Website

Retract refunded and cancelled orders from Google Ads conversions

Confirms the purchase tag sends the order ID, collects refunded and cancelled WooCommerce orders from the last 45 days, builds a retraction and restatement batch with the ROAS impact, and uploads only after approval.

When to use it

When your Google Ads ROAS looks too good because refunded and cancelled WooCommerce orders are still counted as conversions.

What you get

A preview table of retractions and restatements by order ID with the ROAS before and after, uploaded to Google Ads only after you approve.

Does it change anything in my account?

Only with your approval. The assistant first shows a preview of every change and waits for your go-ahead.

Works with

Google Ads, WordPress, Tag Manager

Fill in before sending

[Google Ads account][GTM-XXXX][WordPress site]

Refunds and cancellations are inflating our reported ROAS in [Google Ads account]. Clean them out of the conversion data. 1. Identify the primary purchase conversion action and confirm its settings (counting, lookback window, value). 2. List the tags in GTM container [GTM-XXXX] with their parameters. Check that the Google Ads purchase conversion tag sends the WooCommerce order ID as its transaction ID (orderId parameter). Adjustments only match on that ID. If it is missing or empty, stop and explain the fix instead. 3. Page through the WooCommerce orders on [WordPress site] until you pass 45 days back, and collect every order with status refunded or cancelled, with its date and total. The order list does not show partial refund amounts, because a partially refunded order usually keeps its completed status. So ask me to paste the partial refunds (order ID and remaining order value) from WooCommerce, or skip them if I have none. 4. Build an adjustment batch keyed on order ID. Full refunds and cancellations become RETRACT. Partial refunds become RESTATE to the remaining order value. Note that cancelled orders which never completed payment, and orders that did not come from a Google Ads click, were probably never recorded as conversions and will be rejected. That is expected. 5. Pull 45 days of campaign cost and conversion value for the ROAS before and after. Deliver a preview table with these columns: order ID, order date, original value, adjustment type, new value. Add a total showing how much reported conversion value will drop at most and the estimated ROAS before and after. Upload nothing until I approve the batch. Run the upload as a preview first (confirm=false), then apply it after my approval and report how many adjustments were accepted and how many were rejected, with the reasons the upload returns.

Refunds and cancellations are inflating our reported ROAS in [Google Ads account]. Clean them out of the conversion data.

1. Identify the primary purchase conversion action and confirm its settings (counting, lookback window, value).
2. List the tags in GTM container [GTM-XXXX] with their parameters. Check that the Google Ads purchase conversion tag sends the WooCommerce order ID as its transaction ID (orderId parameter). Adjustments only match on that ID. If it is missing or empty, stop and explain the fix instead.
3. Page through the WooCommerce orders on [WordPress site] until you pass 45 days back, and collect every order with status refunded or cancelled, with its date and total. The order list does not show partial refund amounts, because a partially refunded order usually keeps its completed status. So ask me to paste the partial refunds (order ID and remaining order value) from WooCommerce, or skip them if I have none.
4. Build an adjustment batch keyed on order ID. Full refunds and cancellations become RETRACT. Partial refunds become RESTATE to the remaining order value. Note that cancelled orders which never completed payment, and orders that did not come from a Google Ads click, were probably never recorded as conversions and will be rejected. That is expected.
5. Pull 45 days of campaign cost and conversion value for the ROAS before and after.

Deliver a preview table with these columns: order ID, order date, original value, adjustment type, new value. Add a total showing how much reported conversion value will drop at most and the estimated ROAS before and after.

Upload nothing until I approve the batch. Run the upload as a preview first (confirm=false), then apply it after my approval and report how many adjustments were accepted and how many were rejected, with the reasons the upload returns.

Open in Claude Open in ChatGPT Discuss in the forum

More prompts

WebsiteChanges after your approval

Rewrite meta titles and descriptions for pages with low organic CTR

Finds page-one pages that get impressions but few clicks, compares them to your site-wide CTR, and rewrites their meta title and description around the queries that actually bring them traffic.

