Meta PIP in 2026: Checkpoint, and what replaced the triggers

Read this first: the rating system changed in 2026

Almost everything written online about Meta PIP triggers describes the PSC scale — a single "Meets Some Expectations" rating putting you straight onto a plan, two "Meets Most" ratings doing the same.

Meta replaced that six-tier scale with Checkpoint, announced in January 2026 and rolling out through mid-year. So those trigger rules are history rather than current process, and anyone telling you what a "Meets Most" means for you now is describing a system that has been retired.

What triggers a plan under Checkpoint is not established by any source consulted. That is an unsatisfying answer and it is the accurate one. If somebody offers you a confident version of it, ask where it came from.

The closest thing on the record is one sentence from the chief executive in the January 2026 reporting: people who are not meeting expectations are typically managed out over the course of a year. A statement of intent, not a rule about ratings.

What Checkpoint reportedly is

Four ratings rather than six, and two formal review cycles a year — a midpoint and a year end. The reported target distribution is roughly:

• Outstanding, around 20%
• Excellent, around 70%
• Needs Improvement, around 7%
• Not Meeting Expectations, around 3%, with no bonus

The stated purpose is to reward output over effort, simplify reviews and cut administration. Every figure in that list is reported rather than official.

Three things the rating rename did not change

Three things are unaffected by the rating names, and they are the load-bearing facts:

The quota machinery. Business Insider reported in December 2022, from two people familiar with the situation, that the quota for Meta's bottom rating categories would roughly double. It reported a May 2025 internal memo instructing that for teams of 150 or more, 15 to 20% land in the lowest bucket, up from 12 to 15%, and describing the midyear review as "an opportunity to make exit decisions". That expectation was re-confirmed in 2026 reporting and carried into Checkpoint's own vocabulary.

Two things belong with that, and this page used to carry neither. The first is Meta's own explanation to managers in 2022. It gave three reasons the bottom of the range went up: it now counted twelve months of departures rather than six, a layoff was folded into the count, and the company would be "more rigorous in our assessment of borderline" cases just above the bottom.

Two of those are accounting and the third is not. So a wider bottom range is not by itself evidence that more current employees were marked down — and it is not evidence that none were, because the same explanation says the line on borderline people was about to get firmer.

The second is a limit inside the May 2025 memo. It is reported to have said there would be no company-wide performance terminations that cycle, with managers handling their own reports instead. That memo has not been read directly here, and of the outlets reporting on it that this page rests on, exactly one carries that limit. Treat it as the thinnest thing on this page.

Non-regrettable attrition. People who have already left count toward the target — the same quota accounting as Amazon's unregretted attrition.

The graduation bar. This is the sharpest reported fact about Meta plans: exiting a plan is said to require output at the "Exceeds Expectations" level, two tiers above the rating that triggered it. It comes from several career coaches and from one engineering manager's account of the twenty-odd plans they ran.

That last one generalises well beyond Meta. A plan that demands materially more than the role's normal standard is structurally exit-shaped, wherever you work. It is worth asking, in writing: which rating level does successful completion of this plan correspond to, and how will it be measured?

Selection criteria, where they were reported

Reported criteria for landing in that bottom pool in 2025 included a below-expectations rating in the current cycle, formal discipline in the prior six months, or an open employee-relations case opened in the first quarter — and an employee-relations case is itself reported to mark somebody already on a managed plan.

Note the scope. These are reported as criteria for the bucket rather than for termination. An earlier version of this page called them selection criteria for the cuts, which is a longer reach than the reporting makes. In February 2025, roughly 3,600 people — about 5% — were terminated in a single performance-based wave, with roles backfilled; leadership framed it as moving low performers out faster rather than as reducing headcount.

Former employees later disputed the label, telling Business Insider they had held "Meets All Expectations" ratings and were moved down after a team transfer, a medical leave or a change of manager. Meta's answer was that these were performance-based terminations, and it denied that retroactive downgrading was widespread — a denial about how common it was rather than about whether it happened at all. Both sides are reported here at second hand, and the disagreement is the point: if this happened to you, the written record of what you were rated before is the thing worth having.

Package figures, and what to do with them

Reported separation terms from the February 2025 wave, secondhand: 16 weeks' base pay plus two weeks per year of service, a payout for accrued leave, six months of healthcare continuation, career support, and immigration assistance for visa holders.

Use those as anchors in a conversation, never as expectations. They describe one wave, reported at second hand, and the same terms may not be on the table.

Do not read a bucket as a verdict, or skip the package conversation

Do not read a bucket decision as a verdict on your ability. The quota arithmetic says otherwise, and internalising it spends a window you need.

Do not skip the package conversation because you were rated rather than laid off. The reported 2025 terms show performance exits carrying structured packages too.

If you are in Dublin

A great deal of Meta's EMEA workforce is, and this is where the American machinery meets Irish law. Irish process requirements are demanding and the argument in an Irish case is almost entirely about procedure. Read the Ireland page — it changes what is worth doing during the plan and what a thin file is worth afterwards.

This is journalism and secondhand account

Fairhanded is not a law firm, is not affiliated with Meta, and has no access to its internal policies. Everything above is journalism and secondhand account. Before signing anything, take it to an employment lawyer where you work.