Signal Audit
Your ads are learning
from whatever survived the trip.
Every platform's algorithm optimizes toward whichever conversions actually make it back, not the ones you meant to send. A stripped click ID, a redirect that drops a parameter, a pixel that never fires. None of it looks like an outage. It looks like a slightly worse audience every quarter, and almost nobody traces it back to where it actually happens.
We run six checks against your domain and tell you which ones pass. Every row comes with the instruction to confirm it yourself in about a minute.
Run the six checksYour domain and a work email. Nothing to connect, no call.
Who this is for
Written for the person
who owns the spend.
You're running paid on LinkedIn, Google, Meta, or all three, and it's real money on a monthly cycle. The account is instrumented, somebody set it up, and it has mostly been left alone since. When performance moves you can name three things that might explain it, and you can't rule any of them out. Nobody looks, because it reads as a reporting problem, and reporting problems get filed.
If you can't say for certain that a click on your ad arrives at your form with everything the platform needs to learn from it, this is for you.
- Paid owners who inherited the tracking setup and have never had a reason to open it
- Founders and revenue leaders watching cost per lead climb with no explanation that survives scrutiny
- Teams about to raise spend on an unaudited channel
What we check
Six places the signal breaks
between the click and the form.
None of these need access to anything of yours. They run against your live site, the same way a browser sees it, which is the same way the platform's tag sees it. Each one comes back as a pass or a fail. No scores, no percentages, no opportunities for improvement. A row either holds or it doesn't, and every row tells you how to check our answer yourself.
01
Your click IDs survive the trip
Google, Meta, and LinkedIn each attach an ID to the click. If a redirect strips it before the page settles, the platform never connects the conversion to the ad that earned it.
Confirm it: paste your homepage into the address bar, add ?gclid=TEST&fbclid=TEST&li_fat_id=TEST
, hit enter, and read the URL once it stops moving.
02
Your UTMs survive the same trip
The same failure, one layer up. Your own reporting loses the campaign, so the channel that built the pipeline gets credited to direct.
Confirm it: same test, using ?utm_source=TEST&utm_medium=TEST&utm_campaign=TEST
.
03
Tagging runs on your own domain
When conversions leave through a hostname you don't own, the browser gets to decide how long that data lives. Usually days, sometimes hours.
Confirm it: view source and look for a script loading from your own domain rather than from googletagmanager.com
, facebook.net
, or snap.licdn.com
.
04
Pixels wait for consent
If tags fire before anyone chooses, you have a compliance exposure. If they fire and the choice is later ignored, you have both an exposure and a data problem.
Confirm it: open your site in a private window, don't accept cookies, open the network panel, and watch for requests to Facebook, LinkedIn, or Google.
05
One container, not several
Two GTM containers on one page is the signature of an agency handoff nobody cleaned up. Events double, and the platform bids on a conversion count that never happened.
Confirm it: view source, search for GTM-
, and count the distinct IDs.
06
Every platform you run has a path back
Checked separately for each channel you tick on the form. A pixel with no server-side route is the most common way a conversion reaches your site and never reaches the buyer of the ad.
Confirm it: search your source for fbq
, the LinkedIn Insight tag, and gtag
, then look for a first-party endpoint alongside them.
Six rows. Six answers. Six things you can check in a browser tab.
What a fail costs
What a broken row costs you
in the auction.
Every auction spends your budget by predicting who converts next, and it predicts from whichever conversions actually reached it. A failing row doesn't just lose one form fill. It removes a real buyer from what the algorithm is allowed to learn from, so the next impression goes to someone who resembles your reporting instead of someone who resembles your buyer.
That shows up as cost per lead that creeps for no visible reason, audiences that fatigue faster than they should, and a creative team rebuilding assets that were never the problem. None of it reads as a tracking issue, so nobody checks the six rows that would explain the number.
A failing row spends real budget training the algorithm to find the wrong company.
Run the six checksYour domain and a work email. Nothing to connect, no call.
What this doesn't cover
We can tell you what left the browser.
Not what happened to it after.
Every check here runs against what's publicly there: your HTML, your redirects, your request headers. That's enough to tell you whether the signal leaves your site intact. It's a different question whether it landed somewhere useful once it left, and this audit doesn't guess at that question to look more complete.
- A consent tool or a bot rule that blocks our request the same way it blocks a real visitor gets marked unresolved, not assumed clean.
- A script served from a domain we can't identify gets named as unreadable, not scored as a pass or a fail.
- Anything that happens on your own server once it receives the request sits past the edge of a public scan.
If a row can't be confirmed, the page says so instead of filling in a guess.
What closes a fail
Closing a fail is a fix,
not a rebuild.
Most of what fails here is a setting, not a system: a redirect rule that drops a parameter, a second GTM container nobody removed, a pixel that fires before consent. Whoever already has access to your site can close those directly, the same day, with nothing new to install.
A smaller number point at something structural: no first-party endpoint at all, or an ID with nowhere durable to land. The report tells you which fails are which, so you're not guessing at the fix.
Fixing what's broken rarely takes a bigger stack. Usually it just takes the right row.
Who runs it
Media buyers who went
and got the other half.
Most people who understand this problem don't run paid media, and most people running paid media don't build the layer that feeds it. That split is why broken tracking survives so long. The agency doesn't own the instrumentation and the RevOps consultant doesn't own the spend, so the thing between them belongs to nobody.
These six checks are the first thing we run before taking on paid work, because we won't run media we can't measure. Publishing the audit is just doing that part in the open.
Run the audit
No account access.
No install. No call.
Your domain, a work email, and a few lines about your setup. We run the checks, then someone looks at the results before sending. Expect your report by email within three business days.