A reasonable assumption, and a wrong one: that if you are paying for a tag governance platform that crawls your site and inventories every script, you would hear about it when a Meta pixel stopped producing conversions. You would not, and the reason is not a shortcoming in the product. It is that a dead pixel and a missing pixel are different events, and only one of them is visible from the page.
This comes up often enough to be worth writing down properly, because the mistake is expensive in a specific way: it produces false confidence. A team with a governance tool believes tracking is covered, so nobody checks, and a break survives longer than it would have at a company with no tooling at all.
What a scanner can actually see
Site-scanning tools drive a crawler through your pages and record what loads. That is a genuinely useful capability and they do it well. From the page, the honest list of what is observable:
- Which tags, pixels and cookies are present, and on which pages.
- Which measurement IDs they carry.
- Whether something fires before consent was given.
- Whether a script appeared that nobody approved.
- Whether a tag is on pages it should not be on.
Every one of those is a property of the page. The crawler loads the URL, watches what happens in the browser, and writes down what it saw. If the Meta pixel snippet is in the source and executes, the scanner reports a Meta pixel.
What a dead pixel looks like from the page
Now consider the ways ad tracking actually fails in a live account, and ask what each one looks like to a crawler.
A dataset gets restricted
Meta restricts a dataset after a category review. The pixel is still installed, still executing, still sending requests. What changed is on Meta’s side: what it will accept and use. From the page, nothing happened. We have written separately about what a restriction actually does to your tracking, and it is invisible to any site crawl.
Match quality collapses
A site change stops passing an email or a phone number with the purchase event. The event still fires. It still arrives. It just matches far fewer people, so attributed conversions fall. A scanner sees an event firing, which it was.
A Google Ads conversion action goes inactive
The tag is on the thank-you page, correctly, and the conversion action in the Google Ads account has gone inactive or unverified. The page is blameless. The failure lives entirely in the account, as covered in what conversion tracking unverified means.
A Conversions API integration loses its token
Server-side events stop. The browser-side pixel is untouched and still visible to the crawler, so a scan is clean while half your event volume disappeared.
Duplicate events start
An app adds a second copy of a tag and conversions double. A scanner is actually good at this one, if it reports measurement IDs per page — it is the exception on this list. But it will report presence, not the inflated count in the ad account.
The pattern: four of those five failures leave the page in exactly the state the scanner expects. The report comes back clean on the same day your purchase volume halved.
What this looks like as a timeline
Abstract versions of this are easy to nod along to and easy to forget, so here is the concrete shape it takes, which is close to how it actually goes.
- Tuesday. A developer ships a consent banner change. It is a small change and it does not touch the pixel.
- Tuesday afternoon. The Meta pixel is now in a category a portion of visitors decline. It still loads for everyone who accepts, still appears in the page source for everybody, still fires.
- Wednesday. The scheduled scan runs. The pixel is present on every page it should be on, with the right ID, and the scanner confirms it holds off before consent, which it now does more thoroughly than before. The report is not merely clean, it looks like an improvement.
- The following week. Reported purchases are down by a third. Nobody connects it to a consent change, because the governance tool said tracking was fine and campaign performance is a noisy number.
- Three weeks later. Somebody compares Shopify orders against Meta-reported purchases and the gap is obvious. Three weeks of bidding has already happened against the degraded signal.
Nothing in that sequence is a tool failing. The scanner answered its question correctly every time. The question was simply not the one that mattered.
A crawler is not a customer
There is a deeper version of this worth stating, because it explains the whole category difference.
A scanner visits your pages as a scanner. It does not add to cart, it does not pay, it does not make a consent choice the way a real visitor in a real region does, and it does not complete the purchase that would fire the event that matters. So it can verify that the mechanism exists. It cannot verify that the outcome happened, because it never generates the outcome.
Everything real users do that a crawler does not — decline consent, use a browser with different protections, buy things — is where the interesting failures live. This is the same vantage-point problem that separates the tools in our Trackingplan vs ObservePoint comparison: watching real traffic and crawling pages have opposite blind spots.
What each kind of tool is honestly for
None of this is a criticism of governance platforms. They do a job, and it is not this one.
- ObservePoint— automated scanning across a whole site for privacy, accessibility against WCAG 2.1 AA, tag accuracy and analytics validation. Buy it to certify a site, page by page, with a record.
- Tag Inspector— privacy risk specifically, matched against a large tag library, with consent policy management and violation tracking. Buy it when privacy compliance is the obligation you have to discharge and someone outside the team will read the output.
- Feroot— client-side security, blocking skimming, formjacking and data exfiltration in real time, with PCI DSS 4.0.1 payment-page script monitoring. Buy it when an unapproved script is a security incident.
All three answer “what is running on my pages, and should it be”. None answers “are my conversions still arriving in my ad accounts”, because that question is not answerable from the pages.
What actually answers the other question
To know whether tracking is working, something has to read the ad platforms themselves — the place where the consequences show up. The options, in increasing order of reliability:
- Looking. Events Manager and the Google Ads conversions summary both show you the truth. The flaw is that they require you to remember, per account, forever. A break on Friday waits until Monday.
- Platform diagnostics. Meta surfaces some issues in Events Manager and Google flags conversion action statuses. Useful, inconsistent about notifying you, and per-account rather than across a roster.
- A scheduled manual review. Putting the check in a recurring process beats relying on memory, which is what a tracking health SOP is for. It still depends on somebody doing it.
- Continuous monitoring of the accounts. Something that reads what the platforms report, on a schedule, and tells you when it changes.
Taglertis the fourth one, and we build it, so read this with that in mind. It connects to each Meta and Google Ads account with a login, rechecks every thirty minutes, and emails you when a pixel stops reporting, Event Match Quality drops, duplicate events start or a conversion action’s status changes. It has nothing to say about cookies, consent banners or accessibility, which is why it is a complement to a governance platform rather than a replacement for one.
Questions worth asking a vendor
If you are evaluating something and want to know which category it is in before the demo talks you round, these four separate them quickly:
- Do you connect to my ad accounts, or crawl my site? The single most clarifying question, and the answer is never ambiguous.
- Would you alert me if my Meta purchases dropped to zero but the pixel was still on the page? A site scanner cannot, and a good vendor will say so.
- Do you report Event Match Quality? It only exists inside the ad platform, so this is a reliable proxy for whether a tool reads accounts.
- How often do you check, and do you push or do I pull? A monthly scan you have to open is a different product from something that emails you on the day.
If you only remember one thing
Ask of any tool you are relying on: does this read my pages, or does it read my ad accounts? It is a cleaner question than comparing feature lists, and it predicts which failures the tool can possibly catch.
Tools that read pages catch what is installed and whether it should be. Tools that read accounts catch whether it is working. Most teams that think they are covered have bought the first kind and assumed it did the second, and the gap between those is where a client’s budget spends against numbers that are not real.
Sources
Stop finding out about broken tracking from your client.
Taglert monitors your pixels and conversion tracking 24/7 and alerts you the moment something breaks.
Start your free trial- First scan in ~5 minutes
- 30-day money-back guarantee
- No credit card
