A number is only as good as the method behind it

Most salary benchmarks match a job title to one dataset and hand you a figure. Titles don't mean the same thing twice, and no single source tells the whole story — so the number looks precise, and quietly misleads. Three principles sit behind every benchmark we produce.

01Beyond titles

Titles are inconsistent. Roles aren't.

A "Director" at one company is a "Senior Manager" at the next. We read the full job description — scope, responsibilities, seniority, skills — to understand what a role actually is before benchmarking it.

Senior Product Manager

Job description · reading role, not label

reading

You'll own the product roadmap end to end, lead a team of six engineers and two designers. Requires 8+ years in B2B SaaS, fluency with SQL, experimentation and stakeholder management.

ScopeResponsibilitiesSenioritySkills

02Multi-source

One source is one opinion.

Every benchmark request puts eight specialised AI agents to work in real time. They search across dozens of sources, gather hundreds of data points, keep only the ones that genuinely match your role, and converge on a single answer. See how that compares to traditional salary surveys.

Live job ads

Historical offers

LLM research

Salary guides

Public data

Onenumber

03Personalised

Apples to apples, every time.

The same salary means something different at a 15-person startup and a global bank. We compare each role only against companies that look like yours — same industry, size, and growth stage.

Where every figure comes from

Live job ads

Scanning live online job ads from a variety of sources that contain salary information and related benefits.

Historical offers

Anonymised data captured by EvenBetter as users run salary benchmarking reports.

LLM research

Running salary queries across a variety of large language models.

Salary guides

Triangulated salary information across publicly available salary reports and industry research.

Online information

Salary signals gathered from across the public web — company pages, industry commentary, and community discussion.

Public databases

Publicly available salary information from industry bodies and government surveys.

Signal Strength

Built bottom-up from quality and quantity.

Before any number reaches your report it passes through two scoring layers that decide how much weight it carries.

Step 1

Quality Score

Every individual data point is given a Quality Score based on relevance and recency.

Step 2

Signal Strength

Every data source is then assigned a Signal Strength rating — Low, Good, or Excellent — based on the Quality Score of the data points it provided and the Quantity of data points found (more data points means a stronger signal).

Anti-promises

What we don't do

Discipline as a feature. Each anti-promise below is a deliberate product choice — saying no to one thing makes the rest more honest.

We don't fabricate single-point estimates — every range is min/median/max with explicit spread.

We don't hide Low signal strength reports in fine print — wide ranges and 'thin market' explainers are surfaced prominently.

We don't share your JD with anyone — PII is redacted via Haiku pre-pass before any persistence; raw text purged after 30 days.

One number you can stand behind

Read the role, not the label. Weigh every source, not just one. Compare like for like. Every benchmark shows its working — the sources behind it, and how confident we are.

See EvenBetter on a live role