Methodology
Where the numbers come from and how to read them
Sources — open data
- Career histories. Public professional profiles and resumes from open sources: where a person worked, from which month to which, in what title. They make up the HÜNT Workforce Panel — a monthly view of hiring across more than a thousand employers.
- Job postings. Open job ads. A pay reference is shown only when the range is stated in the posting itself — we do not estimate it.
- Companies. Name, sector and stock ticker of public companies — from open reference data.
We do not use private employer data, reviews or “culture” ratings. We do not make up revenue, valuations or salaries.
Estimates from change, not levels
Not every employee is visible in public profiles, and the visible share differs by company and country. So the raw “how many people were hired” is almost always understated and hard to compare across companies. We show what is more robust to that bias:
- Hiring percentile — the company's rank by hiring intensity among all panel employers this month: 100 hires the most actively, 50 is in the middle.
- Δ 12 mo — how the percentile changed over a year. This is the main “accelerating / slowing hiring” signal.
- Joined and left — people who started and ended work at the company over 36 months; balance is the difference. A move to another title inside the company is not an exit.
- Tech share — technical titles among those moves.
The latest months are always under-counted: people update profiles with a delay, so recent hires show up later. Recent cohorts are not compared directly with mature ones.
What a proxy is
A proxy is an indirect signal where no direct measurement exists. For example, “became senior” is read from a word in the job title, and “director or above” from the title. Such numbers are labelled as proxies and have known blind spots: internal levels without “senior” in the title (such as SDE III or E5) are invisible, and subsidiary brands count as separate companies.
How to read n
n is how many people (moves) stand behind a number. The smaller n, the less you should trust it:
| n < 10 | count only; percentages and distributions are hidden — “not enough data” |
| n = 10–29 | thin evidence, read as a hint |
| n = 30–99 | medium evidence |
| n ≥ 100 | more reliable — subject to coverage |
Coverage (high / medium / low) is separate from n: how many of a company's employees are visible in public profiles at all. A large flow with low coverage can still mislead. We do not hide low-coverage companies, but we mark them and keep them off the map.
Text written by a model
“Why it fits” explanations, message drafts and the resume review are written by a language model from your data and the panel numbers. The model can be wrong — check before you send. There are no opaque overall scores (“employer rating”, “match score”) in the service.
For employers: remove my data
All company data in HÜNT Career comes from open sources. If you represent a company and want it removed from the service, email us from a company address with the company name, website and what should be removed. We will reply and remove it.
Address: help@huntshare.tech
Write an emailA person who wants data about themselves removed can write to the same address — see “Your rights” in the Privacy page.