Why your hackathon didn't change anything

Hackathons fail in the handover. Energy on Saturday, silence in the steering pack.

The pattern is consistent: no product owner, no integration path, no 30-day experiment budget. The winning team gets a trophy and a thank-you email. We have mapped this across enterprise weekends in Australia, New Zealand, Canada, and the US. The logos change. The Monday does not.

This is a Lean Startup and Product Sprints piece that happens to be about a format everyone already bought. If you want culture, change a Tuesday. If you want prototypes, run a factory with a budget. Do not ask one Saturday to do both.

What we keep seeing

Friday: a keynote about being bold. Saturday: teams that include one engineer, two marketers, and a facilitator who is doing their best. Sunday: demos judged by people who cannot fund. Monday: Slack goes quiet. Day 30: the winning idea is a slide in a pack titled Pipeline.

Inkeros is a Canadian omnichannel sales and engagement company with 600-plus people across Canada and the United States. When we ran a two-day innovation sprint there, the useful move was not "more hack energy." It was problem framing on day one, prototypes and decision criteria on day two, and a shortlist with owners. That is a product sprint wearing a workshop badge. The residue was a list someone would fund, not a mural.

AGL Energy Ltd is a listed Australian generator and retailer. Energy weekends fail when the winning idea needs a production integration and nobody from risk attended. They work when the output is an offer test or a concierge a retail and risk lead can live with. Lean Startup T&E is the name of that honesty.

Atlassian ShipIt is the internal rhyme everyone copies and few copy faithfully. A bounded clock. A demo. A path back into the product conversation. The steal is the path, not the beanbags. If your enterprise hack has no path into a backlog an owner controls, you copied the photos.

Book the follow-up before you book the venue. Put a dollar figure on the two ideas you will actually run.

Lean Startup is the missing operating system

A hackathon is a burst of build. Lean is the loop that tells you whether the burst was a test or a costume.

Write, before the event:

• Hypothesis per team, in one sentence.
• Thinnest test that could run in 30 days without a platform.
• Owner who already has a day job adjacent to the problem.
• Kill criteria a sceptic will sign.
• Budget (even small) sitting in a cost centre, not in a "we will find it" slide.

That is a product sprint brief. Teams that cannot fill it should not be at the event. They should be in a framing session. Inkeros-style rooms that start with framing avoid the Sunday pile of demos that cannot be tested. AGL-style rooms that start with the constraint avoid the demo that is illegal to learn from. Atlassian-style rooms that start with a product conversation avoid the orphan repo.

The Monday job

Monday is the product. Saturday is rehearsal.

On Monday a named owner:

1. Takes two ideas, not twelve.
2. Writes the 30-day test licence (what we will do, who it touches, what we will not change, who can stop it).
3. Books the next demo in the existing cadence. Do not invent a new steering pack.
4. Kills the rest in public, with thanks and a sentence of learning.

If that job has no name on Friday, cancel Saturday. I mean it. Collective Campus will not pretend a venue is a strategy.

Don'ts for the next brief

Don't run it to "build culture." Culture is a wet Tuesday. Run it to produce three prototypes you will fund or kill.

Don't let vendors own the judging. They will reward polish. You need testability.

Don't exclude risk, legal, or ops from the room and then act shocked on Monday. They are users of the experiment.

Don't measure success as number of ideas. Measure funded tests and killed tests. Both are results.

Don't stack a hackathon on top of an already drowning calendar. See productivity. Hours must come back from somewhere. Delete two status rooms to pay for the weekend, or do not run it.

A one-page brief

Event purpose: produce two 30-day tests, not a culture story.

In the room: problem owner, builder, constraint owner (risk/legal/ops), someone who can say no.

Out of the room: a page per funded idea (hypothesis, test, metric, kill date, dollars).

Not out of the room: a mural, a trophy as the only artefact, a promise to "take it to the innovation committee."

Inkeros could run that page. AGL could run that page. An Atlassian team already does a version of that page. If your organisation cannot, you do not have a hackathon problem. You have an ownership problem. Fix that on a Wednesday. Then, if you still want a Saturday, we will run it as a sprint.

Judging without the costume

Judges reward polish because polish is easy to see at 5pm on a Sunday. Change the score sheet.

