Google PIP: GRAD ratings, the retired check-in, and the plan

Every figure here is reported, and the stage this page was named for has been retired

Google does not publish this process. What follows comes from coaching firms with large Google caseloads, journalism and employee accounts.

One caution specific to this page, and it is a correction. The Support Check-In, the documented warning stage this page was written around, is reported to have been discontinued in August 2023. A coaching firm reported the phase-out as planned for that half-year at the time, and a 2026 account states that it happened and that the signal before a plan now comes through GRAD's ordinary cadence of semi-annual reviews and monthly check-ins. No statement from Google was found either way. Two earlier checks could not settle the question; two sources now point the same way, and this page follows them. Do not assume a friend's account from a couple of years ago still applies.

GRAD, and where a plan comes from

GRAD — Google Reviews and Development — rolled out in 2022: annual formal ratings on a five-point impact scale, from Not Enough Impact through Moderate Impact, Significant Impact (the designed norm), Outstanding and Transformative, with quarterly goal-setting, a midyear checkpoint and regular manager check-ins around it.

A formal plan is reportedly triggered by a Not Enough Impact rating, or by persistent Moderate Impact across cycles. That part is consistent across sources and dates.

At the launch of GRAD, reported expectations moved from about 2% of employees in corrective territory to around 6% at the warning stage plus 2% at the lowest rating, while the top two tiers shrank from about 27% to 22%. When employees asked leadership whether this amounted to managed-attrition quotas, leadership said no. Both the question and the denial are on the record.

A separate change lands in the 2026 compensation cycle: an internal note from the VP of global compensation and benefits sets out an approach letting managers award the top rating to more people. That says nothing directly about the bottom of the distribution, but it is evidence GRAD is still being tuned — which is the reason to keep checking the stage names rather than trusting a secondhand account.

The warning stage, without its old name

The pre-plan stage used to be the Support Check-In: structured, documented meetings signalling that performance was trending toward the lower ratings, which coaching firms mapped onto Amazon's Focus. With it retired, the warning is reported to arrive through the ordinary review cadence instead — more frequent, documented feedback rather than a named stage.

A figure circulating from employee forums claimed that around 95% of people who reached a formal plan had passed through the check-in first. It described the retired stage and was anecdote-grade even then. The sequencing claim underneath it — a documented warning precedes a formal plan — is still consistent across sources, in this weaker form.

The formal plan itself is a signed document with written goals, a defined timeline and a binary outcome. How long it runs is not settled. The accounts this page was first built on reported three to six months; a 2026 account reports thirty to sixty days, shorter than the sixty to ninety that is common elsewhere. Both are reported and neither is corroborated, so this page no longer states a runway. Read the deadline off your own document.

One more reported detail from the 2026 account: Google often presents a severance package in the same meeting as the plan, at a reported two to three months' pay. Treat the figure as anecdote. Treat the sequence as worth noticing — the account's own reading is that an exit package at the start of a plan is not the start of a genuine improvement process, which is what this guide says about that sequence everywhere else.

Telling a support conversation from a formal plan, at any employer

This is the most portable thing on this page, and it applies at any employer:

• More frequent one-to-ones with documented feedback is a warning stage.
• A signed document with goals and a deadline is a formal plan.

Different situations, different clocks, different sensible responses. The Google reporting makes the binary explicit, but the question is worth asking wherever you work: is this a support conversation or a formal plan — and if it is formal, may I have the document and the timeline in writing?

What the retired stage changes about the advice

An earlier version of this page said the warning stage at Google carried unusually good recovery prospects, on the strength of a longer runway and a named, documented check-in. Both supports are gone, so that claim goes with them. What remains is the ordinary rule: the stage before a signed plan is the recoverable one, and the mistake to avoid is treating a warning conversation as doom and disengaging before any plan exists.

Once a signed plan with a deadline exists, the ordinary analysis applies: work the plan, keep a dated record, and run a parallel search on a pace set by the deadline the document actually states rather than by a runway somebody reported.

An impact ledger is the right artifact here

GRAD is built on impact language, so a weekly dated log of shipped work mapped to that vocabulary is worth more than a general diary. It is the form the rating system already reads in.

If you hold a work visa

Visa holders surface repeatedly in Google threads asking what a low rating means for their status. The answer is that a rating and a plan have no immigration effect whatsoever. Status turns on employment ending. If it gets that far, the visa page has the clock, and the plan's own deadline — not a reported runway — is what to plan around.

Dublin changes this materially

Fairhanded is not a law firm, is not affiliated with Google, and has no access to its internal policies. If you are in Dublin, Irish procedural law changes the picture materially — read the Ireland page too.