Classification Methodology
How LayoffWatcher classifies every layoff event — with transparency about our evidence standards, sources, and process.
Classification Decision Flow
Layoff event identified
From earnings calls, SEC filings, WARN notices, press releases, or credible news reports.
Does the company cite AI, automation, or technology replacement?
YES → AI-Driven
Direct evidence of AI/automation causation
NO → Continue
Does evidence point to revenue decline, cost-cutting, or financial distress?
YES → Financial Distress
Earnings misses, restructuring charges, bankruptcy
NO → Continue
Is the layoff part of a merger, acquisition, or strategic pivot?
YES → Restructuring
Strategic reorganization, M&A, business pivot
NO → Undisclosed
Insufficient evidence to classify
Evidence Sources
Primary Sources
Official Company & Regulatory Filings
- Earnings calls and investor presentations — Direct company statements about the reasons for workforce reductions
- SEC filings — 8-K, 10-K, and 10-Q filings that disclose material layoffs and their stated causes
- WARN Act notices — Federal and state Worker Adjustment and Retraining Notification filings
- Official press releases — Company-issued statements about restructuring or workforce changes
Secondary Sources
Verified Journalism & Research
- Credible business journalism — Reuters, Bloomberg, WSJ, Financial Times, TechCrunch, The Information
- Industry analyst reports — From firms like Challenger, Gray & Christmas; McKinsey; Goldman Sachs
- Internal communications — When leaked or confirmed by multiple independent sources
Evidence Standard by Classification
AI-Driven
Requires specific evidence that AI or automation directly caused the layoff. This can be: a company statement citing AI, evidence that eliminated roles are being replaced by AI systems, or an AI product launch coinciding with cuts in the roles the product replaces. We do not assume AI causation without evidence — even in tech companies.
Financial Distress
Requires evidence of revenue decline, missed earnings targets, rising debt, restructuring charges, or bankruptcy proceedings. Standard cost-cutting in a profitable company with no stated financial pressure does not qualify.
Restructuring
Applied when layoffs are part of a strategic reorganization not primarily driven by AI or financial distress. Includes mergers, acquisitions, divestitures, business model pivots, or efficiency programs in financially healthy companies.
Undisclosed
Used when available evidence is insufficient to determine the primary cause. This is the default classification — we don't guess. Events may be reclassified as more evidence emerges.
Confidence Level Criteria
High Confidence
Multiple primary sources confirm the classification. The company has made direct public statements. Evidence is unambiguous.
Medium Confidence
At least one primary source or multiple credible secondary sources support the classification. Some ambiguity exists but the weight of evidence favors this classification.
Low Confidence
Classification is based on limited evidence, circumstantial indicators, or a single secondary source. More evidence is needed. These classifications are most likely to be revised.
How the Optimism Note Works
Every layoff event includes an optimism note — a 1-2 sentence note on how AI could positively impact the affected sector or create new opportunities for displaced workers. This is not spin or false positivity. It's a genuine, sector-specific observation about the constructive applications of the same technology that's causing displacement.
We include this because the data shows AI is simultaneously destructive and constructive — and presenting only the destructive side is as misleading as presenting only the constructive side.
Flagging Classification Disputes
We take classification accuracy seriously. If you believe a layoff event has been misclassified — whether as a journalist, company representative, affected worker, or researcher — please contact us:
Email: corrections@layoffwatcher.com
Include the layoff event in question, the classification you believe is correct, and the evidence supporting your position. We review all disputes within 48 hours and publish corrections when warranted.