Where the numbers come from
Jobvin runs no salary database of its own and does not average what users type in. The values come from the Entgeltatlas of the German Federal Employment Agency, and the result page names that source.
Entgeltatlas of the Federal Employment Agency
The basis is the gross monthly pay of full-time employees, taken from official employment statistics. Not from self-reports, not from job ads, not from surveys.
A comparison group, not a job title
Your free text is mapped to an official occupation and from there to the five-digit key of the German classification of occupations. Which group was hit always stands above the numbers - a silent mismatch would be worse than a visible one.
Region is an axis of its own
There are 17 regions to choose from: Germany as a whole and the 16 federal states. For each one the check queries the same comparison group again. The regional comparison is therefore a real comparison, not an extrapolation from a national figure.
When the statistics have nothing
If there is nothing for your region, the source answers nationally. The note about it then sits above the result and survives a reload, because it travels in the saved row. If there is nothing for the occupation at all, the check stops - no neighbouring occupation is substituted.
- Assistant
- Skilled worker
- Specialist
- Expert
The level sits in the fifth digit of the occupation key and is therefore fixed by the occupation itself. It is not something you set alongside, and your years of experience do not move it.
The sample size is shown
Above every distribution it says how many employees it rests on. Below 1,000 the result explicitly flags the thin data instead of letting the values look as sharp as those of a mass occupation.
What we compute from the numbers - and what we do not
For a comparison group the official statistics deliver exactly three values: lower quartile, median and upper quartile. Everything you see beyond that is modelled from them. This is not small print, it is stated on every single number.
The curve is a model, and it has a name
From those three values the check fits a log-normal distribution: the median sets the centre, the distance between the lower and the upper quartile sets the spread. The histogram draws ten bars over that curve, nine on narrow screens - recomputed then, not simply hidden. A hidden bar would silently be a different distribution.
Rounding happens in the open
Your market value is rounded to full 100 euros of gross annual pay, the bar boundaries to full 1,000 euros. Bar heights are normalised to the tallest bar: absolute density values would suggest an accuracy that three support points cannot deliver.
Twelve monthly salaries is a convention
The source delivers monthly values. The check scales them to a year using twelve salaries, without bonuses and without extra payments. That sits as a footnote under the figure, because it is an assumption and not a statistic.
What the salary check is not
It does not predict what your next employer will pay. It knows neither the company nor the industry nor any collective agreement, and it knows nothing about your benefits. It tells you what your official comparison group earns in your region. Everything else you negotiate.
- under 2 years · entry level20th percentile
- 2 to 4 years · early years35th percentile
- 5 to 8 years · experienced50th percentile
- 9 to 14 years · highly experienced65th percentile
- 15 years and up · top tier75th percentile
The requirement level is already baked into the occupation. Experience must not shift it a second time - it only decides at which point of the range your market value is read off. Five steps, fixed in the product:
And where there is nothing to compute, we do not compute
For some occupations the statistics only deliver the median. Lower quartile, median and upper quartile then collapse into the same value, and no distribution can be fitted to a single point. Exactly then the whole distribution section disappears, histogram and top decile included, one sentence takes its place, and your market value simply stays the median. That is the inconvenient answer. Any other one would be a guess.
Place your own salary
What you earn today is a voluntary entry. Add it and the check computes the gap to your market value. Leave it out and the comparison disappears entirely - no empty frame, no "unknown".
The gap in percent, phrased plainly
The gap is computed as a share of the market value, not of your own salary. If the two are no more than three percent apart, it simply says you are inside the market corridor. Above and below there is a number and a sentence that names a negotiating distance instead of issuing a warning.
Your salary appears in no resume
The entry lives in your Master Profile in a place of its own, deliberately separated from your personal details. From there it would otherwise travel into resume documents and PDF exports. Clear the field and the amount is gone - an empty field means deleted, not "zero euros".
The same role across all 17 regions
Below the distribution, Germany as a whole and the 16 federal states sit side by side, sorted by median. One click recomputes the whole page for that region. Regions for which the source fell back to national figures are marked as such - otherwise two states would show the same value as if it were their own.
Your recent checks stay comparable
The twenty most recent calculations remain as a history, each with the percentage gap to the next older row; anything older is cleared away on save. The market value itself is not stored: it is recomputed from the saved raw values every time, so that two truths never end up side by side.
- Upper quartile
- Median
- Lower quartile
Modelled distribution based on lower quartile, median and upper quartile - the official statistics provide these three values, not the full curve.
Take it into the salary negotiation
From a finished check you can have negotiation arguments generated. They do not start by themselves, only on click - and they follow rules that are written into the product.
Three to five sentences you can use verbatim
Every argument is a sentence in the first person, result first. Three strong ones beat five of which two are filler.
Every point says what it rests on
An argument comes either from your profile, from the official statistics, or it is pure negotiating technique. Which one it is stands on the point itself, not in a collective footnote below.
Amounts only as a range or as an anchor
The effect of an argument is given as a range or as the note "negotiating anchor", never as an exact single figure nobody can derive. The target range comes from the statistics, not from a hunch.
Nothing gets invented
Arguments are built only from what is genuinely in your profile. The numbers behind them the server fetches from your saved check, never from whatever the browser sends along.
Switch the region and they are gone
Arguments name the median and the name of their region. Click a different one in the regional comparison and they disappear: leaving them would mean passing them off as those of the new region. Five runs a day are included.