How this tracker works

Who is counted?

People who signed in with a verified email address or a Google, LinkedIn or GitHub account, and agreed to be included in published statistics.

How is AI involvement classified?

Data privacy thresholds

No chart shows any group with fewer than 10 people. Smaller groups are combined into "Other" (and, on time charts, hidden together with a neighbouring value). A filtered view with fewer than 10 people in total shows no charts. Names, employers, job titles, locations and exact salaries are never published.

Limits of the privacy protections

The 10-person threshold is applied to every chart. Each chart's total is checked against the overall total, so a hidden group can't be determinbed by subtracting one from the other. On "Where people are now", small groups are combined separately among people back in work and people not in work, so the number back in work (shown on the salary chart and the home page) can't be used to uncover a hidden group either. Some small counts can still be worked out in other ways. Comparing results across different filter settings (for example, two overlapping date ranges) can reveal small counts of entries defined by those coarse filters (country, industry, month). Comparing the totals of two outcome charts (for example, people with a time-to-work figure and people with a salary figure) can reveal a small count of people who moved between broad outcomes. Neither reveals names or other personal details. These are known limitations of the current version.

After finding new work

When someone reports they are back in work (re-employed or self-employed), we also ask about the new role: its occupation, industry and company size, how much it involves AI, whether pay is higher, about the same or lower than before, whether it feels like a step forward, a sideways move or a step back, and whether they compromised to land it or held out for the right one. These answers are self-reported and charted from each person's latest update. Updates made before these questions existed aren't counted, and if only a handful of people in work haven't answered, the charts are hidden so that group can't be worked out.

Salaries

Salaries are entered in local currency and converted to US dollars using the European Central Bank reference rate for the month of job loss (or the month of the new job). They are shown only as bands.

Classifications

Occupations use the ILO's ISCO-08 sub-major groups. Industries follow the UN's ISIC Rev. 4 sections, so figures can be compared across countries.

Moderation

Suspicious patterns, such as many entries from one network in a short time, are reviewed. Entries judged not genuine are excluded from the statistics.