How to Host a Hackathon: A Guide for Startup Programmes and Corporate Innovation Teams

Researchers followed 11,889 hackathon projects from US events of the Major League Hacking (MLH) network through their code repositories. Seventy seven per cent of all the work happened in the first week after the event. Six months later, only 7 per cent of the projects showed any activity at all (McIntosh et al., 2021). The weekend itself worked. The Monday after did not.

If you run an accelerator, a venture client unit or an innovation programme, that number is your real brief. Somebody signed off the budget for your hackathon: an investment committee, a board, a head of innovation. In three months they will ask one question: what came out of it? This guide shows how to host a hackathon that has an answer. It covers the four types of hackathon, an eight week plan, team building, judging, budget and the 30 days after the event, where most projects are won or lost.

Our view in one sentence: a hackathon is a matching problem with a deadline. The prototypes come from the weekend. Whether they survive depends on who sits in which team before it starts and which decision maker meets which team after it ends.

What is a hackathon?

A hackathon is a time boxed event, usually 24 to 48 hours, in which small teams of three to five people build a working prototype for a set challenge and present it to a jury. The word combines “hack” and “marathon”. Online hackathons often run over several weeks instead of one weekend.

Participants are mostly developers, designers and product people, joined in corporate formats by domain experts from the business. The output is a demo, not a finished product. Judges score what a team built in the time, not what it promises to build later.

Hackathons are no niche. The platform Devpost lists almost 14,000 hackathons, and MLH counts more than 1,000 events a year in 100 countries (both as of September 2026). In 2026, machine learning and AI challenges carry the largest prize pools on Devpost by a wide margin. For you as an organiser, that means competition for good participants. Your challenge has to be worth a weekend.

Four types of hackathon, four definitions of success

The first decision is not the venue or the date. It is the goal, because the goal decides who you invite and what you will count as success. Most failed hackathons tried to be all four types at once.

TypeMain goalWho takes partWhat counts as success
Talent hackathonSpot and recruit peopleStudents, developers, career changersInterviews and hires within 90 days
Internal innovation hackathonNew ideas from your own staffEmployees across departmentsProjects adopted into a product roadmap
Corporate startup challengeFind startups that solve a real business problemStartups and scaleups, corporate challenge ownersPaid pilots or purchase orders
Ecosystem hackathonGrow a community or a startup sceneFounders, students, local developers, investorsNew teams, companies founded, follow-on funding

The third type is the one most startup programmes are really after, and it works differently from the others. Startups do not come for a prize. They come for access to a buyer. If that sounds like venture clienting, it is close: our guide to venture clienting at startup events explains why the corporate in the room acts as a customer, not a mentor.

Be honest about the cost of each type, too. A 2026 overview of open innovation formats by the consultancy Valu puts a two to three day corporate hackathon at 20,000 to 80,000 US dollars and describes its typical output as “prototype, rarely production”. Read that as a warning, not a verdict. The format produces prototypes by design. Whether anything moves beyond that depends on what you plan after the weekend.

Why most hackathon projects stop on Monday

The research on what happens after a hackathon is surprisingly consistent, and it points to things an organiser controls. Three findings matter most for your planning.

After the event

Most projects fade within a week

7% of hackathon projects still show activity six months later 11,889 MLH projects, 2018 to 2019
77% of all code commits happen in the first week after the event same data set
60% of participants would not have started their project without the hackathon online health hackathon, 2020
McIntosh et al., SIGCSE 2021; Braune et al., Journal of Medical Internet Research, 2020/2021

The third figure is the good news. Hackathons do start things that would not exist otherwise. The first two show where the effort goes missing: in the days after the event, when the adrenaline is gone and nobody owns the next step.

What makes the difference? A large study of hackathon projects by Nolte and colleagues (2020) found that short term continuation depends on technical preparation and on winning. Long term continuation depends on something else: skill diversity in the team and the team’s intention to take the project further. A case study of corporate hackathons in a company with operations in several African countries (Ratsoga et al., 2023) names three factors that decided whether projects continued: financing, the skills fit of the team, and how well the project was integrated into existing operations.