• Is there a hypothesis in one sentence?
• Is there a 30-day test that does not need a platform?
• Is there an owner who already has a day job next to the problem?
• Did risk, legal, or ops mark the constraint?
• Would we spend money on this Tuesday?

Atlassian-style ShipIt works when the path back to a product owner exists. Inkeros-style sprints work when the shortlist has names. AGL-style energy weekends work when the test is an offer or a concierge. If your judges are brand, comms, and a vendor, you will fund a video. Change the panel or cancel the event.

Budget on Monday, written like an adult

"We will find funding" is not a budget. A cost centre, a figure, and a name who can release it: that is a budget. It can be small. A landing-page test and a week of someone's time is a budget. A concierge for 30 customers is a budget. A core-system rewrite is not a hackathon output and should not be funded as one.

Put the figure on the invitation. Teams behave differently when they know two ideas will get money and the rest will get a funeral. Lean Startup is that honesty. Product sprints are that honesty. Your Saturday can borrow it.

Forty events, one pattern table in prose

The dek said we mapped 40 enterprise hackathons. Here is the map without a fake leaderboard. We will not invent a "23 of 40 failed" statistic. We will not quote a living judge. We will name the shapes the weekends clustered into, because the shapes are what a sponsor can steal or refuse.

Across Australia, New Zealand, Canada, the US, and the weekends we have sat in or reconstructed from briefs in the UK and Singapore, the 40 events fell into five shapes. Most buildings ran more than one shape over a few years. The logo on the lanyard changed. The Monday did not, unless the shape was the last one.

Shape 1. The culture weekend. Purpose stated as energy, belonging, or "being bold." Keynote on Friday. Mixed teams with one engineer, two marketers, a facilitator doing their best. Judges from brand, comms, and a vendor. Trophy on Sunday. Slack quiet on Monday. Day 30: a slide titled Pipeline. We saw this at telco, bank, energy, and public-sector-adjacent weekends. It is the default if you do not write a Monday job. AGL-shaped energy and NHS-shaped health both produce this shape when risk and ops are invited as guests, not as users of the experiment.

Shape 2. The vendor showcase. Purpose stated as innovation. Purpose in practice: a platform or a model gets a captive demo slot. Teams are steered toward the tool. Judges reward polish that makes the vendor look inevitable. Residue: licences and a photo. Telstra-scale, DBS-scale, and US bank digital academies have run useful and useless versions of tool-heavy weekends. The useful ones started with a hated workflow and a product owner. The useless ones started with a booth.

Shape 3. The ShipIt copy. Atlassian ShipIt is the internal rhyme everyone photographs and few copies faithfully. A bounded clock. A demo. A path back into a product conversation. When ANZ-scale, Canva-scale, or a Canadian operator like Inkeros copies the photos and not the path, you get Shape 1 in nicer merch. When they copy the path (owner, backlog, 30-day test), you get a factory. The steal is the path.

Shape 4. The regulated weekend with the stopper missing. Energy, banking, government, legal. A winning idea that needs production integration, a financial promotion, a clinical pathway, or a client-confidential workflow. Risk, legal, IG, or a partner was not in the room. Monday is a polite no, or worse, a quiet yes that becomes an incident. AGL, NAB, Lloyds, Services Australia, Allens: the physics are the same. The constraint is a design input. If it arrives on Monday, you ran a costume.

Shape 5. The product sprint wearing a hackathon badge. Problem framing before the build. Hypothesis in one sentence. Thinnest test that does not need a platform. Owner who already has a day job next to the problem. Kill criteria a sceptic will sign. Budget sitting in a cost centre. Person who can say no in the chair on Sunday. Two ideas funded. The rest killed in public. This is the Inkeros two-day shape. This is Lean Startup T&E. This is the only shape we will keep running.

A sixth near-shape showed up often enough to name: the student or graduate jam attached to an enterprise brand. Fine as recruiting. Do not file it under transformation. Hours and credibility still move. Residue still needs a home or it is a party with better lighting.

If you are briefing the next event, name the shape in the first line. If you cannot, you are about to buy Shape 1 and call it Shape 5.

