Experiments and validation: Find out what works before you build it

We validate early on whether solutions actually create value.

No idea is successful straight away. However a solution came about – from a workshop, a stakeholder request or a sudden flash of inspiration – at first it is nothing more than a hypothesis, without a single data point showing that it will achieve the desired effect. We help product teams to identify their most critical assumptions, to test them with the right experiment at the right time and to make clear decisions based on the results. In case of doubt, success consists of abandoning an idea early to avoid burning money. Hands-on, pragmatic and in days, not weeks – with a testable prototype by the end of the week.

Typical problems we see

  • Testing to convince, not to learn. The prototype is presented in a way that users can hardly help but like it. Negative feedback is explained away: „This person was not our target group anyway.”
  • No success criteria upfront. The team only decides after the test what counts as success – and, surprise, the result is always good enough.
  • The wrong hypothesis first. Teams test the color of the button while it is still completely unclear whether anyone is willing to pay for the product at all.
  • One MVP, then all in. A single test is run after months of preparation, and its result decides everything. Success becomes a matter of luck.
  • Only the user is asked. But not the buyer who approves the budget, the admin who operates it or the partner who sells it.

Why validation matters more than ever

With AI coding tools, a clickable prototype can be built in hours, and even functional software can be built significantly faster than before. That is great news – but it shifts the risk. When building is almost free, the question is no longer „Can we build it?”, but „Should we build it?”. AI makes building cheaper. It does not tell you whether you are building the right thing. That is exactly what experiments are for.

Of course, we use AI in validation ourselves – wherever it makes experiments faster: prototypes in hours instead of weeks, landing pages for a quick demand test in minutes, several solution variants tested against each other and the analysis of test sessions. What AI does not replace is the context: the real target groups, the success criterion defined before the test and the courage to take a decision. A working AI prototype is not validation – it only proves that something can be built. And a hypothesis „validated” by asking a language model is not validated.

How we support you

We help product teams to turn solution ideas into validated knowledge – quickly and with as little effort as possible. No fixed framework, but exactly the tools that fit your situation:

  • Surfacing critical assumptions: Which assumptions does your solution rely on – across desirability, viability, feasibility and strategic fit? We make them visible and prioritize them by criticality and existing evidence, e.g. with Assumptions Mapping by David J. Bland.
  • Testable hypotheses: „Users will love our new onboarding” cannot fail. We formulate hypotheses that can – with a metric and a success criterion defined before the test, not after.
  • The right test for the level of confidence: The more certain the team is about the direction, the more effort the test may take. We design the simplest experiment that is „good enough” for the next decision.
  • Prototyping: From paper prototypes to AI-generated clickable prototypes and functional software – the right testable prototype in a fraction of time.
  • Running experiments: Usability tests, fake doors and landing pages, concierge and Wizard of Oz tests, pilots and A/B tests – with all relevant target groups, not just users.
  • Documenting and deciding: Lean documentation, e.g. a hypothesis backlog, so the same questions are not discussed again and again. After each experiment, the team decides: persevere, pivot or stop.
  • Feeding results back: We feed the insights back into strategy and mandate, research and ideation – and support the transition to delivery.

 

How we work with you

  • Validation embedded in your team: We work directly with your product team – from the critical assumption to a tested prototype within a week, including the decision on what comes next.
  • Assumption workshops and experiment design: One workshop with the team and stakeholders to surface and prioritize the critical assumptions – or a review of planned experiments before money goes into them.
  • Coaching product teams: We accompany your teams over several weeks while they run their own experiments – so that the skills stay in the team.
  • Interim product leadership: We take over as product manager, head of product or CPO until the right person is on board – and establish evidence-based decisions in the meantime.

Also available as in-house training

Want your team to build these skills itself? Our „Experiments and validation” deep dive – derived from the Product Discovery Masterclass – is tailored to your team and works with your real challenges. Possible contents: identifying critical assumptions, formulating testable hypotheses, choosing the right test for the right level of confidence, prototyping including AI prototyping, experiment types from fake door to A/B test, deciding on results and Lean Startup in practice. In English or German, on-site at your company – e.g. in Berlin, Hamburg, Munich, Frankfurt, Cologne, Vienna or Zurich – or remote.

Frequently asked questions

What is a product experiment?

A product experiment is a structured test of a critical assumption behind a solution – before the solution is fully built. It consists of a clearly formulated hypothesis, a test method, a metric and a success criterion defined upfront. It does not test the product itself, but the assumptions it relies on: Do the target groups have the problem? Do they use the solution? Will they pay for it?

If AI can turn our idea into working software so much faster – what is left to test?

Everything that matters. A working prototype answers the question of feasibility – and even that only partially. Whether the target groups need it, whether it creates enough value to pay for itself and whether it fits your strategy is not answered by the fact that it runs. AI makes experiments faster. It does not make them unnecessary.

Our stakeholders don't like the term „experiment". Should we call it something else?

No. It is tempting to rename experiments as „pilots” or „proofs” so that they sound less uncertain. But this does not solve the real problem – that the views on how product success is created diverge. Instead, make the case openly: Every solution is a hypothesis anyway. The only question is whether you find out early and cheaply, or late and expensively. Small, well-framed experiments with visible results are usually the best argument.

What if an experiment fails?

Then it has done its job. An experiment that invalidates a critical hypothesis saves you from investing months into the wrong solution. Usually, the results lead to adjustments and the next test.

The other areas of product discovery

Experiments and validation is one of four areas every product discovery has to cover. They are closely connected – insights from one area regularly change the others:

All four areas at a glance: What we do · All four areas in one training: Product Discovery Masterclass

Get in touch

Do you want to find out faster which of your ideas actually create value?

Let's talk