Who Is Actually in the Room? Why We Built Estimated Audience Profiles

We assumed for a long time that sponsorship decisions came down to a few obvious numbers. How big is the event? How large is the prize pool? How well known is the organiser? Build a good directory of those, we thought, and teams could decide where to show up.

Then we started listening more closely to the people actually writing the cheques. The first question almost none of them asked was "how many attendees?" The first question, over and over, was a different one.

"Who is going to be in the room?"

That is the question that decides a sponsorship.

A team will happily back a 150-person event if most of the room is their ideal audience, and walk away from a 2,000-person event that is a poor fit. Students or senior engineers. Founders or researchers. AI builders or hardware tinkerers. First-timers or seasoned regulars. The headline numbers do not answer any of that.

The trouble is that organisers almost never publish it. They rarely list attendee seniority, experience level, the industries people work in, or the technologies the crowd actually uses. So sponsorship teams reconstruct it by hand: reading descriptions, scanning themes, recognising the organiser, guessing from the university name or the sponsor list. It works, and it does not scale. Doing it across hundreds of events a quarter is impossible.

What we built

Every hackathon now carries an Estimated Audience Profile.

Instead of one vague label, it breaks the likely crowd into the dimensions that matter: the experience level the event is pitched at, the kind of people it tends to draw, its technical focus, the industries in play, and inclusivity signals like beginner-friendly or women-in-tech. It is built from the details an event already gives off, so it works even when the organiser never spelled any of it out.

One distinction matters here. We are not claiming to know who attended. We are estimating who the event is designed for. That is an honest, useful thing to infer, and it is exactly the question you are asking when you are deciding where to spend a quarter's budget.

Built to be doubted

An estimate you cannot interrogate is just a guess with confidence.

So every profile shows its working. Each one carries a confidence rating, and the evidence behind it: the specific signals that led us to call a crowd mostly students, or beginner-friendly, or focused on AI. When the event description is thin, the confidence drops and we say so, rather than inventing detail to fill the gap.

This is deliberately closer to an audience estimate than to a demographic report. We would rather tell you "mostly students, high confidence, here is why" than hand you a precise-looking 63% that nobody can defend.

What it unlocks

Once every event has a profile, a new kind of search becomes possible.

Find the student AI events. Find the professional developer crowds. Find the hardware hackathons, the climate founders, the beginner-friendly weekends in a particular region. None of it requires an organiser to have tagged anything. You can shortlist on audience fit first and worry about logistics second, which is the order sponsorship teams actually think in.

That is the shift we care about. Hackathon Radar started by answering "what is this hackathon?" Estimated Audience Profiles let it answer the question that actually moves a decision: who is it for.