Complexity was bought. Coherence was not.
Every company runs on decisions nobody wrote down: who owns the customer record, which number is the real number, what happens between a signature and a renewal. Software does not settle those questions, and neither does a reorganisation.
When the questions stay open, the business pays in friction — slower cycles, arguments about data, work done twice. The evidence sits in public, in the job specifications companies write when they decide what work matters. So we read them, and we score them.
- Read a posting before it reads you.
- Compare a band against its region.
- Name the flags in an interview.
- Count the platforms you are asking for.
- Split the req when both scores climb.
- Put the first ninety days in writing.
Read like a rate, not a part count
A chronometer is not certified on how many parts it holds. It is certified on how far it drifts per day. We score businesses the same way.
Counts the disconnected platforms one role must hold together. A reading of complexity bought and not yet resolved.
Counts the distinct skill categories demanded of one person. A reading of design a company asks a single hire to carry.
Hands show the current market averages on a 0–100 scale
Five functions, one train of wheels
Marketing
A definition of a qualified lead that survives contact with Sales.
Sales
A pipeline that reflects reality without a weekly negotiation.
Customer Success
A handoff that carries the promise made during the sale.
Finance
One forecast, one source, and no reconciliation ritual.
Product
Usage evidence that reaches the roadmap before the churn does.
Assembled once, running since
Four stages, no human in the loop. The pipeline was built like a movement: each stage does one thing, hands off cleanly, and is inspectable when it drifts. Open a stage to read it.
A scheduled n8n workflow pulls RevOps postings daily from LinkedIn and other sources. The ingestion layer runs without supervision, so the record grows whether or not anyone is watching.
LinkedIn via Apify · Adzuna API · SerpAPI
Four postings, opened up
Each one read as a design document: what the company believed, what the belief cost, and what a corrected specification would say. Pick a case from the index.
Score shown is Unicorn JD · Illustrative annotations, wire to scored postings
One req, four jobs
The posting asked one person for Salesforce administration, SQL, Tableau, territory planning, enablement content and vendor management — six skill categories against a single mid-market band.
The stack is defensible; the role is not. What the company could not decide internally, it published as a requirement. Split into an administrator and an analyst, both scores fall into workable range and the work becomes ownable.
Red flags detected- Six skill categories in one req
- Engineering work, non-engineering band
- No named owner for data quality
- Band below regional median
Open, checkable, yours to use
Both index scores for every posting in the record, with the red flags that produced them.
Regional breakdowns, score distributions and tool frequency across US, UK, DACH and EU.
The method, stated in full — so a number can be argued with rather than believed.
No sponsors, no affiliate links, no paid placements. Tools rank by market demand only.
Cut the parts to fit, and the business keeps time.
Count the platforms. Name one owner per number. Cut what you cannot justify, and put the first ninety days in writing. Then check your own specification against the record before the market does it for you.