New to Science Fairs? From Our Experience as Judges, What to Know About Winning Science Fairs

AIClub mentors have judged science fairs at every level — from local school fairs up through ISEF itself. This isn't advice pulled from a rulebook. It's what we've actually watched happen, sitting at the judges' table, project after project.

Most first-time science fair advice is either too vague ("pick something you're passionate about!") or too generic — a list of pre-packaged project ideas anyone can copy. Having judged this at every level, here's what we actually see separate a strong first project from a weak one, and what that means for how you should approach yours.

1. The ladder is real, and the same things matter at every rung

Local fairs, regional fairs, state and affiliated fairs, and ISEF itself, we've judged across that entire range. The key is that judges at every level are looking for the same underlying things. What changes as you go up isn't the criteria, it's how rigorously they're applied. A thin project can still place at a small local fair. It won't survive a serious judging panel two levels up. Building good habits early means you won't have to unlearn anything later.

2. The projects that stand out started with a real question, not a kit

The single most common thing we see separating a memorable project from a forgettable one: a real question the student didn't already know the answer to, versus a demonstration dressed up as an experiment. "Does music affect plant growth" and "which brand of battery lasts longest" are fine starting points for curiosity, but on their own they're closer to a demonstration than an investigation.

A useful test: can you finish the sentence "I don't know what will happen when..." with something specific? If the question already has an obvious answer, it's not there yet.

3. Do the background research before you touch an experiment

Before designing anything, find out what's already known. This isn't busywork. It's what lets a student design a question that's actually new, instead of unknowingly repeating something well established. We can tell within the first minute of a judging conversation whether a student did this or skipped it.

Tip: Skipping this step might be accommodated at an entry level fair, but it will not work at the next level. The ideas that succeed at higher levels are truly novel, and the student has done the background work and can explain what is new and what is not.

4. Design and methodology is where most points quietly disappear

This is one of the largest-weighted sections in the official Regeneron ISEF judging rubric we've judged against directly — a well-designed plan with clearly defined variables and controls. It's also the section we see the most avoidable point loss in at every level: a project that doesn't clearly identify what it's changing, what it's measuring, and what it's holding constant loses real points regardless of how interesting the topic is.

Tip: This is particularly true for AI. AI algorithms are now part of more and more science fair projects. The judges (and the fairs) are becoming more critical about whether your AI methodology is sound.

5. Data quantity and honesty matter more than most students expect

"Execution: Data Collection, Analysis and Interpretation" carries more weight in the official rubric than any other single section — 20 of 100 points. In our own experience judging, this is exactly where we see the gap between a good project and a great one. One trial, or a handful of data points, isn't enough to support a real conclusion, and it's usually obvious to a judge within the first question.

Tip: As the grade level increases, the data collection becomes more sophisticated. Students are expected to understand the statistical concepts that drive good data collection and data analysis.

6. Keep a logbook as you go, not after

We look at the notebook. A logbook kept honestly throughout — including the parts that didn't work — tells us more about a student's actual process than a clean summary written after the fact usually can.

Tip: Logbooks (whether physical or online) are a good research practice regardless of whether science fairs are your target.

7. The board matters less than students think

We've judged enough projects to say this plainly: the physical display is secondary to whether the student actually understands their own subject. Spend your remaining time before the fair making sure you understand your own project cold, not making the poster prettier.

8. Picture the room you'll actually be standing in

A physical science fair isn't a quiet one-on-one conversation. It's usually a massive room — hundreds of students, each standing next to their poster, shoulder to shoulder with the project on either side. A judge gets a few minutes at your table before moving to the next one. In that time, you're not really competing against an abstract standard. You're competing with the poster three feet to your left.

What actually cuts through, in that room, in those few minutes: why the problem matters — not just what you did, but why anyone should care — why the idea itself is interesting, and how far you actually went to chase it down. We've seen hundreds of competent, well-executed projects in a single day. What sticks with us is the student who tried something a little unconventional, or who kept going after the first version didn't work.

Tip: That last part matters more than most students realize: a failure doesn't hurt you in that room. A dead end, described honestly, followed by what you did next, reads as real work. Exploration is what wins those few minutes — not a spotless result.

9. What actually impresses us isn't how much a student knows — it's how far they went to find out

This is the thing we notice most, and it's rarely what students expect us to care about. A student who can recite the right answer to every question we ask is fine. A student who tells us about the three methods they tried before one actually worked, or the strange result they got halfway through and the extra experiment they ran just to figure out why, is memorable. That's not in any rubric line item, but it's the difference between a project that feels rehearsed and one that feels real. It shows up in small ways: mentioning a control you added after noticing a flaw in your first design, or a version of the experiment you scrapped and why. Don't hide that. It's usually the most convincing part of the whole conversation.

10. The interview is where we actually find out what a student knows

This is the part first-timers underestimate most. We're not just checking your results — we're checking whether you understand your results, including their limitations. A student who can say "here's where my method has a weakness" comes across as more credible to us, not less. We also routinely ask whether a project done at home or in a school lab had any mentoring or professional guidance — that's a completely normal question, not a trap, and the students who answer it clearly and specifically about what was theirs versus what a mentor helped with are the ones who come across best.

11. The mistake we see most often at every level we've judged

Not lack of talent — lack of real, honest data. A beautiful hypothesis with three data points doesn't hold up under the highest-weighted section of the rubric. If you're choosing between a more ambitious question with less time to collect data, and a more modest question you can actually test thoroughly, choose the one you can test thoroughly. We'd rather see a small project done rigorously than a big one done thin.

12. Why a mentor who's actually judged this makes a real difference

Everything in this post came from sitting at the judges' table ourselves, which is really the point. A mentor who has personally judged science fairs — at every level, from local fairs through ISEF — doesn't just describe what a judge might ask. They can sit across from a student beforehand and actually ask it: push on the limitations of the method, ask what specifically the student did versus where they got help, probe the part of the project the student is least confident about. That's a different kind of preparation than reading a guide, including this one. It's rehearsal against the real thing, run by someone who's actually done the real thing — and it's exactly the kind of prep AIClub's mentors, many of whom have judged these fairs directly, can give a student before they ever walk into that room.

We also know the value of this kind of rehearsal from the other side of the table, not just as judges. Every one of AIClub's founders and advisors holds a PhD, and a PhD is essentially years of mock judging before it counts for real — qualifying exams, committee reviews, practice defenses, being pushed on the weakest part of your argument until you can defend it without flinching. We didn't just judge students through this. We went through it ourselves, over and over, before anything we did was ever evaluated for real. That's not a coincidence in how we prep students. It's the same instinct.

If your student is starting their first project and could use a second opinion

Sometimes the most useful thing isn't more information. It's a mock judging conversation with someone who's actually done this, telling you honestly whether your specific question is testable, and where a real judge would push back.

Book a free consultation here.