with sessions and key events. Use a high enough limit to catch small sources. Keep only YouTube rows: sources containing youtube.com, m.youtube.com, youtu.be or youtube. Sum them into one YouTube line and show the individual rows underneath.
4. If I name a primary key event ([primary key event, optional]), also break the YouTube sources down by event so that event can be counted on its own.
5. From the GA4 overview for the same period, get site-wide sessions, key events and engagement rate. Calculate the site-wide key event rate as key events per session.
6. Build a funnel table with these rows:
   - YouTube views,
   - sessions from YouTube,
   - view-to-session rate %,
   - key events from YouTube,
   - YouTube key event rate (key events per session),
   - site-wide key event rate and YouTube as a % of it.
   Show the site-wide engagement rate as context below the table. GA4 does not give engaged sessions per source here, so do not estimate engagement for YouTube.
7. Verdict rule:
   - If YouTube's key event rate is at least 70% of the site average, call it a qualified traffic source.
   - If it is below that, call it an awareness channel only.
   - If YouTube sent fewer than 30 sessions, say the sample is too small for a verdict.

Note that traffic from paid YouTube campaigns usually appears as google / cpc, so this covers organic YouTube only. Read-only. Make no changes in any account.

Reporting

Weekly cross-channel ROAS scorecard with underperformer flags

A Monday-to-Sunday scorecard across Google Ads, Meta, TikTok, Microsoft Ads and ChatGPT Ads with week-over-week changes, a comparability note per platform and a flag list of campaigns below target ROAS.

Build my weekly paid media scorecard for last week (Monday to Sunday) compared with the week before. Use campaign-level data from Google Ads, Meta, TikTok, Microsoft Ads and ChatGPT Ads, and pull both weeks with exact start and end dates so every platform covers the same days. Use the connected default account on each platform unless I name one here: [account names, if you run more than one per platform].

Data notes:
- Microsoft Ads: use a campaign performance report that includes the Revenue column, so ROAS can be calculated, not just conversions.
- TikTok: request purchase value along with spend and conversions. If TikTok returns no value metric for this account, show conversions and CPA and mark ROAS as n/a.
- ChatGPT Ads: pull campaign-level insights with conversions included. If no conversion value comes back, show spend, conversions, CPA and CTR and mark ROAS as n/a. Do not estimate value.

1. Channel summary table: spend, conversions, conversion value, ROAS, CPA, CTR, plus the week-over-week % change for spend and ROAS.
2. Campaign table under each channel, sorted by spend, with the same columns.
3. Comparability note for each platform:
   - Which conversion counts as a purchase: Google Ads from the conversion actions included in the Conversions column, Microsoft from its conversion goals, TikTok from the pixel's configured events, ChatGPT Ads from its conversion event settings, Meta from the purchase action type in the report.
   - Attribution window: Google Ads from the conversion settings (click and view-through lookback), ChatGPT Ads uses OpenAI's fixed 30-day window. Meta, TikTok and Microsoft do not expose this through the connection, so use what I give you: [Meta attribution setting, e.g. 7-day click 1-day view], [TikTok attribution window], [Microsoft conversion window]. If I left any blank, write "unknown" rather than guessing.

Flag two groups in a separate list:
- Campaigns with ROAS below [target ROAS, e.g. 4.0] and weekly spend above [minimum weekly spend].
- Campaigns whose ROAS fell more than 25% week over week, regardless of spend.
Campaigns with ROAS n/a go in neither group; list them under "cannot be judged on ROAS" with their CPA instead.

End with three lines: blended ROAS across all channels that report conversion value (name any channel left out), the biggest winner of the week and the biggest problem, each with the number behind it. This is a report only. Do not change anything in any account.

Reporting

Monthly Google Business Profile report: calls, searches, reviews

An owner-ready report on how the local listing drives calls and directions, which searches find it, and draft replies to weak reviews.

Create a monthly Google Business Profile report for [location name] that I can send to the owner. Compare the last full month with the month before.

1. From gbp_insights, report Search and Maps impressions, phone calls, direction requests, website clicks and bookings. Show the absolute numbers and the % change for each.
2. From gbp_search_keywords, list the top 15 search terms that showed the listing. Split them into branded searches (the business name) and non-branded service searches, and give me the non-branded share. Non-branded searches are the new customers.
3. From gbp_reviews, report new reviews this month, the average rating and every review that is still unanswered. For each unanswered review of 3 stars or less, draft a short, calm reply that does not argue with the customer.
4. Summarize in 3 bullets what drove calls this month. Then name one concrete action for next month, such as a post, a service list update or new photos.

Show me the reply drafts first. Do not publish any reply until I approve each one.

Reporting

LinkedIn lead gen report for B2B local services: messaging and lead fit

Scores LinkedIn creatives on cost per on-target lead, using real form answers for job title and company location, compares the copy of the best and worst performers, and flags weak forms and off-target creatives.

We sell [service, e.g. commercial cleaning] to businesses in [city/region]. Analyse LinkedIn ad account [account name] for the last 60 days. Read-only.

1. Structure: list the campaigns as I see them in Campaign Manager (campaign groups), their ad sets and the creatives.
2. Performance per ad set and per creative: impressions, clicks, CTR, spend, Lead Gen Form opens, leads, form completion rate (leads / form opens) and cost per lead. If form opens are not returned for the account, say so and leave the completion rate blank rather than estimating it.
3. Read the actual ad copy of every creative with at least 1,000 impressions.
4. List the Lead Gen Forms and download the leads for the same 60 days with their answers. If the download reports missing Lead Sync access, tell me and stop at the performance part.

Count a lead as on-target only if both are true:
- The job title or function matches facility, operations, office or general management, or owner.
- The company location is inside [city/region].

If a form does not ask for job title or company location, mark those leads "unknown" rather than on-target or off-target, and recommend adding the question.

Attribution: if the lead records carry the creative, calculate the on-target rate per creative. If they only carry the campaign or ad set, calculate it at that level and apply it to its creatives only where the ad set runs a single creative. Say clearly which level you used.

Deliverables:
1. A creative table: spend, CTR, form completion rate, leads, cost per lead, on-target rate and cost per on-target lead, sorted by cost per on-target lead.
2. A side-by-side comparison of the copy of the best and worst 3 creatives. Point out the differences in hook, offer and call to action.
3. A list of creatives (or ad sets) with a form completion rate below 10% or an on-target rate below 50%.

Show lead counts only, no personal data. Do not change any campaign or creative.

These prompts work best when your AI assistant is connected to your accounts with Opus Growth.

Connect your accounts free