TL;DR
- Meta Events Manager shows every event a dataset receives; when a dataset is linked to an existing Pixel, the dataset ID used for the Conversions API is the same number as the Pixel ID.
- Meta deduplicates a browser and a server event when the event name and event_id match (with event name plus fbp or external_id as a fallback), within 48 hours of the first event received.
- Event match quality is scored out of 10 per event, and Meta's documentation recommends aiming for 6.0 or higher where possible.
- Server events sent with test_event_code are not dropped: they still count as real events, so the code must be removed before going live.
- An ad account can hold a maximum of 100 custom conversions, and the rules of a custom conversion cannot be edited after it is created.
Meta Events Manager is the Meta tool for viewing and managing the events your website, app or server sends in through the Meta Pixel or the Conversions API. You use it to check that events arrive, that they are not counted twice, and how well the customer data sent with them matches to Meta accounts, before the same numbers are used for measurement and delivery in Ads Manager.
This guide covers where the tool sits, what each tab tells you, and the order to check things in when conversion numbers look wrong. Menu names follow the English interface. Meta renames and moves buttons often, so if your screen differs slightly, go by what each section does.
What Meta Events Manager is and where it sits in a business portfolio
The main unit in Events Manager is the dataset, which collects events from several sources in one place: website, app, offline store data and conversations in messaging channels. For a website, the dataset is what used to be called the Pixel. When a dataset is linked to an existing Pixel, the dataset ID you send Conversions API events to is the same number as the Pixel ID in your site code.
A dataset is an asset of a business portfolio, in the same way as a Page, an ad account or a catalogue. To open Events Manager, use the All tools section in Meta Business Suite on desktop and choose Events Manager. Then check the top of the screen to confirm you are in the right business portfolio. People who manage several brands often open the wrong portfolio and assume the dataset has gone missing.
Inside, the data sources list sits on the left. Click a dataset to see its detail. Access is worth sorting out early. Someone who cannot see a dataset has usually not been granted access to it in the business portfolio settings, and someone with full control of the business portfolio has to add them.
Checking Pixel and Conversions API status in the Overview tab
The Overview tab lists the events the dataset has received, such as PageView, ViewContent, AddToCart, Lead and Purchase, with counts and a chart over time. Look first at the Connection Method column. It shows how each event arrived: from the browser through the Pixel, from the server through the Conversions API, or both.
If both the Pixel and the Conversions API are set up, the events that matter, such as Purchase or Lead, should show both connection methods. An event that shows browser only means the server side is not sending that event yet, or is sending it under a name that does not match.
The counts in Overview are events received before deduplication, and they include every occurrence whether or not an ad was involved. That is why they run higher than results in Ads Manager, which counts only conversions the system can connect to an ad. The gap is normal and does not mean the setup is broken.
Opening the detail for a single event gives two more views. Event Freshness shows how late events arrive compared with when they happened, on a scale from real time to weekly. Event Deduplication shows the rate at which duplicate browser and server events are being removed.
The table below sums up which question each part of Events Manager answers.
| Part of Events Manager | What it shows | When to use it |
|---|---|---|
| Overview | Event list, counts received, connection method and event match quality score | Routine checks that main events still arrive from both browser and server |
| Test Events | Events fired during a test, listed one by one with their parameters | After a new install, a code change or a new checkout page |
| Diagnostics | Problems detected in the Pixel or Conversions API, with recommendations | When event volume drops or a warning icon appears |
| Event Deduplication | Rate of events deduplicated and how often each source sends a deduplication key | When you suspect Purchase or Lead is being counted twice |
What event match quality is
Event match quality is a score out of 10 that indicates how well the customer information sent with a server event helps Meta match that event to a Meta account. It is calculated in real time and shown per event in the Overview tab. Clicking the score opens the detail: which parameters you are sending and which ones to add.
Matching relies on customer information parameters such as email, phone number, the fbp and fbc values from Meta cookies, IP address and user agent. Email and phone must be hashed before sending, as the Conversions API requires. The more kinds of correct parameters you send, the higher the chance of a match.
Meta's documentation recommends aiming for 6.0 or higher where possible. Treat it as guidance and not a pass mark.
The practical ways to raise the score are few. Send hashed email and phone on events where the user fills in a form or pays. Send fbp and fbc whenever the cookies exist. Check that phone numbers include the country code before hashing. For Thai businesses, numbers that start with 0 and lack the 66 country code are one reason phone matching fails.
Testing events with Test Events
The Test Events tab shows events live while you test, so you do not have to wait for counts to appear in Overview. Testing splits into two sides, by connection method.
Testing browser events
Enter your website URL in the Test Events tab and open the site from there. Then follow a real customer path: open a product page, add to cart, submit a form. Each event the Pixel fires appears in the test window. Check that the event name is right, that it fires once per action, and that parameters such as value and currency carry the values they should.
Testing server events
Test Events generates a test ID. Put that ID in the test_event_code parameter of the request you send to the Conversions API, and the server events appear in the test window. Two cautions apply. First, events sent with test_event_code are not dropped. They still count as real events and can be used for measurement, so do not fire fake high-value purchases at a production dataset. Second, remove test_event_code from the payload before going live.
If you use both connection methods, test them in the same run. The test window shows whether the browser and server events were deduplicated, which is much faster than waiting for the deduplication rate to show in Overview.
Fixing duplicate or missing events
Duplicate events: check event_id and event name
When the same event is sent from both the Pixel and the Conversions API, Meta treats the two as one when the event name matches and the Pixel's eventID matches the server's event_id. Deduplication works within 48 hours of the first event received, and the system generally keeps the event that arrived first.
Deduplication usually fails for one of four reasons.
- The browser side and the server side each generate their own event_id, so the values can never match.
