Create salary survey — collect anonymous salary data

A salary survey stands and falls with the trust of participants that their answers cannot be traced back. As soon as the suspicion arises that an individual data record could be assigned to a person, data quality tips over — either through refusal or through embellished information. Anonymity is therefore not a nice addition but the basis on which the entire evaluation rests.

In concrete terms: no personal fields, no IP capture, no cookies with user IDs. Avoid indirect identifiers like "last promotion in quarter X" or "direct supervisor", because in small teams people can be derived from these. Communicate the measures transparently at the start of the survey — a short privacy explanation with bullet points on "What we collect / What we do not collect / Who has access" builds much more trust than a legally clean but unreadable block.

For the evaluation to remain meaningful, you need demographic fields — role, experience level, region, company size. But this is exactly where the re-identification trap lurks: anyone who specifies "senior engineer, DACH region, 8+ years of experience" in a 30-person company may be uniquely identifiable. Solve the problem with broad buckets instead of fine granularity. Experience in 5-year steps, region as country instead of city, salary in 10K bands instead of an exact number.

Define a minimum size per evaluation bucket — for example: only evaluate when at least five people per category have answered. Clusters with fewer answers are collected in an "Other" category or not shown at all. This threshold is known as k-anonymity in the data protection context and best practice in HR benchmarks. Document the rule transparently in the evaluation report so that readers understand why some buckets show no numbers.

Salary data is considered particularly sensitive. Even without direct personal assignment, you must have a clean legal basis — usually the consent of participants according to Art. 6 (1) (a) GDPR. Actively obtain this consent via a checkbox, with a clear description of the purpose and retention period. Hidden consent in the footer is not enough. Also keep consent revocable — even if revocation technically fails with fully anonymized data, the note belongs in the form.

Define a deletion deadline for the raw data and communicate it. After aggregating and publishing the benchmark, you usually no longer need the individual answers — delete them after three or six months and only keep the anonymized statistics. Use database encryption and ensure that only a small group of people has access to the raw data. Anyone running salary surveys repeatedly should document the data protection impact assessment — in case of dispute this is the decisive evidence that the process was thought through cleanly.

A salary survey gains significantly in value when participants receive a benchmark report in return. Promise this right at the start: "Anyone who participates receives the anonymized industry report for free." This dramatically increases participation — especially in tech communities and HR networks, participation in salary surveys is an established exchange of data for insight.

The report itself should do more than provide simple averages. Show median, 25th and 75th percentile per role and experience level — the mean alone is distorted by outliers. Visualize the distributions as box plots or ranges, not as bare numbers. Add a short interpretation per role: "Senior engineers in DACH are at X median, top values reach Y, entry level at Z." Anyone who repeats the report regularly — for example annually — creates a data basis that not only supports fair pay but also serves as a reference in recruiting. Automatically send the finished report via webhook or email trigger to all participants who have provided their address for delivery.