Of 50 company-authored layoff announcements published between June 2022 and October 2026, 48 (96%) gave a numerical scale for the cuts, but only 25 (50%) quantified exit pay. Everything-PR analyzed announcements from 33 employer brands, captured on October 25, 2026, and found that companies state how many jobs are going far more often than what departing employees will receive.
Published October 31, 2026. Last updated October 31, 2026. All 50 announcements were captured on October 25, 2026. Read each employer's own page for current wording.
A numerical scale means the announcement gives a count of people or roles, a percentage of the workforce, or a workforce target. Quantified exit pay means the announcement states a duration, formula or calendar end-date for pay, such as 16 weeks plus one week per year of service. EPR Research ran the analysis.
The two announcements with no numerical scale are from Dell Technologies and GitLab. GitLab's May 2026 letter said the size of its employee reduction was not yet set, and its 30% figure described country footprint, not headcount. Nokia's October 2023 release is the one workforce target, which moves the company's headcount from 86,000 to a range of 72,000 to 77,000 without stating a layoff total.
The figures cannot be added into a total number of people laid off. Amazon's January 2023 letter gave a number covering both an earlier round and the new one, and PayPal's January 2024 message counted the elimination of open roles in its percentage. Shopify's May 2023 announcement combined a workforce reduction with the sale of its logistics business.
How many layoff announcements quantify exit pay?
Twenty-five of the 50 announcements (50%) quantify exit pay, so half of the messages leave the financial terms unstated or generic. Twenty-nine use the word severance somewhere in the body, a broader measure that includes generic promises and cost references.
Examples of how exit pay was quantified
| Employer | Announced | How exit pay was stated |
|---|
| Google | January 20, 2023 | Starts at 16 weeks of salary plus 2 weeks per additional year (US) |
| Spotify | January 23, 2023 | Average of about 5 months of severance |
| Zoom | February 7, 2023 | Up to 16 weeks of salary (US) |
| Dropbox | April 27, 2023 | 16 weeks of pay plus 1 week per completed year |
| Atlassian | March 11, 2026 | Minimum of 16 weeks plus 1 week per year |
| Epic Games | March 24, 2026 | At least four months of base pay, more with tenure |
| Cloudflare | May 7, 2026 | Full base pay through the end of 2026 |
| HubSpot | October 6, 2026 | 20 weeks of base pay plus 1 week per year, capped at 30 weeks |
Source: Everything-PR analysis of 50 company announcements, captured October 25, 2026. Terms are as stated in each announcement and may apply to specific countries.
A minimum, an average and an upper limit answer different questions, so the table is not a generosity ranking. Geography, years of service, notice periods, equity and benefits can all change what a package is worth, and the analysis does not convert packages into a single number.
Amazon, PayPal, Twitch, Fiverr and GitLab are among the announcements that did not quantify exit pay in the statement analyzed. That does not mean employees received nothing. It means the public message did not say how much.
Support for departing employees named in the 50 announcements
| Item disclosed | Announcements | Share of 50 |
|---|
| Quantified exit pay | 25 | 50% |
| Word severance used in the body | 29 | 58% |
| Health-related support | 30 | 60% |
| Career, placement or alumni networking help | 31 | 62% |
| Visa or immigration support | 12 | 24% |
Source: Everything-PR analysis of 50 company announcements, captured October 25, 2026.
These are disclosure counts. An omitted benefit may have been described privately or on another page, and a published promise does not show that the benefit was delivered.
Did layoff announcement disclosure change by year?
Scale disclosure stayed near-universal in every year of the sample, while exit pay disclosure varied from year to year. The sample is purposive and the yearly groups are small, so the figures describe these announcements and do not show a market trend.
Disclosure by announcement year
| Year | Announcements | Gave numerical scale | Quantified exit pay |
|---|
| 2022 | 6 | 6 | 5 |
| 2023 | 25 | 24 | 14 |
| 2024 | 11 | 11 | 2 |
| 2025 | 3 | 3 | 0 |
| 2026 | 5 | 4 | 4 |
Source: Everything-PR analysis of 50 company announcements, captured October 25, 2026.
Half of the sample (25 announcements) dates from 2023, so that year dominates the totals. Only 2 of the 11 announcements from 2024 quantified exit pay, against 4 of 5 in 2026.
How often do layoff announcements use the word layoff?
Eleven of the 50 announcements (22%) use the word layoff or layoffs in the body, and widening the search to laid off, lay off and laying off added no records. The count includes references to earlier cuts as well as the current announcement.
The employers that used the word include Meta, Stripe, Twilio, Coinbase, Epic Games, Riot Games and Indeed. Other announcements describe the same events as role eliminations, workforce reductions or changes in team size. Vocabulary alone does not show that a message hid what was happening, and the 96% scale disclosure rate shows why a single word count is not a full measure of clarity.
Nineteen announcements (38%) contain sorry or apology wording. That is a reproducible language measure and not a rating of sincerity. Fifteen announcements (30%) explicitly assign personal or founder responsibility for the decisions or circumstances. That category is narrower than general accountability talk, because signing a letter or announcing a decision did not qualify on its own.
Do layoff announcements blame AI for the cuts?
Most do not. Sixteen of the 50 announcements (32%) mention AI or artificial intelligence, and only 5 of those link it to strategy, skills, role sizing or staffing needs. Two explicitly deny an AI connection.