A practical classification line for the sponsor:

This event is Shape 5 or it is cancelled. Two 30-day tests, named owners, a dollar figure on the invitation, risk in the room. If you want Shape 1, do not hire us to pretend.

The budget page

"We will find funding" is how 40 events turned into photographs. Write a budget page before the venue. Circulate it with the invitation. Teams behave differently when they know two ideas will get money and the rest will get a funeral.

Cost centre. Not leftover lab dust. A line a CFO at NAB, Telstra, AGL, or RBC can see. If the lab is paying and the line owner is not, you have already decided Monday will fail.

Figure for tests, not for catering. Catering, travel, and facilitation are costs. They are not the budget that matters. The budget that matters is the envelope for two (maybe three) 30-day tests: a landing-page channel, a week of named people's time, a concierge for a bounded cohort, a script in one bay. A core-system rewrite is not a hackathon output and must not be funded as one. Infinitev-style offer tests and AGL-style concierge tests live in this envelope. Platform fantasies do not.

Name who can release it. A director, a GM, a product owner. They sit in judging or they do not get to be surprised. Inkeros-scale rooms fail when the weekend is energy and Monday is nobody's job. Put the job on Friday.

What the money may not buy. Production integration. A new vendor. A panel the procurement lead has not seen. A claim legal has not marked. Write the exclusions. Shape 4 dies here, on purpose.

Kill date for the envelope. Unused test money returns. It does not roll into "more ideation." Unused money is a result. It means you were honest about what was testable.

Calendar tax. Delete two status rooms to pay for the weekend hours, or do not run the weekend. See productivity. A drowning team cannot also be a prototype factory.

A practical budget paragraph:

Cost centre [x], owner [name], envelope [figure] for two 30-day tests. Money may not buy production change or a new vendor. Unused money returns on [date]. We deleted [rooms] to pay for the hours. If we cannot fill this paragraph, we will not book the venue.

The judging sheet that funds tests

Judges reward polish because polish is easy to see at 17

on a Sunday. Change the sheet. Score testability. Make the person who can say no use the same sheet as everyone else. Vendors do not own the panel. Brand and comms may sit, they may not outvote testability.

Score each team 0, 1, or 2 on five lines. No extra categories. Extra categories are how polish sneaks back.

1. Hypothesis in one sentence. If you cannot read it aloud, it is a vibe.
2. 30-day test without a platform. Script, concierge, shadow, throwaway page, letter. If the next step is a core release, score zero.
3. Owner with a day job next to the problem. Named. Present or confirmed. "The innovation team will take it" scores zero.
4. Constraint marked. Risk, legal, ops, clinical, conduct, or claims sat with the idea or marked a page. Missing stopper scores zero in a licensed shop.
5. Would we spend Tuesday money? The budget owner answers this one. Everyone else can guess. Their guess does not release funds.

Fund the top two that clear a bar you wrote before the event (for example, no zeros on lines 3 and 4). Kill the rest in the room, with thanks and a sentence of learning. Publish the kills. GDS-style and DBS-style cultures already know a public stop earns the next window. Energy and banking can steal it.

Atlassian-style ShipIt works when this sheet maps to a product owner. Inkeros-style sprints work when the shortlist has names. AGL-style energy weekends work when the test is an offer or a concierge. Capital One-style and Canva-style rooms already have a path into a backlog. If your judges cannot use this sheet, change the panel.

A practical Sunday close:

We are scoring testability, not cinema. Two ideas get an envelope. The rest get a funeral sentence tonight. If you want a trophy only, say so now and we will stop calling this Lean Startup.

What the 40 events shared when they worked

The working weekends, whatever the logo, shared a boring spine. Book the follow-up before the venue. Put a dollar figure on two ideas. Invite the person who can say no. Write hypotheses before the build. Treat risk as a user. Measure funded tests and killed tests, not number of ideas. Do not stack the weekend on a drowning calendar.

They also shared a refusal: they refused to let Sunday be the product. Monday was the product. Saturday was rehearsal. Collective Campus will keep running Shape 5 and sending this piece to Shape 1 briefs. If your organisation cannot fill the budget page and the judging sheet, you do not have a hackathon problem. You have an ownership problem. Fix that on a Wednesday.