Why civic tech projects fail (and succeed)

People build technology to improve democracy all the time. Most of it doesn’t last. Matt Stempeck has spent a decade running the Civic Tech Field Guide — a crowdsourced directory of more than 11,000 democracy-tech projects — which means he’s also catalogued how and why they fail. The patterns repeat.

Three ways civic tech projects fail

• Building what the government should build, then asking the government to adopt it. A team builds a better voting tool, a better permit system, a better consultation platform — often genuinely better than what exists — and then tries to convince a city or agency to take it. This very rarely works. “Government likes to adopt their own solutions,” and it has its own ways of buying and building technology. So much civic tech has died on this rock that a generation of builders eventually went inside government instead, and the unglamorous topic of procurement (how government buys technology) became a movement of its own.
• Building your own social network for civic engagement. A recurring, well-funded failure. Network effects are brutal: being the first user on a new social platform is lonely, and “telling people to go somewhere and be the first user” almost never reaches critical mass. These projects also tend to carry a contradiction — “we want you to talk about democracy, but don’t be political” — when people interested in democracy often want to be political, and can already do that on ten platforms they’re already on.
• Building what you want to build, instead of what people actually use. The oldest mistake. “It’s always a sobering experience to watch someone use what you built.” Teams fall in love with the thing before spending time with the people the thing is for.

What actually works

• Go where people already are. When Microsoft wanted to support civic tech, it didn’t start a community from scratch — it sent people to participate in the city-based communities that already existed. Meeting people in the channels and groups they already use beats asking them to adopt one more platform.
• Go inside. That’s where the resources, the decisions, and the power to actually adopt something live. Many former “outside activists” became more effective once they were in the bureaucracy.
• Fix the plumbing. Helping government buy technology better — from smaller, more agile companies, including open source — unlocks more than any single app.
• Build to last. Stempeck’s guide survived for a decade partly by being “too cheap to fail.” He’s sharply critical of funders (the EU especially) whose projects vanish “two years and one day after the grant” — the domain lapses and even the collected knowledge disappears. Citizen infrastructure is only infrastructure if it’s still there next year.

It’s the politics, not the tech

A run of 2025–26 European studies keeps reaching the same verdict from outside the Field Guide. A European Parliament STOA foresight study reviewed 94 digital participation tools and 11 case studies and concluded that success depends far more on the political framework than on the technology: tools work when they carry a clear mandate and give participants visible feedback, and AI’s upside is swamped by its risks (bias, opacity) wherever legitimacy is already thin. Finland’s Sitra lands in the same place from the practitioner side, naming three changes needed to make digital participation stick: convince decision-makers (close the leadership gap with well-documented best practice), link platforms to actual decisions (build them into how government decides, rather than bolting them on), and streamline procurement so public bodies can actually buy open-source participation tools. The throughline with Stempeck’s “fix the plumbing”: what decides whether a tool matters is the interoperability, procurement, and political will around it, not its feature list.

The one piece of advice

Tip

“Talk to your intended users sooner, and spend as much time with them as possible before building the thing.”

It’s the most-repeated lesson in technology and still the most-ignored. For citizen infrastructure the stakes are higher than usual: a tool nobody uses isn’t neutral, it’s a community’s wasted time and trust.

This connects to how platforms touch democracy at all — see civic products, externalities & features — and to the work of letting the tools that do survive talk to each other, in connecting the tools.

Sources

• Matt Stempeck — Democracy Innovators Podcast (2026): youtube.com/watch?v=jQ-PZUNkNfg.
• Civic Tech Field Guide — the directory behind these observations.
• “The Use of Digital Tools and AI to Promote Citizen Participation in EU Policymaking” — European Parliament STOA study, 2026: futures4europe.eu
• Antti Lehtinen, “Three changes needed to strengthen digital citizen participation” — Sitra, 2026: sitra.fi (and the report “European civic technology and citizen participation in the age of AI”)