A year ago, I wrote about how Google hadn't produced any adoption numbers for Bazel. However, companies like ours must make informed investment decisions, and so must a lot of individuals deciding whether the Bazel ecosystem is "big enough". This could be as simple as "should I convince my husband that it's worth me spending my nights and weekends to write a book about Bazel", or "is it economical for my consulting company to advertise our services by hosting a Bazel podcast". On the other hand, it could involve institutional investors who are putting millions of dollars into a business plan that is counting on projected revenue from users to make Bazel more usable by enterprises.

Here's that post from last year: blog.aspect.dev/estimating-bazel-adoption. The summary: in the absence of data from Google, we used estimated 600 companies were using Bazel.

This year

Fast-forward to BazelCon this year. In the keynote, there was a wide open opportunity for the Bazel team to provide some adoption numbers. Even though Bazel still doesn't have telemetry built-in, there's a simple proxy - the same one we always used for the Angular JavaScript framework: distinct active users of the documentation site. Sadly, while the Bazel team is interested in providing this data, there are bureaucratic obstacles in privacy policy and legal standing in the way, so they still haven't published anything. I remain optimistic that we'll get it eventually!

Fortunately, my co-founder Greg is willing to do all the work to reproduce our analysis from last year, and our result is that there are now about 950 companies using Bazel, a 56% increase from last year.

Details

I won't repeat the methodology we used, as you can read about it in the post from last year. Essentially, we take two different approaches to list which companies use Bazel, then see how many companies appear in both lists. Assuming a uniform "density", this gives us an approximation of how many Bazel users appear on neither list, and so we can extrapolate a total.

Last year we published this data:

TBA = [# public list] * [#private list] / [# overlap] = 218 * 127 / 46 = 602

Updating for November 2023, we now arrive at:

TBA = [# public list] * [#private list] / [# overlap] = 311 * 228 / 75 = 945