How to Check Your Meta Pixel and Conversions API Health

To check your Meta pixel and Conversions API (CAPI) health, compare Meta's purchase count against your real store orders for the same dates, compare browser events against server events, confirm the two are being deduplicated, review Event Match Quality and match key coverage, and test that the Facebook click ID survives your checkout. If Meta's numbers do not line up with real orders, fix tracking before you change anything else, because every optimization and every report sits on top of this data.

Meta's delivery system learns from the purchase events you send it. If those events are missing, doubled or poorly matched to real people, Meta learns from bad information. This guide walks through the checks we run, in order.

Why does pixel and CAPI health matter so much?

Because bad event data quietly corrupts everything downstream. It does not throw an error. It just makes your numbers wrong in ways that look believable.

Here is how each failure hurts:

  • Purchases under firing. Meta sees fewer sales than you actually make. It has less signal to learn from, and prospecting usually suffers first, because finding new buyers needs the most data.
  • Purchases counted twice. Meta sees more sales than you make. ROAS looks healthier than it is, which can delay a real diagnosis for months.
  • Weak matching. Meta receives the purchase but cannot connect it to a real person who saw an ad. Reported ROAS drops even though the customer journey really did start on Meta.

The good news is that these problems are fixable, and fixing them often reveals performance that was there all along but invisible.

What are the Meta pixel and Conversions API?

The Meta pixel is a piece of code on your website that sends events (like page views, add to carts and purchases) from the shopper's browser to Meta. The Conversions API, usually called CAPI, sends those same events from your server or store platform directly to Meta.

You want both. Ad blockers and browser privacy limits stop the pixel for some shoppers, and CAPI fills those gaps. Running both creates one new job: telling Meta when two events describe the same purchase so it counts it once. That is deduplication, and it is where many setups break.

Step 1: Does Meta's purchase count match your real orders?

This is the most important check. Pull Meta's Purchase event count for a recent window, then pull your actual order count from your store platform (for example, Shopify) for the exact same dates and time zone. They should be close.

If Meta is well below your orders, purchases are under firing somewhere. If Meta is well above your orders, events are being double counted or something other than real purchases is firing the Purchase event.

Common causes: a theme or app update silently broke the tag, the event fires on page load instead of on order completion, two tag tools both fire it, or the order value is missing, sent as text or tagged in the wrong currency.

Do not scale spend while this check is off.

Step 2: How do browser and server events compare?

Look at three numbers side by side for Purchase: browser (pixel) events, server (CAPI) events, and the deduplicated total Meta actually counts. Then put your real order count next to them.

In a healthy setup, the pattern looks roughly like this:

  • Real store orders. What healthy looks like: Your ground truth
  • Server (CAPI) events. What healthy looks like: Close to real orders, since the server sees almost every order
  • Browser (pixel) events. What healthy looks like: Somewhat lower than server, because some browsers block the pixel
  • Deduplicated total. What healthy looks like: Close to real orders

Two broken patterns show up again and again.

Server events far above real orders while browser events look normal. This is the classic duplicate CAPI problem. Usually either the browser and server events are not sharing a matching ID, or two separate server sources are both sending purchases (for example, your store's native integration and a separate tracking tool). The fix is one server source and a matching event ID on both channels.

Server events far below browser events. CAPI is under firing. Check that the integration is connected, active and sending Purchase, and that it has fired recently.

Also check when each channel last fired. A stale one is effectively down.

Step 3: Is deduplication actually working?

Deduplication works when the browser event and the server event for the same purchase carry the same event name and the same event ID. Meta uses that pair to recognize them as one purchase.

If the event ID is missing on one side, or the two sides generate different IDs, Meta cannot match them. It counts both, and your purchases double.

How to check: use the Test Events tool in Events Manager, place a real test order, and watch both events arrive. Confirm they share the same event ID. Then step back and run the order comparison from Step 1 again. The deduplicated total should land close to your real orders. If it lands meaningfully above, deduplication is not doing its job, and your optimization data is inflated.

Step 4: What is Event Match Quality and what does a good score look like?

Event Match Quality (EMQ) is Meta's score, out of 10, for how well the customer information sent with an event can be matched to a real Meta account. Higher is better, because a matched event can be connected back to the ad that person saw.

Purchase is the event that matters most, because it feeds optimization and reporting. It should be the strongest score you have. A well built CAPI feed sends rich customer details with each purchase, and you will see that reflected in a high Purchase score.

Do not panic about lower scores on upper funnel events like page views and add to carts. Most of those shoppers are anonymous; they have not given you an email or phone number yet. A modest score there is normal, not a defect.

The score is driven mostly by which customer details you send. Strong identifiers like email and phone number carry the most weight. Your own hashed customer ID (external ID) helps too. Name, location details, IP address and browser information add more. Meta's click ID and browser ID round it out. If your Purchase score is weak, the fix is almost always sending more of the strong identifiers, more consistently.

Step 5: Which match keys should your purchase events carry?

Match keys are the pieces of customer information attached to an event. Check what share of Purchase events carry each. Nearly every online order collects an email, so nearly every Purchase should carry one; low coverage points to a checkout data problem. Phone and your hashed store customer ID (external ID) should be present as often as you can manage. Customer details must be hashed per Meta's rules, which most native integrations handle for you. The click ID gets its own section.

Step 6: Why is the Facebook click ID (fbc) so important?

The fbc value is how Meta connects a purchase to a specific ad click. When someone clicks your ad, Meta adds a click identifier (fbclid) to the landing page URL. Your pixel turns that into a first party cookie, and that value should travel with every later event from that shopper, including the purchase.

