Seniority mix by role
Seniority is inferred from title keywords, so "mid" means "neither junior nor senior stated". Entry programmes (Ausbildung, duales Studium, Werkstudent, Praktikum, Trainee) are excluded from the dataset, so junior counts reflect junior-titled permanent roles only.
Role and seniority, side by side
The same data as a heatmap. Darker means more postings.
Does seniority change how long a job stays open?
Average days open by seniority level.
Which data role is which?
Roles are classified from the job title, not from the search that found the posting. Volume shows where demand is; average age shows which roles employers struggle to fill.
Absolute postings per job title
Average days open per job title
How old is an active posting?
Days between the posting date and today, across all three countries. The average is – days. The median is –. That gap is the whole story: half the market moves in weeks, and a long tail of very old listings drags the average up.
How many are still open after X days?
A classic survival curve. Of every 100 live postings, how many have already been open at least this long. The steep drop in the first month is healthy churn. The flat tail is what the average is really measuring.
Three countries, one language, but clear differences
In the figure below, the main insighjts for every country can be seen
Which country asks for which role?
Role mix as a share of each country's postings, so a small market is comparable with a large one. This is where the three countries actually differ.
Where are the jobs located?
Bubble size is the number of live postings, colour is the country. Hover any bubble for the city name and its exact posting count. Postings with only a country-level location are excluded from the map.
Top cities by listings
| City | Listings | Avg days |
|---|
Compare two cities head to head
Pick any two cities carrying at least 20 live postings. Each axis is scaled against the highest value across every city, so the shape shows relative position rather than raw units, and the two shapes are directly comparable. The exact figures are in the table beside it.
Actual figures
Most mentioned tools read the details
The API hard-caps descriptions at 500 characters and 99.5% of postings are truncated. This measures tools mentioned in the title or opening paragraph, not tools required. These are floors, not requirement rates. The truncation window is identical for every posting, so comparisons between tools hold; absolute percentages do not.
Which tools does each role ask for? not every tool is extracted
Pick a role family below. The left chart is the share of that role's own postings mentioning each tool. The right chart is the same figure minus the market-wide rate, so it shows what makes a role distinctive rather than what is simply common everywhere. Denominators are per role, so analytics engineer with 69 postings stays comparable with AI / ML at 1,435.
Not every skill in a posting is extracted, so read these as floors rather than requirement rates. The API hard-caps descriptions at 500 characters and 99.5% of postings are truncated, so any tool named further down an advert is never counted. Anything outside the curated 47-term dictionary is invisible too. The truncation window is identical for every posting, so comparing tools with each other holds; treating any single percentage as the real requirement rate does not.
Share of this role's postings mentioning it
Difference from the market-wide rate
Tools by category
The same mentions grouped into cloud, database, BI, processing and so on. Same truncation caveat applies.
Are the popular tools also the stuck ones?
Every tool in the dictionary, plotted by how often it is mentioned (across) against the average age of the postings mentioning it (up). Bubble size is the number of postings. High and left means a niche tool attached to roles that sit open for months; low and right means a common tool in roles that move. Click any category in the legend to show or hide it, and hover a bubble for the exact figures. The same truncation caveat applies.
Agencies vs direct employers
Split by whether the poster looks like a recruitment agency. Detection is keyword based and flags only –%, which is almost certainly an undercount. Treat it as a lower bound.
Which company has the most listings?
Employers by number of live postings.
How concentrated is each market?
Average number of live postings per employer. A higher number means fewer, larger employers doing the hiring; closer to 1 means a long tail of companies each advertising a single role.
Who publishes a salary?
Share of postings stating any salary figure. The EU pay transparency directive had a transposition deadline of June 2026. Swiss postings quote francs and German and Austrian ones euros, so only the disclosure rate is comparable, never the amounts.
Week over week
Tracked metrics since collection began. Appears once there are three or more weekly snapshots.
This pipeline's own error rates
Most dashboards do not publish how wrong they might be. These are measured on every run: the share of records rejected by quality rules, the share of source descriptions the API truncated, and the accuracy of the tool matcher where it has been measured.
What this does not tell you
- Postings are not hires. An old posting may be a real unfilled role, a pipeline-building advert, or a listing nobody took down.
- One aggregator is not the whole market, and its coverage is not equally deep in all three countries. Cross-country comparisons should be read as indicative.
- The posting date is the aggregator's, which may be when it indexed the job rather than when the employer published it.
- Swiss salaries are in francs and are never averaged with euro figures. Only the disclosure rate is compared across countries.
- Roles are classified from the job title with keyword rules. The AI/ML family includes software engineering roles that merely mention AI, which inflates it.
- Entry programmes are excluded (Ausbildung, duales Studium, Werkstudent, Praktikum, Trainee), along with Controlling and finance roles and data-centre infrastructure jobs pulled in by keyword search.
- Reposts are detected by content hash (title, company, city, first 200 characters).
- Agency detection is keyword based and undercounts.
- Only aggregates are published here. Individual listings are not redistributed.