Standalone product
LayoffAnalyzer
A 0–100 read on development output.
Why it exists
“Are we shipping enough?” is usually answered by whoever speaks first.
Executives ask the question, engineering answers with anecdotes, and the conversation ends in a headcount argument nobody can check. A score against a fixed baseline moves the argument onto ground both sides can inspect.
What it reads
Windows, not vibes.
- Rolling windows
- Delivery is scored over fixed windows rather than a single sprint, so one quiet week does not move the verdict.
- Per repository
- Each connected repository gets its own score and history. A monolith and a service are not averaged into one number.
- Per developer
- The same windows resolve per developer, so the report shows where the aggregate is coming from.
- One algorithm
- The score is implemented once and shared with Autodevelop. The product does not compute a second percentile.
How it works
Connect, compare, read the call.
01
Connect the repository
An operator connects the GitHub App for your organization. The product lists the repositories that connection covers.
02
Queue the compare
Pick a repository and queue the weekly compare. The job reads delivery history over each window.
03
Read the score
A 0–100 score against a house midpoint of 50, with the table of inputs behind it and the history over time.
04
Get the call
Devrecated or Not Devrecated. Meeting the midpoint is the line, and the report shows what put you either side of it.
The house position
A score is evidence. It is not a decision.
We named it to make a point about the industry, not to hand anyone a termination list. The number tells you where an organization sits against a current baseline. What to do about it is judgment — which is exactly the work we argue does not go away.
Boundaries
What it deliberately does not report.
- No dollar amounts. The product reports a 0–100 score and its history, never a cost-per-developer figure.
- No second algorithm. The score comes from the same package Autodevelop uses, so two surfaces cannot disagree about one repository.
- No raw source code leaves the connection. The score reads delivery history, not the contents of your files.
Access
Owners and executives, explicitly granted.
Sign-in is an organization owner or executive account on the Autodevelop host. Reading the score requires its own grant — it is not implied by general executive insight or organization-management permissions, so access is a decision somebody makes on purpose.
Autodevelop subscribers already have it. See the flagship.
Get the baseline before the conversation.
Tell us how many repositories you want scored and who should be able to see the number.