Blog

Judging rubric template for hackathons and build days (copy and edit)

· buildtolaunch

This guide is for student organizers who need a fair way to pick winners at a hackathon or build day, and for judges who have been handed a blank form. You get a copy-paste rubric, a way to combine scores, and a no-judges option for beginner events. It sits inside the full guide to build days and builder communities.

Short answer

Use four to five criteria, score each from 1 to 5, and write one plain sentence for what a 1, 3 and 5 look like. Give every project at least three judges. Instead of adding up long scores, have each judge pick their top three and award 3, 2 and 1 points. That ranking method is how MLH suggests picking winners. For beginner events, a public vote can replace judges.

What a rubric is for

A rubric answers one question before the event starts: what are we rewarding? Without it, judges reward whatever they personally like, and teams find out on stage.

A good rubric does three jobs:

  • It tells teams what to aim for, so you publish it on day one.
  • It helps judges agree, because "5 = a stranger can open it and finish the task" is clearer than "good execution".
  • It gives you something to point to when a team asks why they did not win.

Keep it short. Judges at a student event often have only a few minutes per team, so a ten-line form will not get filled in properly.

The template (copy and edit)

This is our suggested starting point, not an official standard. Change the names and weights to fit your event.

CriterionWhat the judge asksWeight (suggested)
WorksCan I open it and complete the main task without help?30%
Real problemIs there a specific person with this problem, and is it clear who?25%
ClarityDid the demo make sense to someone seeing it for the first time?20%
CraftIs it easy to use, and did the team make good choices within the time?15%
Next stepDoes the team know what they would do in the next week?10%

Anchors for the 1 to 5 scale, using "Works" as the example:

Works
1 = Does not open, or only a slide or mockup exists.
3 = Opens, but the main task breaks or needs the builders to explain.
5 = A stranger can open the link and finish the main task alone.

Write one line for 1, 3 and 5 on every criterion. Judges fill in 2 and 4 themselves.

Why these four or five

For a beginner build day, "Works" carries the most weight because the goal is a live link, not a clever idea on paper. For a hackathon with experienced teams, you can add a criterion such as "Technical depth" and lower "Works". If the event allows AI tools, say so in the rules and judge the team on how well they directed the tool and checked the result, not on typing speed. Teams will find your rules easier to follow if you also share the hackathon submission guide beforehand.

The score sheet

One row per project, one column per criterion. A plain spreadsheet is enough.

Judge name:            Round / table:
Project | Works (1-5) | Real problem (1-5) | Clarity (1-5) | Craft (1-5) | Next step (1-5) | One-line comment

Rules to print at the top of the sheet:

  1. Score each criterion on its own. Do not let a great demo lift the "Real problem" score.
  2. Write one comment per project, even if it is only "needs a clearer first screen". Teams value the comment more than the number.
  3. If you know the team well or mentor them, say so and skip that project.
  4. Scores are final once you submit the sheet.

How to combine scores without arguments

Averaging raw scores has a problem: one judge gives everyone a 4, another gives everyone a 2. MLH's judging guide recommends stack ranking to cancel that out:

  1. Each project is seen by at least 3 judges.
  2. Each judge picks their top 3 projects.
  3. First place gets 3 points, second gets 2, third gets 1.
  4. Add the points. The highest total wins.
  5. Organizers review the top 3 to 5 and make the final call.

You can use the rubric to help each judge decide their top three, then use stack ranking to combine them. This is our suggested way to use both together; MLH's page itself does not give a rubric.

How many judges and how much time

MLH gives a formula for the number of judges: projects times rounds per project times minutes per project, divided by total judging minutes, rounded up. Its recommended numbers are 3 rounds and 4 minutes per project, split as about 2 minutes of presentation and demo, 1 minute for questions and scoring, and 1 minute to move. Its example: 175 projects, 3 rounds, 4 minutes and 120 minutes of judging needs 18 judges.

For a student event, run the numbers with your own project count. Twenty projects, 3 rounds and 4 minutes is 240 judge-minutes, so with 60 minutes of judging you need 4 judges. MLH also suggests science-fair style judging, with judges walking between tables, and asking teams for a short demo video as a backup for deliberation.

A no-judges option for beginner events

If you cannot find three fair judges, skip them. At Vibe-a-thon Saigon, buildtolaunch ran an event with no judging panel: builders posted their projects to the Launch League board and voted on each other's apps. At that event, builders tried each other's projects and then voted; asking voters to try a project first is a step worth copying. Read the recipe in run your own vibe-a-thon.

Public votes reward popularity, so many organizers use a hybrid: the crowd votes for a shortlist of three, and a small panel scores only those three with the rubric above.

Common mistakes

  • Secret criteria. Publish the rubric with the event announcement.
  • Too many criteria. More than five and judges stop reading them.
  • No anchors. "Quality: 1 to 10" gives you noise.
  • Judges who are also mentors. They end up grading their own advice.
  • No comments. Students learn from a sentence, not from a 3.8.

For Vietnamese readers, there is a companion on what real hackathon boards score: Giám khảo chấm hackathon dựa trên tiêu chí gì? A 3-minute pitch script is in Mẫu kịch bản pitch và demo hackathon 3 phút. Planning the whole day? See the checklist and timeline for hosting a student hackathon or build day and how to run a demo day.

How buildtolaunch fits in

buildtolaunch is a free community for university students, open to any major, with no coding needed to start. If you run a build day, the Launch League board is a ready-made place for the vote: members post launches and it resets every Monday. Participants can keep building in project Spaces with an AI teammate, and the opportunities page lists hackathons and competitions for students, where they can read the real judging criteria before they enter.

FAQ

How many criteria should a hackathon rubric have?

Four or five. Fewer than four misses important differences, and more than five leaves judges rushing through the form. If you want more detail, group related ideas under one criterion and put the sub-questions in the anchor lines.

Should I use weights?

Weights are useful when one thing matters much more than the rest, such as a working link at a beginner event. If every criterion matters equally, skip weights and keep the math simple. Whatever you choose, publish the weights before teams start building.

How many judges do I need for a small student event?

Use the formula from MLH: projects times rounds times minutes per project, divided by total judging minutes. With its recommended 3 rounds and 4 minutes per project, 20 projects in 60 minutes needs 4 judges. If you cannot find that many, shorten the demos or use a public vote for a shortlist.

Can participants vote instead of judges?

Yes, and buildtolaunch did this at Vibe-a-thon Saigon, where builders voted on the Launch League board. Require voters to try a project before voting, and consider a small panel for the final three so the winner is not simply the most popular team.

What do I do if judges disagree?

Use stack ranking to combine their picks, then let the organizers review the top three to five. If two projects tie, pick the one with the higher "Works" score, since it is the criterion that proves the team shipped. Tell teams in advance how ties are broken.

Should the rubric change if teams use AI tools?

Say in the rules whether AI tools are allowed, then keep the same criteria. Add a line to the judge sheet asking whether the team can explain what the tool built and what they checked themselves. A team that can explain its project scores higher on clarity regardless of the tool.

Read the full guide to build days and builder communities.