Skip to content

Agency PlaybookPublished 12 min read

Conversion Tracking Audit Checklist for Agencies

The full cross-platform checklist for taking over a client account: GTM, Google Ads, Meta, GA4, CRM, consent, real test leads and a severity-scored report.

Muhammad Qitmeer, founder of Augere Labs
Muhammad Qitmeer
Co-founder and CEO, Augere Labs
Share

A conversion tracking audit for a new client account checks every place a conversion is defined or sent: the GTM container, Google Ads conversion actions, Meta events and deduplication, GA4 key events, CRM source capture and consent. Then it proves what actually happens by submitting real test leads on every form and comparing platform counts with the CRM. Each finding gets a severity score so the client sees what to fix first.

Taking over an account without auditing tracking means inheriting someone else's mistakes and being blamed for them a month later when numbers shift. A structured audit in week one sets a baseline, gives you a documented reason for every change, and often finds the reason the previous agency's results looked the way they did. This checklist is the order we work in. Adapt it, but do not skip the test submissions.

1. Access you need

Ask for the least access each check requires, in the client's own accounts:

SystemAccess for an auditAccess for fixes
Google Tag ManagerRead on the containerEdit or Publish on the container
GA4Viewer or AnalystEditor
Google AdsRead onlyStandard
Meta BusinessView access to the dataset (pixel) and ad accountManage access to the dataset
CRM (HubSpot, GoHighLevel, other)A user who can view contacts, forms and propertiesA user who can edit properties and forms
Website CMS or storeOften not needed for an auditNeeded if forms or code change

Also ask for: a list of every form, calendar and checkout on the site; which phone numbers are used in ads; and the CRM's definition of a qualified lead and a won deal.

2. GTM container review

  • Is there exactly one GTM container on the site, loaded once? Check the page source and Tag Assistant. Two containers, or one container plus hard-coded tags, is a common source of duplicates.
  • Are there hard-coded Google tags, Meta Pixels or plugin-injected scripts outside GTM?
  • List every tag that sends a conversion. Note its trigger.
  • Triggers based on thank-you page views: can the page be reached without converting, by refresh, back button or direct link?
  • Triggers based on form submission: does the built-in form trigger fire on submit attempts that fail validation?
  • Custom HTML tags nobody can explain.
  • Paused tags, unpublished workspaces and the last publish date and author.
  • Consent settings on tags and whether a consent initialization trigger is used.

3. Google Ads conversion actions

  • List every conversion action with its source (website, GA4 import, calls, offline import), category, count setting, click-through window and value.
  • Which actions are primary? Primary actions are what bidding optimizes toward. Page views, scrolls or micro-events as primary are a high-severity finding for most lead-gen accounts.
  • Lead forms counted "Every" instead of "One".
  • The same action tracked twice: a Google Ads tag and an imported GA4 key event, both primary.
  • Enhanced conversions: on or off, and the diagnostics status.
  • Calls: calls from ads and calls from the website, with a sensible minimum duration.
  • Offline imports: is there one, when did it last upload, and what is its match rate?
  • Account-level goals versus campaign-level goal overrides.
  • The Conversions column versus All conversions, last 30 days, per action. Anything unexpected?

The detail behind these checks is in our Google Ads conversion tracking setup guide.

4. Meta events and deduplication

  • In Events Manager, list every event received for the dataset, by browser and server.
  • For events received from both, what is the deduplication status? Low or no deduplication on Lead or Purchase means double counting.
  • Who sends server events: a Shopify or WordPress integration, a CAPI gateway, server GTM, a CRM integration? More than one is a red flag.
  • Event match quality per event and which customer parameters are sent.
  • Events firing on page load that campaigns optimize for.
  • Domain verification and any diagnostics warnings.
  • Which event each active ad set optimizes for.

See the Meta CAPI deduplication guide for how to fix what you find.

5. GA4 key events

  • Which events are marked as key events? Are any of them noise, like page_view or scroll?
  • Event naming: consistent, lowercase, recommended event names where they fit (generate_lead, purchase).
  • Duplicate events: the same lead firing twice from GTM and an enhanced measurement form interaction.
  • Data streams, cross-domain configuration and unwanted referrals, especially from payment providers and booking tools.
  • Google Ads link status and whether GA4 key events are imported into Google Ads.
  • Data retention setting and internal traffic filters.

For GA4 specifics, use our GA4 audit checklist.

6. CRM source capture

  • Do new contacts have a gclid, fbclid and UTM values stored? Pull the last 50 paid leads and count blanks.
  • Are hidden fields present on every form, including popups and embedded forms?
  • Do values survive navigation from landing page to form page?
  • Are lifecycle stages or pipeline stages defined with dates, so outcomes can be sent back to ad platforms?
  • How are duplicate contacts handled, and does a second submission overwrite source data?

7. Consent

  • Is a consent banner present where the client's audience requires one?
  • Is consent mode implemented with all four signals (ad_storage, analytics_storage, ad_user_data, ad_personalization)?
  • Do defaults load before any tag fires?
  • Do Meta and other non-Google tags respect the banner choice?
  • Does anything fire before consent is given that should not?

Our Consent Mode v2 guide covers the implementation details.

8. Phone calls and offline outcomes

Lead-gen accounts often get a large share of conversions by phone, and they are the most commonly untracked.

  • Which phone numbers appear in ads (call assets, call-only ads) and on the site? Is each tracked?
  • Calls from ads: is the call conversion action set up, and does the minimum call length reflect a real inquiry rather than a misdial?
  • Calls from the website: is there a website call conversion with a Google forwarding number, or a call tracking tool that sends conversions back to Google Ads?
  • Click-to-call taps tracked as conversions: are they marked secondary, so a tap is not treated like a completed call?
  • If a call tracking tool is used, does it pass the source into the CRM?
  • Offline outcomes: are qualified leads or closed deals sent back to Google Ads or Meta? If so, when did the last upload succeed?

