Is Your Tracking Lying?
The day we launched our own campaign, the platform said four leads. We got two. A $26 cost per lead was reporting as $13. The 15-minute check that catches this before you scale a lie.
The day we launched our own campaign, the platform reported four leads. We'd actually received two.
A tag was firing twice, so every cost in the account was reported at half what it really was. We caught it the same day because we check. Most accounts never do, and tracking errors don't look like errors. They look like great results.
The platform happily optimizes toward the fiction, and you scale the wrong ad with total confidence. This is the fifteen-minute, read-only check we run before any budget decision. The one that matters most is the simplest: put what the platform says next to what your CRM actually got, and let the gap speak.
The three levels
Trust the dashboard. Make decisions on it. Find out six months later that half the leads were double-counted.
Connect the ads platform read-only plus your CRM, run the skill, and get a verdict with the reality ratio and the three fixes ranked by impact.
The truth check runs weekly inside the account audit, so a tracking break gets caught the week it happens instead of the quarter after.
The free skill
It checks whether the pixel fires on the right action, whether a server-side feed exists, whether events deduplicate, how strong match quality is, which event campaigns actually optimize toward, and — the one that matters — how the platform's count compares to what your CRM really received.
---
name: tracking-truth-check
description: A 15-minute audit that tells you whether the numbers your ad platform reports are real — pixel firing, server-side events connected, deduplication working, match quality, and whether reported leads match what your CRM actually received — before you make a single decision on them. Read-only. Trigger on "is my tracking right", "check my pixel", "my leads don't match", "why does Meta say more leads than I got", "verify conversion tracking".
---
# Is Your Tracking Lying?
Every decision in an ad account sits on top of the conversion data. When that data is
wrong, the platform optimizes toward fiction and you scale the wrong ad with confidence.
We've caught double-counted leads that made a $26 cost per lead look like $13. This
audit takes fifteen minutes and should run before any budget decision.
You are read-only. You inspect, you compare, you report. The human changes settings.
## Before you start (connector-first)
You need: the ads platform connected read-only (Meta's official ads connector or
equivalent), the site or funnel URL, and — this is the part everyone skips — the CRM or
inbox where leads actually land, connected or exported for the same date window. The
audit's most important check is platform vs reality, and you can't do it with platform
data alone.
## Check 1: Is the browser pixel actually firing?
Load the landing page and confirm the pixel script loads and a page-view event fires.
Then trigger the money action (opt-in, booking, purchase test) and confirm the matching
standard event fires — the correct event name, on the correct page, once. A pixel that
fires "Lead" on page load instead of on submit is the most common lie in small accounts.
## Check 2: Is there a server-side feed?
Browser-only tracking loses a meaningful share of conversions to ad blockers and mobile
privacy settings. Check whether a server-side connection (Conversions API or the
platform equivalent) exists and is receiving events. If it doesn't, this is usually the
single highest-value fix in the account — say so plainly and put it first.
## Check 3: Are events deduplicated?
If both browser and server send the same conversion, they must share an event identifier
so the platform keeps one. Trigger one test conversion and count what the platform
received. Two events for one action = double-counting, which inflates results and
corrupts optimization. State the observed count.
## Check 4: Match quality
Look at the platform's match-quality score for the money event. Low match quality means
the platform can't connect conversions to people, so it optimizes on a fraction of the
truth. Note which customer signals are being sent (email, phone, browser identifiers)
and which are missing — each missing one is a concrete fix.
## Check 5: Platform vs reality (the one that matters)
Pick the last 7 or 30 days. Count the platform's reported leads/purchases. Count what
the CRM, booking calendar, or payment processor actually received from that traffic in
the same window. Put the two numbers side by side:
Platform says: 48 leads Reality: 24 new contacts Ratio: 2.0x
A ratio near 1 means the account can be trusted. 1.5x or more means every cost-per-result
in the account is understated by that factor — and every "winning" ad might not be.
## Check 6: The optimization event
Confirm which event each campaign is actually optimizing toward. Sending an event and
optimizing toward it are different settings. A lead-gen campaign optimizing toward
link clicks will happily buy you cheap clicks that never convert.
## The report
1. **Verdict:** trust these numbers / fix tracking first — one sentence.
2. **The reality ratio** from Check 5, with both raw counts.
3. **Fixes ranked by impact,** each with exactly where to click. Never more than three.
4. **What's healthy** — name one thing that's right.
## Never
- Never guess a number the tools didn't return. "Could not verify" is a valid finding.
- Never change a setting. Never add or remove a pixel. Report; the human acts.
Want this checked before every dollar you spend?
Inside the AI CEO Lab I teach the full measurement setup — pixel, server-side, dedup — and how agents keep it honest. Or we install and verify it for you.