6 Self-Service Data Lessons from Everyday Platforms

If you'll get into a stranger's car, then why won't you trust your colleague's sales report?

It's a fair question, and it gets at something I think about a lot when helping organisations with their self-service data strategy. We trust platforms like Uber, Airbnb, and Tinder with some pretty significant decisions, often involving strangers, money, or both. But inside our own organisations, people won't touch a dataset they didn't build themselves. Self-service data strategy succeeds when organisations apply the same design principles that make consumer platforms work: network effects, discoverability, trust, scalable starting points, governance, and accessible documentation. So I looked at six platforms people use every day and pulled out the principle behind why they succeed, because it maps directly.

Network effects

On Airbnb, hosts become guests and guests become hosts. Value flows both ways. Every time another team in your organisation reuses what's already been built, whether that's a report, a data model, or a shared metric definition, you increase the return on that original investment. The more teams participate as both creators and consumers of data, the more valuable the whole environment becomes. This is the foundational shift: moving from "my team's reports" to "our organisation's data."

Discoverability

Tinder shows you options you didn't know existed until the platform surfaced them. You swipe through, you figure out if you want to commit. Most BI environments have the opposite problem: useful reports and datasets exist, but nobody outside the team that built them knows they're there. If people can't find what's been built, it may as well not exist.

A Forrester study found that knowledge workers spend nearly 30% of their week just searching for the information they need to do their jobs. That's almost a day and a half every week lost to looking for things that should be findable. There's a concept in platform economics called the cold-start problem, basically the chicken-and-egg challenge of getting both sides to show up. The answer is almost always to make the first thing people find so useful they come back for more, and people will figure out if they want to commit.

Trust

Uber's verification, ratings, and in-app safety features make it feel safe to get into a stranger's car. That's a genuinely remarkable thing when you think about it. And yet inside most organisations, people won't use a dataset they didn't build themselves because they have no way of knowing whether it's accurate, current, or maintained.

People only use data they trust, and they rarely want to spend time figuring out whether they should. Make it obvious which sources are verified, which datasets are certified, and who owns what. If the trust signals aren't there, people will just rebuild it themselves (and you're back to the same analysis being done four times across the business).

Start small, scale fast

Facebook started at Harvard before world domination. The organisations I've seen get self-service right don't try to do everything at once. They solve a specific problem for a specific team first, and they make sure the solution is genuinely better than the current duct-taped workaround. If it's not clearly better, adoption won't happen regardless of how good your strategy deck looks. Prove the model with one team and one use case, then scale from there.

Governance

Vinted has clear rules and buyer protection, but if you break them you're out. Data governance works the same way. Rules are the easy part, most organisations have a governance policy somewhere. What actually matters is how fast you know when they're broken and what happens next.

There's a good piece of research from HBR that found network effects alone don't determine whether a platform succeeds, governance does. Gartner estimated that poor data quality costs organisations an average of $12.9 million per year, and a big part of that comes down to unclear ownership and inconsistent standards. Governance that lives in a document nobody reads isn't governance, it's wishful thinking.

Guides and resources

The App Store's published guidelines, documentation, and review processes make building standardised. Developers know what's expected, what's allowed, and how to get through the review process.

There's a concept I like called the "thinnest viable platform," the simplest thing you can build that still reduces friction for the people using it. That's exactly the right mindset for your data standards and documentation. The how has to be obvious. Meet people where they are, not where you think they should be. If adopting your data standards requires reading a 40-page manual, adoption will be close to zero.

Self-Service Data has to be deliberate

None of these six things happen on their own, and giving everyone a Power BI licence without thinking about discoverability, trust, governance, and support is how you end up with 200 reports and nobody using any of them.

The good news is you don't need to tackle all six at once. Pick the one your organisation is worst at (for most, it's discoverability or trust) and start there. Self-service doesn't happen by accident. It's designed.

References

• Parker, G., Van Alstyne, M., and Choudary, S.P., Platform Revolution (2016)

• Hagiu, A. and Altman, E.J., "Finding the Platform in Your Product," Harvard Business Review (2017)

• Zhu, F. and Iansiti, M., "Why Some Platforms Thrive and Others Don't," Harvard Business Review (2019)

• Skelton, M. and Pais, M., Team Topologies (2019)

• Forrester Consulting / Airtable, "The Crisis of Fractured Organizations" (2022)

• Gartner, "Data Quality" research (2021)