Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Meta Ads › Troubleshooting

Low Event Match Quality on Meta

Why Meta event match quality is low: few customer parameters, hashing errors, browser-only events, missing click identifiers, consent limits and bad formats.
Troubleshootingintermediate6 min readUpdated

First, confirm which events score low

The rating is reported per event, not per dataset, so start in the event list in Events Manager and note which events carry a low rating. A weak score on page views is common and rarely urgent; a weak score on the purchase or lead event that campaigns optimise for is the one to work on. Open that event's detail and read the parameter coverage: which customer information fields arrive, and on what share of events. Note the connection method too, since a browser-only event and a server event with the same name score differently.

Then send one test event through each route with the test event code and inspect the payload. The point is to see the actual values, not only their presence: an email field that is present but formatted wrongly shows as coverage yet contributes nothing to matching. The diagnostics tab lists formatting faults Meta detects, which narrows the table below quickly.

Diagnosis table

Diagnosis

SymptomCheckLikely causeFix
Parameter coverage lists only one or two fieldsEvent detail coverage; the payload of a test server eventToo few customer information parameters sentAdd the identifiers you already hold lawfully, starting with email and phone
Coverage looks healthy but the rating stays low; diagnostics warn about formatsRaw value before hashing for one test eventValues hashed without normalising, or in the wrong formatNormalise to Meta's documented rules, then hash with the specified function
Event lists only a browser connectionConnection methods for the event in the overviewBrowser-only event without a server copyAdd a server event through the Conversions API with deduplication
Server events arrive but browser identifiers are absent from coverageServer payload for the click and browser ID fieldsClick and browser identifiers not passed to the serverRead the identifiers from the visitor's cookies and pass them with the event
Rating varies by region or falls after a banner changeConsent state recorded with the order; banner configurationConsent limits what may be sentRespect the limit; improve what consented events carry

Fix, verify and prevent each cause

Thin customer information

A server event with only an IP address and user agent gives Meta very little to work with. Shops often hold an email address, phone number, name and postcode for every order but send none of them because the integration's defaults were never extended.

Fix: add the identifiers you already collect and may share, starting with the strongest documented fields.

Verify: the event detail's coverage shows the new fields on most events after a few days, and the rating moves.

Prevent: list the fields each event will carry in the integration specification, using the parameter set described in the Conversions API guide.

Hashing and format errors

Meta requires personal fields to be normalised before hashing: lower case, no surrounding spaces, phone numbers in the documented format with a country code, and so on. A value hashed with capitals or a trailing space produces a different hash and matches nobody.

Fix: apply the documented normalisation to each field, then hash with the specified function, and send non-personal fields such as the IP address unhashed.

Verify: the diagnostics tab clears the format warnings and coverage stops overstating usable data.

Prevent: keep a unit test in the integration that hashes a known value and compares it with the expected output; the normalisation notes in the Conversions API guide are the reference.

Browser-only events

An event sent only by the pixel carries what the browser knows and nothing from the order record.

Fix: add a server copy of the key events with customer information, sharing the event ID so the two are deduplicated; the duplicate conversions page covers what happens when that pairing fails.

Verify: the event lists both connection methods and the rating for it improves.

Prevent: make a server copy the default for every event used for optimisation, as the Conversions API guide recommends.

Click and browser identifiers left behind

The pixel sets a browser identifier and, when someone arrives from an ad, a click identifier. If the server event omits them, it loses the link to the session the pixel saw. Integrations miss them because the server never reads the cookies, or because the checkout runs on a domain where those cookies are not available.

Fix: capture both identifiers when the session starts, store them with the cart or order, and send them with the server event.

Verify: coverage for those fields rises on server events.

Prevent: include them in the field list for every new form or checkout flow, as set out in the Conversions API guide.

A visitor who declines advertising tracking should produce no pixel event and no server event, and the visitors who remain may be fewer and less varied in some regions. This lowers what can be matched, and it is correct.

Fix: confirm the consent decision travels into the order record and that the server respects it; then improve formatting and identifier coverage for consented events only.

Verify: test orders with consent declined send nothing, and consented orders carry the full field set.

Prevent: document the consent rule for every route, following the privacy and consent guide.

Prevention: a parameter contract with developers

Most low ratings come from integrations nobody revisits. A written parameter contract keeps them honest. For each optimised event it names the fields sent, where each comes from in the site or order record, the normalisation applied, whether it is hashed, and the consent condition under which it is sent. The contract sits beside the tag container and the server code, and changes go through it rather than around it.

Pair the contract with a short review cycle. After each checkout or form release, send a test event and compare its payload with the contract field by field. Once a month, record the rating and coverage for each key event so that a decline is visible as a trend rather than discovered by accident. When a new data source appears, such as a loyalty sign-up that captures phone numbers, check whether the contract allows that field for advertising before adding it. Better matching improves more than reporting: matched events feed Attribution, website audiences and the seeds behind each Lookalike Audience, so thin matching often shows up later as the shrinking pools described on the custom audience too small page. The Conversions API integration is the place where most of these fields are added.

Illustrative example — not a real result

A furniture retailer with a weak purchase score

A furniture retailer sends purchases from both the pixel and a server integration, yet the purchase event carries a low rating. Coverage shows email on 96% of server events, phone on 0% and the browser identifier on 4%. A test payload shows the email hashed before lower-casing, so mixed-case addresses never match. The developer adds normalisation before hashing, reads the browser and click identifiers from cookies at checkout and stores them on the order, and adds phone numbers for the 70% of customers who supply one. Three weeks later coverage shows the browser identifier on 81% of server events, the format warning has gone, and the purchase event's rating has moved up two bands.

If the browser identifiers are missing because the site sends server events without the pixel, the Meta Pixel versus Conversions API comparison explains why Meta recommends running both.

Common mistakes

Chasing the dataset-wide picture instead of the optimised event. Adding fields without checking the consent basis. Hashing IP addresses or user agents that Meta expects in plain form. Assuming coverage means usable data when the format is wrong. Treating a regional dip after a consent change as a fault to engineer away.

Frequently asked questions

Which parameter usually makes the biggest difference to match quality?

Meta's guidance and practitioner experience both point first at the browser identifiers the pixel sets, passed through to server events, followed by a correctly normalised and hashed email address. Both tie a server event to a person Meta can recognise. Check the parameter coverage for the event before adding anything, because the gap is often one field rather than many.

Is a low score a problem if the event counts look right?

It can be. An event that arrives but does not match cannot be attributed to an ad or used to build audiences, so reporting understates results and delivery learns from fewer examples. A low score on an upper-funnel event matters less than a low score on the purchase or lead event the campaigns optimise for.

Can I send more personal data to raise the score?

Only data you hold lawfully and have a basis to share for advertising. The score rewards more parameters, but each one is personal data, and the consent the visitor gave or refused applies to the server route as much as to the pixel. Raise the score by fixing formats and passing identifiers you already collect, not by collecting more.

Terms used on this page

Where this page is used

Part of these learning paths

Also related

Sources

  1. About the Conversions API — Meta Business Help Centre
  2. Meta Business Help Centre — Meta Business Help Centre

Related