What we collect
You send a pay file as Excel, CSV, or an HR system export. We need three things for each employee:
- Job title
- Base pay (annual salary or hourly rate)
- Work location (city and state, or ZIP)
These help, but are not required: employee ID (not names), bonus or other cash pay, full time or part time, hire date, department, and exempt or non-exempt status. We ask for your industry and company size once, when you sign up, not in the file.
How jobs get matched
Every company names its jobs a little differently. Our AI reads each job title in your file and matches it to a job in a standard job catalog, so that the same work is compared across companies.
You do not have to match jobs by hand or use a template. You can review the matches if you want to, but you do not have to.
How we clean the data
- Names dropped on arrival. If your file has employee names, we remove them as soon as it arrives.
- Duplicates. We remove rows that appear more than once.
- Outliers. We check pay that looks far out of line, such as an hourly rate entered as a salary, and fix or leave out rows that are clearly errors.
We will publish the exact outlier rules before the first report.
What a median is and why we use it
The median is the middle number. Line up every pay rate for a job from lowest to highest, and the median sits right in the middle: half are above it and half are below. It is also called the 50th percentile, or P50, which is where our name comes from.
We lead with the median because one very high or very low number barely moves it. An average can be pulled off by a single outlier. We also show P25 and P75, the points a quarter of the way up and three quarters of the way up, so you can see the range.
The 5-company rule
We never show a result for a group smaller than five companies. If fewer than five companies have a job in a location, that result is not shown. This protects every participant: no one can work out a single company's pay from a P50 number.
How often data is updated
P50 is an annual survey. Every result carries a "data last updated" date so you know how fresh it is. We will publish the collection and report schedule before the first report.
What we never do
- We never show your company's data to anyone else.
- We never publish a result built from fewer than five companies.
- We never ask for employee names.
- We never sell or share your raw data.
- We never deny who operates P50. The operating company is named in our terms.
The full data promise is on the your data page.