Meta's Eight-Event Limit Did Not Quietly Disappear

Search for the eight-event limit today and you will find confident posts saying Meta removed it, that prioritisation is automatic now, and that the whole thing is a relic of 2021. Meta’s own help centre still documents the cap and still documents event priority. Both cannot be right, and the difference matters, because if the limit applies and you have ignored it, some of your conversions are not being counted for some of your buyers.

This is worth getting right rather than taking on trust in either direction. What follows is what Meta publishes, what it means in practice, and the check that settles it for your account in about ten minutes.

What the limit actually is

Aggregated Event Measurement is the system Meta built when Apple’s App Tracking Transparency arrived. It is how Meta reports conversions from people who opted out of tracking, and because it aggregates rather than attributing individually, it has to be told in advance which events it should bother with.

Meta’s documentation states that Aggregated Event Measurement limits domains and mobile apps to eight conversion events that can be configured and prioritised. Two details in that sentence do most of the damage in practice.

The first is “domains”. Meta applies the eight-event limit to the primary domain, which it defines as the effective top-level domain plus one — eTLD+1. So client.com, shop.client.com and booking.client.com do not get eight events each. They share one pool of eight between them. If you run ads for a business with a marketing site on the root domain and a store on a subdomain, two sets of events you think of as separate are competing for the same slots.

The second is “prioritised”, which is not a synonym for “listed”. The order is load-bearing, and the next section is about why.

Why the “it was removed” claim is circulating

There is a real reason for the confusion. Meta has made genuine changes to website conversion campaigns over the last couple of years, some of which reduced how much manual configuration a buyer has to do, and coverage of those changes has been summarised, re-summarised and eventually flattened into “the eight-event limit is gone”. Several widely-shared posts now state that outright.

What we can say with a source is narrower. Meta’s help centre currently describes the eight-event limit for optimisation as applying to the primary domain, and separately maintains a page explaining event priority and noting that only prioritised events can use Aggregated Event Measurement data to help improve conversion optimisation. Those pages are the ones linked at the foot of this post. They are not dated on the page, which is a fair criticism of Meta’s documentation rather than evidence they are stale.

So the honest position is this: the limit is documented as current, we have found nothing from Meta retiring it, and the secondary claims that it is gone do not cite Meta saying so. That is good enough to act on, and not good enough to stop checking — Meta rolls changes out in stages and by account, so the only authority on your account is your account.

How priority actually behaves

This is the part that produces the surprise, because it is not a reporting nicety. For a person who opted out of tracking on iOS, Meta counts only the highest-priority event that person triggered. Not all of them. One.

Work through what that means for a normal funnel. Someone clicks your ad, views a product, adds to cart, begins checkout and buys. Five events fired. If your priority order has Purchase above the rest, Meta records the Purchase and discards the other four for that person. That is usually what you want.

Now invert it. Put AddToCart above Purchase — which happens more often than you would think, because the list is easy to reorder and the consequence is invisible — and the same buyer is recorded as an AddToCart. The sale happened. Meta did not count it. Your Purchase volume quietly understates, your cost per purchase looks worse than it is, and nothing anywhere reports an error.

Events outside the eight are not silently discarded altogether. Meta’s guidance is that non-prioritised events remain usable, but with limited delivery and reporting for some iOS users. That is a meaningful distinction: an event off the list is degraded rather than dead, and the degradation applies to a subset of your audience rather than all of it. Which is exactly why it is hard to spot by eye — the numbers are wrong by an amount, not absent.

Two things that eat slots without announcing it

You can be under eight events and still be out of room, because not every slot is spent by something you added to the list.

Value optimisation

Meta’s documentation notes that turning on value optimisation for app or web events consumes at least four of the event slots. Half your allowance, for one setting, in service of value-based bidding. That is not a reason to avoid value optimisation — it is usually worth it — but it does mean a store running value optimisation has roughly four slots left for everything else, and should plan the list accordingly rather than discovering the ceiling.

What does not count against you

Landing page views are not subject to the cap. Worth knowing, because they are the event people most often consider sacrificing to free up room, and removing them buys nothing.

The ten-minute check

Do this per client rather than reasoning about it, because the configuration lives in the account and not in anybody’s memory.

1. Open Events Manager, select the dataset, and find the Aggregated Event Measurement configuration for the domain. Write down the events in their current order.
2. Separately, list what your live campaigns and ad sets are actually optimising for. Not what the strategy deck says — what the delivery settings say.
3. Compare the two lists. Any event a campaign optimises toward that is not in the configured list is the finding. That campaign is optimising on degraded signal for part of its audience.
4. Check the order, not just the membership. If a lower-funnel event sits below an upper-funnel one, you are recording the cheaper action and discarding the valuable one for opted-out users.
5. Count the slots, including the four or more that value optimisation takes if it is on. If you are at the ceiling, decide what to drop deliberately.

On a domain with subdomains, do step one once for the whole eTLD+1 rather than once per subdomain, or you will convince yourself there is more room than there is.

How to order the list

The common advice is to rank by revenue impact, and it is sound as far as it goes: the event that represents money should outrank the events that represent interest. In practice two refinements matter more than the general principle.

• Rank by what you optimise toward, first. If every campaign optimises for Purchase, Purchase belongs at the top regardless of how interesting your other events are. The list exists to serve delivery, not to describe your funnel.
• Remove events nobody uses. Most accounts carry one or two events configured for a test that ended, a campaign type nobody runs any more, or a report somebody asked for once. They occupy slots and contribute nothing.
• Do not reshuffle casually.Reconfiguration affects delivery, so it is a change to make deliberately and then leave alone, not something to fiddle with weekly. Check Meta’s current guidance on what happens to active ad sets before you reorder a live account.

What this does and does not explain

If a client’s Purchase volume in Meta looks persistently lower than their own order data, and the gap is a proportion rather than a cliff, a priority order problem is a reasonable suspect and worth ten minutes before you go looking for anything more exotic.

If the number went to zero on a particular day, this is not your answer. That is a pixel that stopped firing, a Conversions API integration that lost its token, a consent banner recategorising the pixel, or a dataset restriction. Priority order degrades a number; it does not switch it off.

It is also worth saying plainly what the eight-event limit is not: it is not a limit on how many events you may send. You can fire as many as you like. It is a limit on how many Meta will use for optimisation and reporting under Aggregated Event Measurement. The data leaving your site and the data Meta acts on are different quantities, and conflating them is how people conclude the cap does not affect them.

Keeping an eye on it

An Aggregated Event Measurement configuration is a thing somebody set once, probably before you inherited the account. It does not drift on its own, which makes it easy to forget, and it is not visible from outside the account, which makes it easy to miss during onboarding. The audit above belongs in whatever checklist you use when you take over an account.

What does change without anybody telling you is whether those events are still arriving at all. Taglertrechecks every connected Meta and Google Ads account every thirty minutes and emails you when a pixel stops reporting, match quality drops or duplicate events start. It reads what Meta reports about the account, so it will tell you the day Purchase events stopped coming in — it will not read your priority order for you, which is the one in this post you have to go and look at yourself.

If you are auditing an account you did not build, our fifteen-minute tracking audit covers the rest of the sweep, and what ATT actually changed is the background to why any of this exists.