- Tracking
- Agency
- Onboarding
- Audit
The New Client Tracking Check for Agencies
By Olam Sule · Published 2 Sept 2026
TL;DR
Before an agency takes on a new client, scan the site from the outside first: which tags load (GA4, GTM, Meta Pixel, Google Ads), whether any fire twice, and whether consent is set so data is collected at all. Run it before you get account access, so broken tracking is on record as inherited, not caused by you. We run this check for agencies most weeks.
Olamide Sule, founder of Dolphin Analytics, a London analytics consultancy that sets up and fixes GA4 and GTM tracking for agency and in-house teams.
Taking on a new client means inheriting whatever tracking the last team left behind, and it is rarely clean. We run this pre-onboarding check for agencies most weeks, usually in the first day or two, before anyone has account access. The point is plain: find what is broken now, write it down, and make sure the numbers you are about to be judged on were not already wrong when you arrived.
Why run a tracking check before you onboard a client?
Run it because the alternative is quietly owning someone else’s broken data. The moment you take over an account, the next monthly report carries your name, even if the conversion behind it stopped firing months before you were hired. A pre-onboarding scan records the starting state, so a gap you inherited stays on record as inherited.
Agencies feel this as a specific fear: getting blamed for numbers that were already wrong. The check below is how you get ahead of it. It moves from what you can see without a login, to what needs GA4 and Google Tag Manager access once the client grants it.
What can you check before you even have account access?
Quite a lot, because most tracking runs in the visitor’s browser where anyone can read it. From the outside you can see which analytics and ad tags load on the client’s site (GA4, GTM, Meta Pixel, Google Ads, TikTok), whether the same tag fires more than once, and whether a consent banner blocks storage until the visitor opts in. No login required, which is the whole point on day one.
Sonar is our free external scanner for exactly this first look: one click over the client’s URL, no account, and it reports the tags, pixels and consent setup it can prove from the outside. If you would rather read it by hand, our guide on how to check what tracking a website uses walks through the browser tools that show the same thing.
Three questions the outside-in pass answers:
- What is actually installed? GA4, Universal Analytics still lingering, a tag manager, ad pixels, a consent platform. A site running two analytics tools or a dead pixel tells you the last setup was never tidied.
- Is anything firing twice? A duplicated GA4 tag or a pixel loaded through both GTM and the theme inflates every count. It is the single most common fault we find on a first scan.
- Is consent even letting data through? A banner that blocks tags until opt-in, with a low opt-in rate, means the client has been under-counting for as long as it has been live.
Which conversions actually fire, once you have access?
This is the check that decides whether the account is trustworthy, and it needs
GA4 and GTM access. List the conversions the client actually cares about (a
purchase, a qualified lead, a booking), then confirm each one fires, fires once,
and carries the right value. A purchase event that fires with no revenue, or a
lead form that was never wired into the container, is worth more to catch now
than in month three.
Use GTM Preview mode and GA4 DebugView to watch events in real time as you click through the site. Our GA4 audit checklist covers the event and key-event passes in order, and the GTM audit checklist covers the container side: duplicate tags, triggers firing where they should not, and variables resolving to nothing.
Do the numbers reconcile with a source of truth?
A conversion that fires is not the same as a conversion that is right. Compare GA4 against the system that holds the money: Shopify or Stripe for ecommerce, the CRM for lead gen. GA4 almost always shows fewer orders than the store, because consent blocking, ad blockers and dropped events cost it some data, so a small gap is normal. A large or growing gap is a fault, not a rounding error, and it is usually a tag or consent problem underneath.
Do the same for the ad platforms. Meta and Google Ads both report conversions their own way, and both tend to claim more than the store confirms. Knowing that gap on day one stops you from optimising a campaign against a number that was never real. Our post on conversion tracking for agencies goes deeper on where those platform numbers diverge from booked revenue.
Who owns the accounts, and what access will you actually get?
Before any of the deeper checks, sort out ownership. Confirm the client, not a former agency, owns the GA4 property, the GTM container, the Google Ads account and the Merchant Center. Get Admin or Editor access in your own name, not a shared login, and note anything you are refused, because missing access is its own risk to flag. An account you cannot fully see is an account you cannot fully stand behind.
Turn the check into your baseline
Save what you found. A dated record of the tags loading, screenshots of the key GA4 reports, and the access you were granted is your evidence that the broken conversion or the consent gap was there before you started. That baseline is cheap to make now and impossible to reconstruct later. It is also the difference between “we found this and fixed it” and “we might have caused this”.
Fix it early, and it stops being a liability
This is the work we do for agencies most weeks, often as a white-label partner, so you can hand a client clean measurement without building a data team of your own. Caught early, broken tracking turns from a risk into a service you can charge for: on one media agency engagement the measurement work gave the team twenty hours a week back from manual reporting and £50K in dashboard upsells to its own clients.
For the fastest first-day version, run Sonar over the new client’s URL: free, one click, no account, and it reports the tags, pixels and consent setup it can prove from the outside. Once you have GA4 and Google Tag Manager access and want every fault named in a written report, that is the paid audit, scoped per property. Either way, see how we approach tracking for the order we work through it.
The onboarding tracking check, in order
- Scan the site from the outside: which tags load, what fires twice, whether consent blocks data. No access needed.
- Confirm the key conversions fire once, with the right value, in GA4 and GTM.
- Reconcile GA4 against the store or CRM, and against the ad platforms.
- Sort out account ownership and get named access.
- Save a dated baseline before you change a thing.
Work it in that order and you start the engagement knowing exactly what you inherited, which is the only position from which a broken number is the last team’s problem and not yours.
Frequently asked
Should you check a new client's tracking before or after signing them?
Do the outside-in scan as early as you can, ideally during onboarding and before your team touches the account. The first monthly report you send is yours, even when the tracking behind it was already misfiring when you arrived. Scanning early documents the starting state, so a conversion that never fired is on record as inherited rather than introduced by you.
Can you audit a client's tracking without account access?
A surprising amount, yes. From the outside you can see which analytics and ad tags load in the browser (GA4, GTM, Meta Pixel, Google Ads), whether the same tag fires more than once, and whether a consent banner blocks them until opt-in. None of that needs a login, which matters because on day one of an engagement you usually do not have one. Confirming why a tag duplicates, or whether a conversion value is correct, needs GA4 and GTM access later.
What should an agency do if a new client's tracking is already broken?
Write down exactly what you found and when, then decide who fixes it. We pick up this work for agencies most weeks, often as a white-label partner, so you can hand the client clean measurement without building a data team. Fixing it early also turns a liability into a service line: on one media agency engagement the measurement work gave the team twenty hours a week back and £50K in dashboard upsells to its own clients.
How do you prove broken tracking was inherited, not your fault?
Capture a dated baseline before you change anything: a scan of the tags loading on the site, screenshots of the key GA4 reports, and the account access you were granted. That record is your evidence that a broken conversion or a consent gap existed on day one, so a bad number three months in cannot be pinned on your setup work.