- The event names differ, for example one side sends Purchase and the other sends a custom event with another name.
- The same event_id is reused for every order, so real orders get discarded as duplicates.
- Two plugins or tags fire the Pixel on the same page. That is browser-side duplication, and event_id cannot fix it.
There is a fallback method that uses fbp or external_id together with the event name, but it only works when the browser event arrives first. Matching event_id is the more dependable route. Check the result in the Event Deduplication view, which shows a warning when the deduplication rate is too low.
Missing events: check in this order
Meta's Conversions API documentation says server events should be verifiable in Events Manager within 20 minutes of being sent, and its help page for browser and app events says data may take 30 minutes or more to appear. If it has not, work through these checks.
- Confirm the Pixel ID in the code or the request matches the dataset you have open.
- Open Test Events and run a test action. If the event shows in the test window, the code fires and the problem is in the conditions under which it fires in real use.
- Open the Diagnostics tab to see whether the system has flagged missing parameters or badly formatted data.
- Check the cookie consent banner. If it blocks the Pixel until the user accepts, browser events fall in proportion to the people who do not accept.
- For the Conversions API, read the response returned by the request. The error message says whether the access token has expired or the payload is malformed.
How to create custom conversions
A custom conversion is a conversion you define with rules, without changing code on the site. It suits cases where a broad event needs narrowing, such as counting only people who reach the thank-you page for one service, or only purchases above a set value.
You create one from the Custom conversions menu in Events Manager. The steps are: pick the dataset as the data source, pick the event to build on, then set the rules. Rules come in two kinds. URL rules, for example URL contains thank-you. Event parameter rules, for example value or content_type.
Three limits are worth knowing before you start.
- An ad account can hold a maximum of 100 custom conversions. Once full, you have to delete unused ones first.
- After creation you can edit only the name, description and value. The rules cannot be edited, so a wrong rule means creating a new custom conversion.
- A "URL equals" rule counts only URLs that match exactly. Pages with parameters appended are not counted, so "URL contains" is the better choice in most cases.
Meta also reviews custom conversion names and rules that may suggest sensitive information, such as health conditions or financial status. A flagged custom conversion cannot be used in new campaigns. Clinics, insurers and lenders should keep names and rules neutral.
Connecting a dataset to an ad account
A dataset with complete events still cannot be used in campaigns until it is connected to the ad account that will run the ads. If no dataset or conversion event is available to select when you build an ad set, this connection is the first thing to check.
The connection is made in the business portfolio settings, in the data sources section. Select the dataset and add the ad account as a connected asset. The same page controls who can access the dataset, both team members and outside partners. Button names here change between interface versions, but the principle holds: one dataset can connect to several ad accounts, and you need full control of the business portfolio to connect them.
Businesses that work with an agency should keep the dataset in the brand's own business portfolio and share access with the partner. If the dataset was created in someone else's portfolio, the event history and the audiences built from it do not follow the brand when the account manager changes.
Worked example: tracing an inflated Purchase count
This is an illustrative case to show the order of checks. The numbers are not from a real account. An online shop has 100 orders in its back office for one week, but Events Manager shows 190 Purchase events.
- Open Overview and look at Connection Method for Purchase. It shows both browser and server, which is as intended.
- Open the Event Deduplication view. The deduplication rate is very low and the server side is hardly sending the event_id key at all.
- Place a test order with Test Events open. Two Purchase events appear per order and neither is marked as deduplicated.
- Change the server side to use the order number as event_id, the same as the Pixel, and retest until one of the two is deduplicated.
In a case like this the extra count affects more than the report. Ad delivery uses the same events as its signal. If Purchase is double counted, the cost per purchase shown in Ads Manager is lower than the true figure.
What this means for marketers in Thailand
Many Thai businesses close sales in chat, on Messenger, Instagram Direct and LINE, more often than on a website checkout page. On that path the Pixel alone sees only the early part of the purchase. A dataset can take events from Meta's messaging channels and offline events too, but a sale that finishes in LINE or by bank transfer only arrives if the business sends it in through the Conversions API or an offline upload.
Consent under the PDPA is the other point. A site whose cookie banner blocks marketing tags until the user accepts will record fewer browser events than its real visitor count. Events sent through the Conversions API have to respect the same consent choice, so the server route is not a way around it. How much goes missing depends on each site's banner design, and there is no single figure that applies to every business.
Finally, Events Manager shows only what Meta received. It does not tell you what you sold. Always compare it with a second source, such as the shop back office or Google Analytics 4. Teams without dependable web measurement can see the scope of the Google Analytics 4 setup and migration service. Campaign structure and measurement on Facebook and Instagram fall under Relevant Audience's Facebook Ads service and social ads service.
Frequently asked questions about Meta Events Manager
What is the difference between a Pixel and a dataset?
Dataset is the broader term: it groups events from website, app, offline and messaging sources. The Pixel is the code on the website that sends browser events into the dataset. When an existing Pixel is linked to the dataset, the dataset ID is the same as the Pixel ID.
Why do event counts in Events Manager differ from Ads Manager?
The two screens count different things. Events Manager counts events received before deduplication, whether or not an ad was involved. Ads Manager counts only conversions it can connect to an ad within the attribution setting in use.
What event match quality score is enough?
Meta recommends aiming for 6.0 or higher out of 10 where possible. Upper-funnel events such as PageView, which carry no email or phone number yet, normally score lower. Put the events your campaigns optimise for first.
Can Test Events be used on a live dataset?
Yes, but server test events sent with test_event_code still count as real events. Test with low-value transactions, test only as much as needed, and remove test_event_code from the payload before going live.
If you want a team that handles both event setup and ad delivery to review your dataset, see Relevant Audience's Facebook Ads service page.
