Gerrard's Insurance · Google Ads · 3 September 2026

Why Google Ads showed a conversion drop, and why the leads never actually dropped

A plain-English account of what happened between 24 August and 3 September, what it did and didn't affect, what has been fixed, and what happens next. Prepared by Greg Dickson.

1The short version

Leads went up. The counting broke. On 24 August we moved several campaigns onto a new quote page. The new page worked well and produced more leads than the old one. But the small piece of code that tells Google Ads "a lead just happened" was never told about the new page, so it stayed silent there. Google Ads saw the clicks and almost none of the leads, and reported a conversion drop that did not exist in the business.

The gap was found and fixed on 3 September. Every lead reached the brokers as normal throughout. Nothing was lost except ten days of accurate reporting inside Google Ads, and the automated bidding that depends on it.

What Google Ads reported
107 → 101
conversions, 6–19 Aug vs 20 Aug–2 Sep
Leads actually delivered
110 → 196
same two fortnights, from the lead system and the CRM
Leads from the new page
52
24 Aug–3 Sep. Google Ads counted 2 of them.
Fixed
3 Sep
counting resumes from the next lead

2How we know the leads were real

Three separate systems record leads independently of Google Ads: the lead-routing system that sends each enquiry to a broker, the CRM, and the quote pages themselves. All three agree with each other, and all three disagree with Google Ads from 24 August onward.

PeriodGoogle Ads spendConversions Google Ads reportedLeads actually delivered
6 – 19 August (before the change)$10,601107110
20 August – 2 September (after)$14,046101196
Last six days, 28 Aug – 2 Sep3280

Before 24 August the two columns matched almost exactly, day after day. From 25 August they split: on one day the brokers received 28 leads and Google Ads recorded 13; on another, 14 leads against 3.

The clearest proof is which campaigns were affected. Only the campaigns we moved onto the new page show the gap. The campaigns left on their existing pages still match.

Campaign (20 Aug – 3 Sep)Leads deliveredGoogle Ads countedMoved to the new page?
Commercial Motor Vehicle296yes
Commercial Property103yes
Trades52yes
Retargeting30yes
Rural & Lifestyle Block3228no
Contract Works1111no
Landlord77no
Truck1213no

3What actually went wrong

Every quote page carries a small tracking container. Inside it is the instruction that reports a completed enquiry to Google Ads, so Google can attribute the lead to the ad click that produced it. That instruction has a safety list: it only fires on pages whose ID appears on the list. The list is there so that test pages and unrelated forms don't get counted as leads.

When the new quote page went live on 24 August, its ID was not added to that list. The tracking container loaded on the page exactly as it should. The Google Ads instruction inside it simply never fired, because the page wasn't on its list. The same list also controls Microsoft Advertising, so Bing was blind to the page too. Google Analytics is not affected by the list and kept recording normally, which is one of the reasons the gap wasn't obvious at first.

Nothing about the page itself was faulty. The enquiries were completed, sent to the routing system, delivered to brokers and confirmed by email and text as usual.

4Timeline

5What it affected, and what it didn't

Not affected

Affected

One thing that is not a counting problem. The new YouTube prospecting campaign has produced 2 leads for $1,263 on real numbers, not just Google's. Its reporting will now work, but its result so far is genuinely poor and it will be reviewed on its own.

6What has been done

7What happens next

  1. 4 September: confirm Google Ads is counting leads from the new page again, by matching the day's leads in the lead system against Google Ads.
  2. Expect an apparent "recovery" in Google Ads. It is counting resuming, not a change in demand. Read the lead system for the true trend.
  3. Give automated bidding one to two weeks to relearn on the five affected campaigns before judging them.
  4. Close the gap that let this hide. Two additions are proposed: a daily check that compares leads delivered against leads Google Ads counted, per campaign, and raises a flag the moment they diverge; and a pre-flight check whenever an ad is pointed at a new page, so the page is confirmed on the tracking list before it goes live. Either would have caught this on day two rather than day ten.
Technical appendix

Container GTM-KNHW726H (LeadsHook account, container 202519681). Tags Google Ads Conversion Tracking (#137), LH Google Ads Enhanced Conversion Tracking (#114) and LH Microsoft Conversion (#58) fire on custom events whose first condition is {{LH.trigger}} equals true. LH.trigger is a custom JavaScript variable returning {{LH.token}}.includes({{LH.incoming_token}}), where LH.token is a constant listing allowed LeadsHook page tokens and LH.incoming_token is the token embedded in the page HTML.

Old page /e/v/299 token 20b02e57… was on the list. New page /s/SSmXUbD1… token 342454ec… was not. Both pages have Google Ads enabled in their LeadsHook settings with the same $55 value, so the token was the only difference. Version 38 published 3 September adds the token (15 entries). Verified in the served gtm.js.

Sources: Google Ads API (campaign, conversion action, landing page and change-history reports), the lead-distribution database, the GoHighLevel contacts API, and the Tag Manager API live-version read. Full working: clients/gerrards-insurance/reports/2026-09-03-conversion-drop-investigation.md.

8Acknowledge & feedback

Acknowledge this report

Read it and happy with the explanation? Give it the tick. Have questions, or want something looked at further? Note it below. Every acknowledgement and request is recorded.

Sent straight to Greg. One click, no email app needed.