The employer brand
intelligence database
A standing record of what the world's largest employers publish about working for them, and what their employees say back. Built and owned by Additive Force from public data.
Employer brand intelligence databaseLast readaugust 7 2026What it is
The Employer Brand Intelligence database is a continuous record of employer brand activity at major multinational companies
It holds what those companies publish about working for them, on the channels they own, alongside what their employees say about working there in public reviews.
Both sides are collected continuously
On the publishing side, every post on a company's owned pages is recorded and classified by who it speaks to and what it is for.
On the employee side, public reviews are mapped to a single fifty-attribute taxonomy, so that what people value — and what they complain about — can be compared across companies, industries and markets rather than read one company at a time.
It exists because the two halves are rarely held together
What a company says about itself as an employer is public. What its employees say is public. Keeping both in one structure, at scale, is what makes the distance between them measurable.
Additive Force builds and owns the database
It is not licensed from a third party and it is not a syndicated panel. No one else is reading from the same source.
Scope
Channels
LinkedIn, Instagram and YouTube: company-owned pages, not careers or "Life" pages
Employee voice
Public employer reviews, mapped to a custom-made global attribute taxonomy
Companies
Major multinational employers, currently weighted to North America, Europe and India
Period
May 2025 to present, collected continuously
What gets recorded
| Category | What it holds | Scale |
|---|---|---|
| employer publishing | Posts on company-owned pages, each classified by content type | 132,256 |
| audience response | Reactions, comments and shares, per post | every post |
| employee reviews | Public reviews, mapped to fifty EVP attributes | ~475,000 |
| company reference | Industry, employee count, headquarters, follower base | 375 |
How it updates
Collection runs continuously and the index refreshes weekly
The panel grows as companies are added. Every figure on this page carries the date it was last read.
What it is not
Not a survey
Nothing here is self-reported to us. Every record is something a company or an employee published themselves.
Not a one-off study
It is a standing record that keeps growing, not a piece of research with a fieldwork window.
Not a claim about any platform as a whole
Findings describe the employers in the panel and what they publish on their own pages.
Not a ranking or a score
We record what was published and how audiences responded. Where we compare companies, we say so and show the basis.
Not evenly deep across channels
Coverage begins at different points on each channel and the volumes differ. Comparisons across channels are made like-for-like, on companies active on all of them, over the same window — and we say when that is what we have done.
Not complete at any moment
Companies still being indexed are excluded from measurement until their pages are finished, so a mix is only ever measured on complete records.
How our articles use this data
This page describes the database as it stands today
Our articles do not. Each article carries a frozen snapshot of the figures it used, taken on the day it was published and never quietly changed afterwards.
That means an article's numbers and this page's numbers will differ, and the gap will widen as the panel grows. This is deliberate.
A figure published in an article is a claim about what was true when it was made, and it stays auditable against the snapshot that shipped with it. If we revisit an article, we re-run the snapshot and declare it.
This page is current
Refreshed weekly. Moves as the panel grows.
Articles are frozen
What was true at the date of publication.
How these articles are made
Automated analysis with three human-in-the-loop stages
Each article starts as a question put to the database. A large language model pulls the figures and drafts a preliminary outline. A human reviews the outline, refines the scope and sets the direction. An LLM writes a first draft based on the approved structure, which is then edited by a human. Nothing is published without human judgement.
Mechanical calculation and numerical validation
All charts and calculations are generated from a frozen snapshot of the database taken at the time of writing. Final numbers are mechanically validated against the datafile prior to publication. No claims are made that the numbers cannot show. Where data is thin, we say so.
Automated data analysis
Figures, the first draft and every chart are generated from the data.
Human editorial judgement
The angle, the approval to write, and the final read are a person's.
Citing this data
The data is free to cite, quote and reproduce with attribution
Charts published in our articles carry their own source line and can be downloaded, including the underlying figures as CSV.
Additive Force. (2026). Employer Brand Intelligence Database (edition 1.0). additiveforce.com/journal/methodology
Research record
No spam. No sales.Just insights.
Get Additive Force articles sent right to your mailbox.