Put those findings next to each other and a pattern appears. Two of the success factors are about people: the right mix of skills in each team, and the right owner on the business side. Both are matching decisions. Both are yours.

How to host a hackathon in eight weeks

Knowing where projects fail changes the order of your planning. Commit the follow-up before you pick the pizza. The plan below works for a 24 to 48 hour in person event with 50 to 150 participants. For larger events or several challenge partners, add four weeks.

Planning timeline

A hackathon in eight weeks

  1. Week 8 Goal and challenge owners Pick one of the four types. Name a person on the business side for every challenge.
  2. Week 7 Budget and follow-up commitment Agree what winning teams receive after the event: a pilot budget, a meeting, a place in the programme.
  3. Week 6 Jury and criteria Mix business and technical judges. Publish the criteria before registration opens.
  4. Weeks 5 to 3 Recruit with skills data Ask every applicant for role, skills and interests, not only name and company.
  5. Week 2 Form teams Mix skills on purpose and leave room for people who arrive with their own team.
  6. Week 1 Brief and prepare Send challenge briefs, data sets, test accounts and access to application programming interfaces (APIs).
  7. Event Build, pitch, decide 24 to 48 hours of building, mentor check-ins, then short demos and a two round jury.
  8. Days 1 to 30 Follow-up Booked meetings between teams and challenge owners, check-ins on day 7 and day 30.
Converve editorial synthesis based on McIntosh et al. 2021, Nolte et al. 2020, Ratsoga et al. 2023 and organiser guides by AngelHack and Qmarkets (2026)

Write challenges a team can finish in 48 hours

A good challenge is narrow enough to demo on Sunday and important enough that a challenge owner wants to see the result on Monday. “Improve our customer experience” fails both tests. “Build a tool that tells a field technician which spare part to order from a photo of the broken component” passes both.

Hackathon ideas that work well for startup programmes and corporate challenges share three traits: real data, a named user and a clear definition of done. A few examples from the formats above:

  • For a logistics corporate: predict late deliveries from two years of anonymised shipment data.
  • For a city or regional development agency: help visitors find accessible venues using open data.
  • For an insurance venture client unit: turn a claim photo into a structured damage report.
  • For an accelerator’s own community: build the tool your founders complain about every cohort.

Give every challenge an owner who attends the kick-off and the final pitches. If no one on the business side will give up a weekend for it, drop the challenge.

Build teams on purpose

Many hackathons leave team building to chance: a kick-off, a few minutes of pitching ideas, and whoever is standing nearby joins the team. That is how you get four backend developers and nobody who can explain the result to a jury. Research on continuation points the other way. Skill diversity predicted which projects survived in the long run (Nolte et al., 2020).

Collect skills at registration and use them. Most organiser guides recommend teams of three to five, and MLH notes on its own event pages that teams with a maximum of four people tend to perform best. A workable rule: every team needs someone who builds, someone who designs or tests with users, and someone who understands the business problem. Participants who arrive as a complete team keep it. Everyone else gets matched before the event, not during the kick-off.

Choose judges and publish the criteria

Your jury decides which teams get attention after the event, so it should include the people who will give it: challenge owners, an investor or two for ecosystem formats, and at least one technical judge who can tell a real demo from a slide. Publish the criteria when registration opens. Teams build towards what they know will be scored.

CriterionWhat judges askSuggested weight
Problem fitDoes it solve the challenge as the owner described it?30%
Working prototypeDoes the demo run, with real or realistic data?25%
FeasibilityCould this run in production within six months?20%
Team and next stepDoes the team want to continue, and is the next step clear?15%
OriginalityIs the approach new for this company or this problem?10%

The fourth criterion is unusual, and deliberate. It rewards the intention to continue, which Nolte and colleagues identified as a driver of long term survival. With more than 15 teams, use two rounds: judges visit the tables during the last hours and pick a shortlist, and only the finalists pitch on stage. Twenty five teams presenting for five minutes each would take more than two hours, and the room would empty long before the end.

