What actually makes an offer good
A good offer is not the biggest base salary; it is the strongest total package for the role, level, and location you are being hired into. The single most common mistake is comparing one number, the base, against a half-remembered figure from a friend in a different city or a different level. Compensation is local and laddered: a senior backend engineer in San Francisco and the same title in Berlin are paid on entirely different scales, and a one-level difference inside the same company can move total comp by thirty to fifty percent. This tool removes that noise by benchmarking your annualised total against the specific slice you selected, so you are comparing like for like.
To compare offers fairly you have to convert everything to one annual number. Base salary is already annual. A target bonus is annual cash, so it adds in directly, though you should treat it as a target rather than a guarantee. Equity is the part people get wrong: a grant is almost always quoted as a total value over a vesting period, usually four years, so the correct annual figure is the grant divided by that period, not the whole grant. A one-off signing or relocation bonus should be amortised over the same window rather than counted as recurring pay, because it does not repeat. Once all four components are on an annual footing you have a number that can be placed on a market distribution honestly.
How to read the percentile
The result places your annualised total on the published percentile band for the role, level, and city: the 10th percentile is the low edge of the typical range, the 90th is the high edge, and the median is the midpoint. If your offer sits below the 25th percentile it is genuinely under market and there is a strong, data-backed case to negotiate before you accept. Around the median, the offer is competitive and the upside is usually in equity refreshers, the level you start at, or non-cash terms rather than base. Above the 75th percentile the offer is strong, and pushing hard on headline cash has diminishing returns; the leverage moves to the terms with asymmetric upside.
Percentiles are an estimate, not a verdict on your worth. The bands are built from real reported pay, but every dataset has gaps, especially for senior and staff levels outside the US and UK where verified samples are thin. Use the percentile to decide how hard to negotiate and where, not as a precise score. The number is most reliable in the middle of the distribution and least reliable at the extreme tails, which is exactly where the underlying sample sizes shrink.