Meta event match quality: understanding a 6/10 score and improving the inputs
A Meta event match quality score of 6/10 does not judge your product or predict the profit from your next advertisement. It invites you to examine the information accompanying an event and whether that information can help connect it with a Meta account. A beginner merchant in Tunisia should start with data genuinely available at each
A Meta event match quality score of 6/10 does not judge your product or predict the profit from your next advertisement. It invites you to examine the information accompanying an event and whether that information can help connect it with a Meta account. A beginner merchant in Tunisia should start with data genuinely available at each stage, then investigate information being lost. You can do this without launching a campaign, using legitimate visits and orders. Invented contact details and artificial traffic are not improvements.
Separate four questions before changing settings
Ask whether the event arrives, whether it arrives at the appropriate time, whether multiple versions represent one operation, and whether useful matching information accompanies it. These are different questions. Reliable server delivery can coexist with limited customer information. A detailed event can still be duplicated. Record the event name, source and reporting period beside every observation, rather than treating all diagnostics as one score.
A visitor viewing a page may not have provided a telephone number. An order can make that information available later. Comparing PageView with Purchase as though they had identical inputs is misleading. Check monetary value and currency too: extra identifiers cannot repair an event reporting the wrong commercial amount.
Know what the current Matjri integration actually sends
The store integration tracks browser activity including page visits, product views and the beginning of interaction with the order form. The configured order event can also be sent from the server at placement, confirmation or delivery. Enabling Conversions API does not mean every browser navigation event is automatically mirrored by the server. Order information can supply a telephone number and other genuinely available details; an anonymous visit may not have those same inputs.
Check the Pixel ID, product scope and saved token in the integration settings. Matjri helps organize products and orders, giving you an identified operation to trace. Start there before adding another tracking installation. Additional code can complicate diagnosis instead of solving a problem that has not yet been identified.
Use a five-step audit
- Select one event and a fixed reporting period, and write down what business action the event represents.
- Review the availability of customer information reported in Events Manager, distinguishing browser and server sources.
- Compare a known order with its corresponding event: time, available phone number, amount and currency. Keep customer details out of public notes.
- Check that each field is genuine and consistently formatted, then repair its source when evidence shows a problem.
- Record the change and its date. Compare new, similar events rather than assuming an old reporting window immediately reflects your correction.
Do not borrow someone else’s telephone number or place the same invented email address on every order. A real first name is preferable to a guessed surname, and missing information is preferable to false information. Hashing a contact field does not remove the need to use it appropriately for how it was collected and what the customer expects.
A fictional example with ten orders
Suppose a shop receives ten orders worth 550 TND in total. Telephone numbers are present in all ten order records, but an event review appears to show a phone field in only five cases. The useful question is not how to move the score from 6 to 8 immediately. It is why information that exists might be missing from half the relevant events. Different reporting periods, an older sender or comparison with different actions are possible explanations. None should be presented as the confirmed cause without evidence.
| Observation | Next check |
|---|---|
| Phone in order but absent from event | Sending source, reporting period and field format |
| Browser identifier present, phone unknown | Was this a visit before form completion? |
| Low Click ID availability | Did incoming links genuinely contain fbclid? |
After a repair, compare new orders of the same kind. The 550 TND remains order value rather than earned profit. Some orders may be cancelled, and delivery or collection may occur later. Better measurement explains these stages; it does not change the underlying commercial result.
Useful improvements without advertising
You can verify a token, remove repeated installation, choose a consistent order-event time and check the telephone format, currency and destination domain. You can also improve how clearly the form explains the information needed to buy. Adding an unnecessary birth date or demanding another contact field only to satisfy a technical indicator may make purchasing harder. A small event sample also makes score changes difficult to interpret, so avoid attributing every weekly movement to your latest edit.
When delivery is accepted, values are correct and no information is missing without explanation, return to customer questions: product suitability, dimensions, shipping and mobile ordering. Follow one order from its product page through delivery. If you are preparing that workflow, organize your products and orders with Matjri before connecting measurement to a clearly defined purchase journey.
Frequently asked questions
Must I reach 10/10 before advertising?
No single score proves that every store is ready. Start with genuine events, clear definitions, correct values and appropriate deduplication. Separate information that is naturally unavailable from information your implementation is actually losing before deciding whether to spend.
Does missing fbc mean tracking is broken?
Not necessarily. It depends on genuine click provenance. Direct or search visits may not contain it, while some organic Meta links can carry fbclid. Examine the actual journey and do not manufacture identifiers to improve a percentage.
Why might Purchase score higher than PageView?
A completed order can provide details that were unknown during an initial visit. Compare information availability, event types and reporting periods. The difference alone is not evidence of failure; a genuinely available field disappearing from the same operation deserves investigation.
Sources and review
Editorial review: 15 October 2026. Numerical examples are illustrative learning exercises, not promised results or market prices.