What does a hackathon cost?

Budgets vary widely, but the main cost items are predictable. The organiser guide hackathon.guide estimates venue costs at around 10 to 30 US dollars per person per day in a major city, and food at 7 to 15 dollars per person. Add prizes, travel for judges and mentors, swag, and staff time, which is usually the largest item nobody budgets for.

For a corporate format, the 20,000 to 80,000 dollar range from Valu is a realistic frame. The line most budgets miss is the one after the event: the pilot budget or meeting time that turns a winning prototype into a project. If you have to cut something, cut the swag.

The 30 days after: where the matching happens

The commit data shows the pattern. The week after the event is when teams still have energy, and it is exactly when most organisers are busy with thank you emails and photos. That is the week to schedule meetings.

Plan the follow-up as part of the programme, not as an afterthought:

  • Meetings booked before the event ends: every finalist leaves with a date for a meeting with the relevant challenge owner or investor, set on site rather than promised by email.
  • One owner per project: someone on the business side is responsible for the next step, with the budget agreed in week 7 of your plan.
  • Check-ins on day 7 and day 30: a short call with each team. Day 7 catches the teams that are still moving. Day 30 tells you which projects your board will hear about.
  • A result for the report: by day 30, you can show which teams moved to a pilot, a programme place or a hire. That is the answer to the question your budget holder will ask.

The same logic applies to investors at ecosystem hackathons. If you invite them as judges, give them a way to meet the teams they liked afterwards. We described how to track those conversations in our guide to investor follow-ups after a demo day. The mechanics are the same.

Solution: Converve helps you plan the people side of a hackathon. Participants create profiles with their skills and interests, which you can use to form balanced teams before the event. During and after the hackathon, challenge owners, investors and teams can request and confirm one-to-one meetings in a shared schedule, based on rule based matching that your team controls. Our solution for startup and investor events shows how this works in practice.

Frequently asked questions

How long does a hackathon last?

Most in person hackathons last 24 to 48 hours, often over a weekend. Online hackathons can run for several weeks, because participants build alongside their normal work. Internal company formats are sometimes shorter, with one or two focused days.

What is the best platform for hosting a hackathon?

It depends on what you need to manage. Platforms such as Devpost handle submissions, project galleries and judging. For team formation and meetings between teams, challenge owners and investors, you need a tool that works with participant profiles and schedules, such as an event matchmaking platform. Many organisers combine the two.

Can anyone take part in a hackathon?

Most public hackathons are open to anyone who registers, and many are explicitly beginner friendly. Corporate and startup challenges often select participants through an application to ensure a mix of skills. Internal hackathons are limited to employees.

Can participants use ChatGPT or other AI tools?

Usually yes, and in 2026 most hackathons expect it. Say so in the rules. If you allow AI tools, ask teams to disclose what they used, and make sure your judges score the working result and the team’s understanding of it, not the amount of generated code.

How many people should be in a hackathon team?

Three to five people is the common recommendation, and MLH observes the best results with a maximum of four. More important than the size is the mix: someone who builds, someone who designs or tests, and someone who understands the problem.

How much does it cost to host a hackathon?

A small community hackathon can run on a few thousand dollars with a donated venue. Corporate formats typically cost 20,000 to 80,000 US dollars (Valu, 2026). Plan a separate budget for what happens after the event, such as pilots or programme places for the winners.

Conclusion: plan the Monday first

A hackathon produces energy, prototypes and a good photo for LinkedIn. Only 7 per cent of projects are still alive half a year later, and the research explains why: the teams lacked the right mix of skills, or nobody on the business side took the next step. Both are decisions you make as an organiser, before the first line of code is written.

So treat your hackathon as a matching problem with a deadline. Match skills when you form teams. Match winning teams with the owners, buyers and investors who can move them forward. Then book the meetings before everyone goes home. If your programme also includes a stage for founders, our guide on how to run a pitch competition covers the format that often follows a hackathon.

Want to plan the meetings that come after your next hackathon? Get in touch with Converve.

Want matchmaking that fits your event?

Get a free consultation