Meta Pixel, Conversions API and attribution

How do I improve Event Match Quality on my Meta lead events?

Send the Lead from your server with a hashed email, phone, click ID, browser ID and visitor IP, then read the Lead score after 48 hours, not the PageView.

Published

To improve Event Match Quality on Meta lead events, send the Lead from your server with everything the form and the click legitimately provide: hashed email and phone, the click ID, your own external ID, the browser ID, plus the visitor’s IP address and user agent. A lead form can carry both of Meta’s high-priority parameters, so it can reach Meta’s 6.0 recommendation. Wait 48 hours, then read the score of the Lead event, not the PageView.

What the score measures

Event Match Quality is Meta’s score from 0 to 10 for how well the customer information on a server event can be matched to a Meta account. Two ingredients go into it: the quality of the information you send with the event, and the share of those events Meta matched.

  • It applies to website events sent through the Conversions API, with the action source set to website, standard or custom.
  • Every event has its own score. A weak PageView says nothing about your Lead.
  • The window is the last 48 hours, so the number moves with fresh traffic and not with a fix alone.
  • A Pixel-only setup has no score to read. The Browser can send the same values through advanced matching, which helps matching, but the number in Events Manager belongs to server events.

Why it matters is what Meta does with an event it cannot match. Only matched events can be used for ads attribution and delivery optimisation, while unmatched ones stay useful for basic measurement. A Lead that does not match is a lead the system behind your campaigns never learned from.

The parameters Meta ranks, and where a lead form gets them

ParameterMeta priorityWhere a lead landing page finds itHashed
Emailhighthe form’s email fieldyes
Click IDhighthe _fbc cookie, or built on the server from fbclid in the URLno
Phonemediumthe form’s phone fieldyes, digits only with country code
External IDmediumyour own lead or click IDyes
Browser IDmediumthe _fbp cookieno
Countrymediuma form field, or the country code of the phone numberyes
First and last namelowthe form’s name fieldsyes
City and postal codelowonly if the form asks for themyes
Birthdatemediumrarely on a lead form, and not worth a new fieldyes
Facebook Login IDmediumonly if visitors log in with Facebookno
Lead IDlowonly for Instant Form leadsno
IP addressnot rankedthe request that submits the formnever
User agentnot rankedthe same request’s headersnever

The last two rows are not optional in practice: the user agent is required for website events sent through the Conversions API, and Meta’s developer guidance counts the IP address among the high quality parameters. Both are free, because the server handling the submission already has them.

Count what a form gives you: email and phone from the fields, the click ID from the URL, your own ID, the browser ID from the cookie, and two request headers. That is both high priority parameters and several medium ones. The one thing a PageView can never have is the email, because nobody has typed anything yet.

Normalise and hash before sending

Meta’s systems are not designed to accept plain contact information. Normalise first, then hash with SHA-256, because a value that is hashed in the wrong form is a valid-looking string that matches nothing.

  1. Trim spaces, then lowercase the email. Send the hash, not the address.
  2. Send the phone as digits only, with the country code, then hash.
  3. Send the country as a lowercase two-letter code.
  4. Send names lowercased and without punctuation.
  5. Leave a field out when it is empty rather than filling it with a default. A placeholder that repeats across events is worse than a missing value, and Meta reports it as one.
  6. Keep the same normalisation on the browser and on the server, so a value sent by both sides produces the same hash.

Meta’s reference shows the expected hash for its own example values, which makes a useful test before you blame the audience for a low score.

Six fixes, in order of effect

  1. Send the Lead from your server, with the form data. The score is calculated on Conversions API events, so this is where the gains are.
  2. Capture the click and browser identifiers on the landing page visit. If there is no _fbc cookie, build the click ID on the server from the fbclid in the URL and keep it with the lead.
  3. Pass the visitor’s real IP and user agent, taken from the request that submits the form, never your own server’s or a proxy’s address.
  4. Add an external ID, hashed, and use the same format in the browser copy of the event.
  5. Check the event name and ID pairing. A Lead sent twice under different names produces two unmatched events rather than one matched pair, and Meta Pixel and CAPI deduplication shows how to read that tab.
  6. Wait 48 hours, then read the score of the Lead event.

Why the score stays low after a fix

Meta’s Dataset Quality API reports diagnostics per match key, and they map closely onto the usual lead generation mistakes.

DiagnosticUsual cause on a lead setup
Empty IPthe server never copies the visitor’s IP into the event
Mismatched IPyour server’s or proxy’s address is sent instead of the visitor’s
Relay IPthe visitor browses through a privacy relay; complete the rest of the event
IPv4an IPv4 address sent where IPv6 would match better, often because the landing page domain answers over IPv4 only
FBC coveragethe click ID is taken only from a cookie that never gets written, and never built from the URL
Email placeholdera default value filled in when the field was empty
Phone placeholderthe same on the phone field

A wrongly normalised value appears in none of these. It hashes to something valid and matches nothing, which is why a test against Meta’s example outputs is worth ten minutes.

One more thing before you change the form: Meta warns that sending a parameter does not make a match certain. One parameter alone may not be enough, and the person may not have a Meta account. For leads from your own ads the second case is unlikely, since they clicked one of your ads, but the email typed into a form is not always the one on the person’s account. That is why the phone and the click ID go next to it.

What the score does not tell you

A high score means Meta recognised the people behind the events. It says nothing about whether those people were worth calling, and it does not make Meta’s lead count equal to the leads you received. Budget decisions still belong on your own count, as Facebook lead quality sets out.

DropCPL sends the Lead to Meta from the server with a shared event ID, the click cookie and the browser ID, and counts the lead on the landing page that received it, so the score and the cost per lead describe the same submissions. The mechanics of the payload are in Event Match Quality for lead generation, the browser half is in Meta Pixel setup, and the diagnostics behind the score are in what is the Meta dataset quality API.

Questions on this topic

All questions on this topic →

FAQ

What is a good Event Match Quality score?+

Meta recommends aiming for 6.0 or higher out of 10. The bands you see in blog posts, from poor to great, are not Meta definitions and they disagree with each other. Read the score against the parameters you could legitimately send, not against a threshold someone invented.

Why does my Lead score differ from my PageView score?+

Every event carries its own score, because the information available differs. A PageView arrives before anyone types anything, so it has no email and no phone. Judge the score of the event your campaigns optimize for, which for lead generation is Lead.

Does the score improve immediately after a fix?+

No. Meta calculates it on the last 48 hours of events, so you need up to two days of fresh traffic before the number reflects the change. On an event that fires a few times a day, it moves slowly.

Do I need IPv6 to raise the score?+

Meta lists IPv4 sent where IPv6 is available as a diagnostic, because a mismatch costs matches. Its suggested fix is IPv6 support on the web server and in DNS. It is worth doing if the rest of the event is already complete, and not the first thing to change.

Ready to find out what
converts, for real?

Free to start.

Start free →