Performance Marketing
Published 2026-04-02 · Updated 2026-07-23 · 9 min read
Sapun Lamichhane
Founder & CEO of Arcetis
When a new ad account comes to us underperforming, the instinct — for a client and often for a less experienced media buyer — is to assume the fix is a bid adjustment, a new audience, or fresh creative. Something visible, something that feels like action.
Our first move is almost always none of those things. It's auditing whether the conversions the account is already reporting are actually real, because every other decision downstream — every bid change, every audience test, every piece of new creative — depends on that number being trustworthy first. Optimizing anything on top of a broken tracking number just produces a more confident-looking wrong answer.
“A bidding algorithm optimizes toward whatever conversion signal it's given. If that signal is inflated by a duplicate-firing tag, it will faithfully and efficiently optimize toward more of exactly the wrong thing — and it will look like it's working the whole time.”
This is the first stage of the Signal-to-Revenue Framework — Account Architecture & Tracking Foundation — and it's first for a structural reason, not a stylistic preference for caution. A bidding algorithm optimizes toward whatever conversion signal it's given, full stop. It has no independent way to check whether that signal represents something real.
If that signal is inflated by a duplicate-firing tag, or is counting a page view as a purchase, the algorithm will faithfully and efficiently optimize toward more of exactly the wrong thing — and because the number keeps going up, it will look like the campaign is working the entire time. Nobody gets an alert saying the growth is fake. The dashboard just keeps reporting a healthy trend line built on a broken foundation.
In practice, this audit means three specific things, not a vague "check the tracking" gesture. First: GA4 event configuration reviewed against what the business actually considers a converted customer — not what a developer assumed a conversion meant when the tag was first installed, possibly years earlier by someone no longer at the company.
Second: the GTM container reviewed for duplicate or orphaned tags accumulated over time — containers managed by several different people across several years reliably accumulate tags nobody remembers adding, some of which double-count the same real action as two separate conversions.
Third: each Google Ads conversion action confirmed to be mapped to a genuine revenue event, rather than a proxy that merely correlates with one — a confirmation-page view correlates with a completed purchase most of the time, but "most of the time" isn't the same as "is one," and the gap between those two is exactly where inflated conversion counts live.
It sounds like plumbing work, and it is. There's no version of a tracking audit that produces an exciting before-and-after screenshot the way a creative refresh or a bold new targeting strategy does. But skipping it means every later optimization decision — every bid change, every budget shift, every campaign pause — is being made against a number that might not mean what it appears to mean, and no amount of skill applied to bidding and targeting compensates for optimizing against the wrong input.
This is also why a tracking audit is the right first move even on an account that looks like it's performing well. A healthy-looking cost-per-lead built on duplicated conversions isn't a healthy account being managed conservatively — it's a broken account that hasn't been checked yet.
Only once tracking is verified do we move to the framework's later stages: segmenting audiences by buying-stage intent, aligning ad creative with the landing page it resolves to, governing budget against a documented decision rule instead of daily reactive changes, and reconciling ad performance against CRM pipeline data so the loop closes against what the business can actually verify closed.
Doing any of those first, on top of unverified tracking, just means making better-informed-looking decisions against a foundation that was never solid to begin with — a more sophisticated way of building on sand.
For most small-to-mid accounts, a few focused days to review GA4 events, GTM container hygiene, and conversion-action mapping. Reconciling historical performance against CRM data, where available, usually takes longer than the audit of the tracking setup itself.
It's most urgent then, but it's just as necessary after any change to the site, CRM, or tracking setup on an account you've always managed — a site migration, a new form tool, a CRM switch — since any of those can silently break tracking that was previously accurate.
Duplicate conversion counting — most often a form submission and its confirmation-page redirect both firing as separate conversion events for the same single real action, doubling the reported lead count without a single extra lead existing.