How the 50 announcements related AI to the cuts
| AI relationship | Announcements | Employers |
|---|
| AI not mentioned | 34 | Not applicable |
| AI mentioned as business or technology context | 9 | Google among them |
| AI linked to strategy, skills, role sizing or staffing | 5 | Cloudflare, Dropbox, Atlassian, GitLab, Fiverr |
| AI connection explicitly denied | 2 | Epic Games, HubSpot |
Source: Everything-PR analysis of 50 company announcements, captured October 25, 2026.
The five linked statements say different things. Fiverr's September 2025 letter tied AI productivity to needing fewer people, Cloudflare described an AI-enabled operating model, and Dropbox pointed to a changing skill mix. Atlassian linked its restructuring to investment and skills while rejecting a simple replacement description, and GitLab combined role-sizing language with a denial that its program was an AI optimization exercise.
Epic Games' March 2026 message denied that its layoffs were related to AI, and HubSpot's October 2026 letter denied that AI-related efficiencies drove its reduction. Counting all 16 mentions as AI-caused layoffs would misstate what the employers said. The statements record management's explanations and do not independently establish the causes of job losses.
Do the results hold for employee letters alone?
Yes. Among the 47 public employee letters, 45 (96%) disclose a numerical scale and 25 (53%) quantify exit pay, which matches the headline pattern. The other three records are two corporate press releases (Nokia and Unilever) and one short corporate statement (LinkedIn).
Results for all 50 announcements and for the 47 employee letters
| Measure | All 50 | 47 letters |
|---|
| Numerical scale disclosed | 48 (96.0%) | 45 (95.7%) |
| Exit pay quantified | 25 (50.0%) | 25 (53.2%) |
| Word layoff used | 11 (22.0%) | 11 (23.4%) |
| Apology wording used | 19 (38.0%) | 19 (40.4%) |
| Personal or founder responsibility | 15 (30.0%) | 15 (31.9%) |
| AI mentioned | 16 (32.0%) | 15 (31.9%) |
Source: Everything-PR analysis of 50 company announcements, captured October 25, 2026.
Removing the three corporate statements reduces the influence of genre differences. It does not resolve the sample's concentration in technology employers or its selection bias.
How does this fit earlier layoff communication research?
Earlier peer-reviewed work measured how layoff messages affect investors, while this analysis measures what the messages disclose. Professors Nita Chhinzer and Agnes Zdaniuk of the University of Guelph's Department of Management examined 388 downsizing announcements covered by Canadian news outlets between 2006 and 2015, as the University of Guelph reported on their Journal of Organizational Change Management study. They found that the explanation offered and the person delivering the news were associated with how shareholders reacted.
The two designs answer different questions. The Guelph study looked at share price the day after an announcement, and the Everything-PR analysis looks at the text of company-authored messages. The contribution here is a new descriptive dataset with a transparent disclosure comparison, not a claim to have started the field. More studies of this kind are collected in the EPR research index.
A companion Everything-PR study, Layoff Statements: 97% Quantified Cuts, 12% Support Terms, reviewed 100 official statements from 95 U.S. companies dated 2023 to 2025. It compares employee letters, press releases and SEC filings, while this analysis covers 50 announcements from June 2022 to October 2026 and counts exit pay, apology wording and AI mentions.
For the playbook side, see EPR's guide to internal communications during layoffs and its case study of Boeing's job cut announcement. Teams building employee messaging can start from the internal communications hub, and related coverage sits under crisis communications.
What can this layoff announcement sample not show?
The sample is a purposive selection of 50 announcements and cannot show how all companies communicate layoffs. Forty-seven of the 50 records come from technology, digital platform and video game employers, so the findings describe these messages and not the employment market.
- The analysis does not show who heard the news first or how employees were treated.
- It does not show whether promised support was delivered or how audiences responded.
- It does not rank packages by generosity, because the terms are not comparable.
- Counts of people or percentages cannot be added into a total number of layoffs.
Coding was done by one researcher with a consistency review and no independent second coder, so it is not double-coded academic research. Percentages use a denominator of 50 unless another denominator is stated.
How were the announcements collected and coded?
EPR Research reviewed 50 English-language, company-authored announcements in five batches of ten. Search engines and news coverage were used to find candidates, and accepted evidence came from company websites, employer investor systems or an employer-filed employee letter on the website of the U.S. Securities and Exchange Commission. Full-page text, not search snippets, was required.
The sample was chosen within a January 2022 to October 25, 2026 window, and the records run from June 2022 to October 2026. It holds 47 public employee letters, two corporate press releases and one short corporate statement from 33 employer brands, with no more than two announcements per brand. Parent companies and separately branded subsidiaries can both appear.
The coding definitions, record IDs, source URLs and query log are held in the accompanying workbook, and every published percentage can be recalculated from the dataset. Full source texts are not republished, and a live page can change after capture.
What can communicators do with these findings?
Communicators drafting a workforce announcement can use the 50 messages as a checklist of what readers and employees look for. Each step below follows from a count in the analysis.
- Publish the pay terms in the announcement itself. Only 25 of 50 messages did, and the 8 examples above show that a single sentence with a duration and a formula is enough.
- State what the number includes. Amazon, PayPal and Shopify each combined layoffs with other changes, and a percentage that mixes attrition, open roles or a business sale invites misreading.
- Say plainly whether AI drove the decision. Sixteen messages mention AI and only 2 deny a link, so a reader cannot assume what an unlinked mention means.
- Name who is responsible. Fifteen messages assigned personal or founder responsibility, and the analysis counts only statements that said so explicitly.
Every count on this page can be recalculated from the 50 coded records in the accompanying dataset.
