Methodology
How the benchmark works
Companies answer a short survey, a person checks every submission, and results are only ever shown for groups of at least 8 companies. This page explains each step and the reasons behind it.
How does a survey round work?
Each survey covers one topic and runs in rounds: a round opens, collects answers until its closing date, and is then published.
- The questions are frozen when a round opens. Nothing is reworded while answers are coming in, so every company answers exactly the same thing.
- One set of answers per company. A company is identified by its work email domain and answers each round once.
- Answers can be changed until the round closes. After that they are fixed, so published results do not shift under people who have already read them.
How do you take part?
In three steps, and it takes about ten minutes.
Create your account
Enter your work email and we send you a link. No application, no waiting.
Answer a short survey
About 10 minutes, once per survey. Every question explains exactly what to count, so your numbers can be compared with everyone else's.
We check it, then you get the results
A person looks at every submission, usually within two working days. Genuine answers unlock that survey's results: the typical range, and where your company sits in it.
How do you know everyone answered the same question?
“ARR” means different things to different founders. So every number question comes with a plain-English definition — what to include, what to leave out, which period — right next to the answer box. That's what makes the results mean something.
And a person checks every submission before it counts, so made-up numbers don't end up in the results.
How are submissions checked?
By a person, one submission at a time, before it counts towards anything. Until a submission is accepted it is in no result and unlocks no result.
- The reviewer checks that the company is real and that the answers are genuine: not all zeros, not placeholder numbers, not typed in at random.
- Answers are not removed for being surprising. Unusual but genuine figures stay in; removing them would bend the results towards what was already expected.
- A submission that is not genuine is deleted, and the company is told.
Why is no group under 8 companies shown?
Because with fewer than 8 companies, a result starts to give away individual companies' numbers. A smaller group is reported as “not enough data” instead of a figure, and that applies to everyone, including the companies in it.
The rule is enforced where the data is stored, not only on the website, and it also covers combinations of filters: narrowing a result until fewer than 8 companies remain hides it. When results are split into groups, a group is also hidden if showing it would let the size of a smaller one be worked out.
How are companies grouped?
By five characteristics each company states about itself when it signs up. They are deliberately broad, so that no combination of them singles out one company.
| Characteristic | What it means |
|---|---|
| Country | where the company is headquartered |
| ARR band | annual recurring revenue, in broad bands |
| Headcount | everyone on payroll, in broad bands |
| Engineering headcount | people building the product, in broad bands |
| Funding | bootstrapped, angel, venture-backed, private equity or listed |
A company can correct these until it sends in its first survey. After that they are locked, because they decide which groups its answers are counted in.
What is reported, and why not the average?
For every number: the midpoint, and the range the middle half of companies sit in. The average is not reported.
| Reported | Also known as | What it tells you |
|---|---|---|
| Midpoint | Median, 50th percentile | Half of the companies report more, half report less. |
| Lower quarter | 25th percentile | A quarter of the companies report less than this. |
| Upper quarter | 75th percentile | A quarter of the companies report more than this. |
| Share of companies | For multiple-choice questions | How many companies chose each option. |
Why not the average? Two reasons. One or two very large figures pull an average away from what is typical, which a midpoint resists. And in a small group an average, combined with what a company knows about itself, says more about the other companies than a midpoint does.
The highest and lowest figures are never shown, for the same reason: each belongs to one company.
Who can see a company's answers?
The person who filled them in, and the small team that checks submissions. Nobody else: not other companies, and not colleagues at the same company.
- Other companies never see your answers
- They only see group results, like the typical range. Never one company's figure.
- No group under eight companies is ever shown
- If too few companies match your filters, you see “not enough data” instead of a number. This rule lives in the database itself, not just on the website.
- We watch for anyone trying to single out one company
- Every lookup is logged and rate-limited, so someone repeatedly narrowing the filters to isolate one company gets noticed.
- Our admins can see raw data — and every look is recorded
- We need this for support and moderation. Each access is written to a log that nobody, including us, can edit or delete.
Who runs it?
We Love SaaS, a curated network of software and AI founders and executives, rooted in the Benelux. You do not need to be a member to take part. Questions about the method can go to info@welovesaas.io.