The common mistake is expecting fbc on every purchase. It only exists for buyers who clicked a Meta ad. People who came from email, search or a direct visit will never have one. So judge fbc coverage against the share of your orders that come from paid Meta traffic. If roughly that share of purchases carry fbc, your tracking is probably fine.

If coverage is well below that share, the click ID is getting lost. The usual causes, in rough order of how often we see them:

  1. Server side setup with no access to browser cookies. Some custom or middleware CAPI setups send purchases from a backend system that never sees the shopper's cookies. The fix is to capture the click ID and browser ID at checkout and store them on the order so the server can send them. Many native store integrations do this automatically; custom setups often do not.
  2. Redirects or a separate landing page domain. If your ad goes to one domain and then redirects to the store, the click ID can be stripped along the way. Send ads directly to the store domain, or pass the click ID through explicitly.
  3. A consent banner that blocks the pixel on landing. If the pixel cannot fire on the first page, the cookie never gets set. Work with your consent setup so the click ID can be captured in a compliant way.

How to test it: click one of your own live ads, buy, and watch Test Events and your browser's developer tools. Confirm the click ID is in the landing URL, the cookie is set, and fbc appears on both browser and server Purchase events. Where it disappears is where the leak is.

One limit to know: some browsers, especially on Apple devices, cap how long script set cookies last, so long consideration purchases lose some click IDs regardless. Server set first party cookies (through Meta's CAPI Gateway or server side tagging) can extend that.

What if you also sell on Amazon or in stores?

Then some of the people your ads move will buy somewhere your pixel cannot see. This is not a tracking bug, but it must be part of the conversation, or you will expect pixel fixes to close a gap they cannot close.

Those buyers never touch your site, so they do not affect your deduplication math, unless someone is feeding marketplace or offline orders into your dataset through CAPI. Those arrive with customer details but no click ID, inflating server purchases and dragging fbc coverage down. Also, when ad clickers leak to a marketplace, the purchases left on your site tilt toward email, organic and returning shoppers, which can lower fbc coverage legitimately.

Either way, platform ROAS will understate Meta's real impact. The fix is measurement, not pixel work: a Conversion Lift test or marketplace attribution tools. For how tracking problems ripple into every attribution window, see 1 day click, 7 day click and view.

What is the one rule for reporting pixel health?

Never report a number you cannot verify. If a figure is not available, say so and note how to get it. Do not estimate an EMQ score or guess at event counts.

Key takeaways

  • Every Meta optimization and report depends on accurate purchase events, so check tracking before you change budgets or creative.
  • Start by comparing Meta's purchase count to real store orders for the same dates.
  • Compare browser events, server events and the deduplicated total; server far above real orders usually means duplicate CAPI.
  • Deduplication needs a matching event name and event ID on both browser and server events.
  • Purchase should be your strongest Event Match Quality score; email, phone and external ID are the biggest levers.
  • Judge click ID (fbc) coverage against your share of paid Meta orders, not against all orders.

FAQ

Do I need both the pixel and the Conversions API?Yes, in most cases. The pixel captures browser detail, and CAPI fills the gaps when browsers block the pixel. Running both gives Meta the most complete signal, as long as deduplication is set up correctly.

Why does Meta show more purchases than my store?Usually because purchases are being counted twice. The most common causes are missing or mismatched event IDs between browser and server, or two server sources sending the same purchase. Fix that before trusting any ROAS number.

Is a low Event Match Quality score on add to cart a problem?Usually not. Most people who add to cart have not shared contact details yet, so there is less to match. Focus on your Purchase score, which should be the strongest.

Why is my fbc coverage so low?First, check it against the share of your orders that come from paid Meta clicks, since only those buyers can have it. If it is still low, the usual causes are a server setup that cannot see browser cookies, redirects that strip the click ID, or a consent banner that blocks the pixel on landing.

How often should I run a pixel health check?After any change to your site theme, checkout, apps or tracking tools, and on a regular schedule otherwise. Tracking tends to break quietly after updates, so a routine check catches problems before they distort months of data.

Make sure your data is telling the truth

If you are not sure your pixel and CAPI are reporting real sales, that is worth knowing before your next budget decision. Our free Meta account audit includes a full account audit plus net new reach and cumulative reach reports, and we will flag any tracking issues we find along the way. Request yours here.

Keep reading

More on measurement:

See every article on measurement

Ready for your free Meta audit?

We will review your account and run net new reach and cumulative reach reports, free.

Get your free audit

Check what you learned

Tap an answer to see if you got it right.

1. What should you check first in a pixel health check?

If Meta's count doesn't match real orders, every report built on it is off.

2. Server events are far above real orders. What does that usually mean?

Too many server events usually means the same purchase is being sent twice.

3. What does deduplication need?

Meta only knows two events are the same purchase if the name and ID match.

Ready for the full test? Take the Measurement exam and get certified. Want this run on your account? Get my free audit.

About the author

Jake Mehani, founder of Creative Analytics

Jake has spent 15 years running Meta and paid social ads for consumer brands. Meta published his RBX Active Advantage+ case study, he taught the DMAnc Meta and TikTok ads masterclass, and he spoke at Motion's 2024 Creative Strategy Summit. His senior team runs media buying, creative and measurement like an in house team.

Jake on LinkedIn  |  Case studies  |  Free Meta audit

Previous
Previous

Persona Scaling: How to Grow a Winning Ad Through New People and Angles

Next
Next

1 Day Click vs 7 Day Click vs View Through: Which Meta Attribution Window Should You Trust?