Methodology
How PinkSlip calculates the layoff risk score — a transparent, weighted heuristic using public data.
Each company receives a risk score from 0 to 100, where 0 is safest and 100 is highest risk. The score is a weighted sum of six signal categories, plus an additive penalty when there is direct evidence that a job family was targeted.
Every layoff and every community report is weighted by how recently it happened, on a half-life rather than a fixed expiry: a cut announced last month counts for about twice as much as the same cut two months earlier, and its influence tapers to nothing instead of vanishing on a particular day. Counts feed a saturating curve, so the tenth confirmed layoff still moves the score even though it moves it less than the first.
An absent public announcement is not treated as proof of safety. If employees are reporting cuts and no layoff has been confirmed anywhere, the “no confirmed layoffs” signal is downgraded from reassuring toward neutral rather than counting in the company’s favour, because that is exactly the shape an unannounced layoff takes in the data. It also stops being allowed to headline the page: a score driven by the crowd should say so.
Two things make unconfirmed reports a pattern rather than an incident, and either is enough. How many separate people filed them — reports are de-duplicated per person per day, so two reports are two people, and two strangers describing cuts at the same employer is not one event being re-told. And how many separate days they landed on, which is what a rolling layoff looks like and what a single co-ordinated push does not.
Vibe Check votes are the second channel, and they now count on their own. An office voting nervous or panicked with nothing announced and nobody yet filing a report will still read above a quiet employer, because that is a real statement about what the people inside expect. It is deliberately worth less than a first-hand report and is capped lower, since a mood tells you what people fear rather than what has happened. Two anxious votes inside a visibly calm office count for nothing. When both channels agree in the same month, the agreement adds on top of whichever spoke loudest — that pattern takes distinct people on two different channels at one company, and it is the hardest thing here to manufacture.
Signal Categories
Tracks confirmed layoff events, WARN Act filings, and headcount reductions reported in the last 180 days, weighted so recent cuts count for more. How severe a cut is comes from two readings, and we take the worse of them. The first is its share of the workforce, scaled so the bands land where the judgement changes: under 1% is trimming, 5-10% is alarming, and past 10% is severe enough to hold a company in the high-risk band on its own. The second is the raw number of people, because the same percentage does not mean the same thing everywhere — 2% of a megacap is thousands of people looking for work at once, while 2% of a smaller firm is a bad quarter. The size of a cut is carried through everything it touches: a small confirmed layoff still counts as confirmed, but it does not weigh what a large one weighs. This is the strongest predictor — companies that have recently laid off are statistically more likely to do so again. A statutory filing and a tracker reporting the same cut within a couple of days are reconciled into one event, so a single decision is never counted twice.
Layoff reports submitted by employees and job seekers, weighted by recency and de-duplicated per person per day. Two different people reporting cuts, or the same reports arriving across separate days, are each read as a pattern rather than an incident — and a pattern raises risk on its own even when no layoff has been announced publicly, which is how unannounced cuts and quiet performance-managed exits show up here first. Vibe Check votes are a second, independent channel: an office voting nervous or panicked lifts the score by itself, though by less than a first-hand report and up to a lower ceiling. When both channels say the same thing in the same month, the agreement counts for more than either alone.
Monitors stock price trajectory (30-day trend), quarterly earnings surprises, and year-over-year revenue growth. Financial pressure is a leading indicator of workforce reductions.
Tracks whether the company is actively hiring (open roles count) and whether overall headcount is growing, stable, or shrinking. A hiring freeze often precedes layoffs.
Aggregates layoff-related headlines from major tech news outlets, anonymous employee discussions on community forums, search interest trends, and employer review ratings. A quiet week counts as an absence of reporting, not as proof of safety — silence can never push a company's risk to the floor.
Analyzes whether the company has a pattern of repeat layoffs and whether the current season (Q1/Q4 historically higher) increases risk.
Modifiers
A role only diverges from the company baseline when there is evidence about that role specifically — a disclosed layoff naming the team, or reports from people in it. Applied as an additive penalty on top of the company score, never as a discount, and scaled to how deep the cut naming that team was. A sector label such as “Technology” or “HR” describes what a company sells, not who it let go, so it is not read as evidence about anybody’s team. With no such evidence the score is identical to the company baseline, and the page says so.
Almost every source we read is company-wide — WARN filings are US-only law, earnings and news cover the whole business — so a region cannot be scored independently without inventing a difference. Region therefore changes which employee reports are counted as matching, and nothing else.
Each score includes a confidence indicator based on how many data signals are available:
⚠️ Disclaimer
PinkSlip provides a directional risk index based on publicly available data. It is not a prediction — layoff decisions are driven by internal factors that no external model can fully capture. Do not make career decisions based solely on this score. Think of it as a weather forecast, not a guarantee.