Our guide to Google Ads call conversion tracking goes through the options in detail.

9. Real test submissions on every form

Configuration review tells you what should happen. Test submissions tell you what does. For each form, calendar, phone link and checkout:

  1. Open a fresh private window with a tagged test URL (test gclid, fbclid and UTMs).
  2. Start GTM Preview and Tag Assistant. Open Meta's Test Events tool and GA4 DebugView.
  3. Navigate to a second page, then submit the form with clearly labeled test data.
  4. Record which tags fired, how many times, and with what values.
  5. Refresh the thank-you page and press back. Record anything that fires again.
  6. Check the CRM contact for every source field.
  7. Repeat with consent declined.
  8. Screenshot everything and note the time, so the client or their developer can reproduce it.

Agree with the client before testing, so the sales team ignores test leads, and delete or tag them afterwards.

10. Compare counts to the CRM

Take a fixed window, usually the last 30 days, and put these side by side per conversion type:

SourceWhat to pull
CRM or store backendNew leads or orders created, by source where available
Google AdsConversions per primary action
MetaResults for the optimization event
GA4Key event counts

Platform numbers will not match the CRM exactly. Attribution windows, conversion dates versus click dates, modeled conversions and cross-device behavior all cause differences. What you are looking for is a gap that the normal reasons cannot explain: a platform reporting more leads than the CRM received in total, or paid leads in the CRM with no matching conversion anywhere. See GA4 vs Google Ads discrepancies for the expected differences.

11. Severity scoring

Score each finding so the client knows what to fix first. A simple three-level scale works:

SeverityMeaningExamples
CriticalBidding or budget decisions are being made on wrong data nowPage views as primary conversion; Lead double-counted in Meta; main form not tracked at all
HighSignificant data loss or misattributionEnhanced conversions off; calls untracked; no gclid in CRM; consent mode missing where required
Medium or lowHygiene and future-proofingUnused tags; inconsistent naming; missing documentation; no internal traffic filter

For each finding, record: what was found, the evidence (screenshot or log line), the impact in plain language, the fix, and who needs to do it (agency, client developer, CRM admin).

12. Report structure

Clients and account managers read the first page and skim the rest. Structure it accordingly:

  1. Summary: three to five sentences on whether the account's conversion data can be trusted, and the top three fixes.
  2. Scorecard: one line per platform (GTM, Google Ads, Meta, GA4, CRM, consent) with a status.
  3. Findings: sorted by severity, each with evidence and the fix.
  4. Count comparison: the table from section 10 with a note on what the gaps mean.
  5. Test log: every test submission, what fired, screenshots.
  6. Fix plan: the order of work and what access it needs.

You can see how we lay one out on the sample audit page.

After the audit: fixing in the right order

Fix critical findings before touching campaigns. A sensible sequence:

  1. Stop the bleeding. Demote wrong primary conversions to secondary, pause duplicate tags, and fix Meta deduplication. These change what bidding learns from, so note the date for later analysis.
  2. Close the biggest gaps. Track untracked forms and calls, turn on enhanced conversions, and add click ID capture to the CRM.
  3. Tidy up. Naming, unused tags, internal traffic filters, documentation.
  4. Re-test. Run the same test submissions again and add the results to the test log, so the report shows before and after.
  5. Watch for two to four weeks. Compare counts to the CRM weekly. Smart Bidding and Meta delivery need time to adjust to corrected data, and clients need to hear that before the first weekly report.

Changing conversion actions during a period of heavy campaign changes makes it impossible to tell which change caused what. Where you can, separate tracking fixes from bid strategy or budget changes by at least a week.

Common mistakes in audits

  • Reviewing settings without submitting a single test lead.
  • Testing only the main contact form and missing popups, calendars and landing page variants.
  • Treating every finding as urgent, so the client cannot prioritize.
  • Fixing things during the audit without recording the before state.
  • Asking for admin access everywhere when read access would do.
  • Not telling the client that fixing duplicates will lower reported conversions.

Running it on a client account

If you want this done for you, send us one account for a free tracking check: we review it against this checklist across Google Ads, Meta, GA4 and the CRM and send a short video of what we find, under your agency's name if you want. Details are on the pricing page, and you can try the free tracking health check first for a quick read.

Frequently asked questions

What access do I need to audit conversion tracking?

Read access is enough for most of an audit: GTM read, GA4 Viewer or Analyst, Google Ads read only, view access to the Meta dataset, and a CRM user who can see contacts and forms. Edit access is only needed for fixes.

How long does a conversion tracking audit take?

It depends on how many forms, platforms and conversion actions the account has. The test submissions on every form and the count comparison take the most time and should not be skipped.

Why do platform conversion counts not match the CRM?

Attribution windows, click dates versus conversion dates, modeled conversions and cross-device behavior all cause normal differences. A platform reporting more leads than the CRM received in total usually means duplicates.

What is the most common critical finding?

Commonly, micro-events or page views set as primary conversions in Google Ads, or Meta Lead events counted twice because Pixel and CAPI are not deduplicated. Both mean bidding is learning from wrong data.

Should I fix issues during the audit?

Record the before state first. Fixing without evidence makes it hard to explain later why numbers changed.

Muhammad Qitmeer

Written by

Muhammad Qitmeer, founder of Augere Labs

Augere Labs is the white-label tracking partner behind marketing agencies. If a client account has numbers you don't trust, I'm happy to look at it with you and tell you straight what's wrong.

Connect on LinkedIn