When to use itWhen your pages rank on page one and get impressions but few clicks, so your titles and descriptions are probably not convincing.

Use Search Console property [Search Console property] and WordPress site [WordPress site].

Step 1. Baseline. Pull the last 90 days with dimension "date" and add up total clicks and total impressions. Site average CTR = total clicks / total impressions. Show this number first.

Step 2. Candidates. Pull the last 90 days by page, sorted by impressions. Keep pages with at least 500 impressions, an average position between 1 and 10, and a CTR below 2%. Take the top 10 by impressions.

Step 3. For each of those 10 pages:
- Pull the queries for that page only (dimension "query", filtered to the page URL, sorted by impressions) and keep the top 5.
- Find the matching WordPress post or page ID by searching for its slug. If a URL has no clear match (archive, category, home page), mark it and move on.
- Read its current meta title and meta description from the SEO plugin.

Skip any page whose current title already contains its main query AND whose CTR is within 20% of the site average from Step 1. Say why you skipped it.

Deliver a table with these columns: URL, post ID, impressions, CTR, position, main queries, current title, current description, proposed title, proposed description.
- Proposed title: max 60 characters, main query near the start.
- Proposed description: max 155 characters, one concrete benefit and a call to action.
- Use only facts that are on the page. Do not invent prices, guarantees or numbers.

This table is the preview. Do not write any SEO field until I approve. After I approve, write only the approved rows (overwrite the existing values only for those rows), then purge the cache so the new tags go live, and read the SEO fields back for each changed page to confirm they landed.

WebsiteChanges after your approval

Rebuild the site menu around pages that actually convert

Matches GA4 landing-page performance against your WordPress menus, flags valuable pages missing from the navbar, dead weight and broken links, and proposes a leaner main menu.

When to use itWhen your site menu feels cluttered or outdated and you suspect it does not point visitors to the pages that actually bring conversions.

Compare my navigation menu with what actually converts.

From GA4 property [GA4 property], pull the last 90 days by landing page: sessions and key events. Calculate key event rate as key events / sessions.

Then read every menu on [WordPress site], including the main navbar and the footer, with all their links.

Match menu links to GA4 rows by URL path (ignore the domain, trailing slashes and query strings). Where a link has no GA4 row, treat it as zero sessions and say so.

Check every menu link for problems:
- Fetch each linked URL and record its HTTP status. Anything 4xx or 5xx is broken; redirects are worth noting.
- Flag links that point to draft or private pages, to old slugs, or to http instead of https.

Deliver three lists:
1. Pages with at least 20 key events that are not in the main menu.
2. Menu links pointing to pages with fewer than 50 sessions or zero key events in 90 days.
3. Broken or outdated links, with the reason for each.

Then propose a new main menu:
- Max 7 top-level items, ordered by commercial importance.
- Exact labels of 2 to 3 words, in plain language a customer would use.
- The contact page stays as the last item.
- Move low-value pages to the footer menu instead of deleting them.

Show the current menu and the proposed menu side by side, with the exact list of menu actions you would run (add, rename, relink, reorder, remove), as a preview. Do not edit anything until I approve. After approval, apply the changes, purge the cache and read the menus back to confirm the final structure.

WebsiteChanges after your approval

Plugin and tracking health check before it breaks your conversions

A plugin-by-plugin action list that flags outdated, duplicate and conflicting plugins and double-firing tags.

When to use itWhen your WordPress site has too many plugins, updates piling up, or you fear duplicate tags are inflating the conversions your ads optimize toward.

WebsiteChanges after your approval

Replace an old phone number everywhere on the site, safely

A full map of where the old number appears, a dry-run diff of the bulk replace, and a list of places that need manual fixes.

When to use itWhen your phone number has changed and you need to be sure the old one no longer shows anywhere on your website.