Open Innovation Webinar · Recap · October 28, 2026

Decisions under uncertainty: 5 rules for innovation managers

Most companies don't fail at innovation because they lack ideas. They fail because nobody decides. In our October Open Innovation Webinar, Peter Borchers explained why waiting for more data is usually the riskier move, and how innovation teams can set up decisions that actually get made.

Peter Borchers
Founder of hub:raum, former CEO of Allianz X, Affiliate Professor at ESCP Business School. Hosted by Frederic from GlassDollar.

The project that dies without a no

Peter opened with a question to the audience: have you ever seen a promising innovation project die, not because someone said no, but because nobody decided? Most answers in the chat were yes. That gap between a good idea and a decision that never comes was the topic of the session.

From his time at Deutsche Telekom and Allianz, Peter has seen it many times. Companies rarely fail at innovation for lack of ideas. They fail because they don't decide, and they don't decide because they aren't good at dealing with uncertainty. Customers, competitors and technology keep moving, and in innovation you decide about something that may lie years ahead.

Why waiting for more data is the riskier move

The natural reflex is to wait for more data and more certainty. Peter's point is that this rarely works in innovation. The data you need most is how real customers react to something new, and it doesn't exist yet. It only shows up once you start testing your hypotheses.

Waiting isn't neutral either. It's a decision in itself, just one that nobody owns.

Peter was clear that this is no argument for acting blindly on big bets. It's an argument for small, fast ones.

The 70% rule and two-way doors

He pointed to Jeff Bezos's 2016 letter to Amazon shareholders:

"Most decisions should probably be made with somewhere around 70% of the information you wish you had."

Bezos separates two-way doors, decisions you can reverse, from one-way doors, which you can't. Most innovation decisions are two-way doors. You can test, stop and adjust. Waiting only makes sense when the door opens in one direction and there's no way back if something fails.

Decision sovereignty

Peter calls the skill behind this decision sovereignty: deciding well when you can't know enough, and owning the decision instead of hiding behind more analysis.

For innovation managers it has two sides. You decide well in your own area. And you make sure the decisions you can't take yourself actually get taken at the top.

AI makes this more important, not less. Data has never been cheaper, but certainty costs as much as it always did. AI gives you more options. It doesn't give you more evidence, and it can't take the decision for you.

Five rules for deciding under uncertainty

If waiting isn't the answer and magic isn't either, Peter's answer is a system. Innovation is managing uncertainty, and you manage it with a few clear rules.

1Spread your bets

Nobody can predict which project will work, not even the best venture investors. That's why an early-stage fund usually holds 15 to 20 startups or more. Inside a large company the same logic applies: run several bets in parallel, keep the budget per stage small and add money only once it's clear what works.

With one bet, every mistake is fatal. In a portfolio, a mistake is a data point.

2Learn in short loops

Running many bets only works if each one is cheap. So reduce uncertainty as cheaply as you can: no long project phases, but short pilots one after another, the way startups work because they can't afford anything else.

Everyone knows the lean startup approach, Peter said, but he has rarely seen it done consistently in large companies. There, the loudest opinion often wins when it should be the best and fastest evidence. You don't wait for data. You create it with small, cheap experiments.

3Start with the problem

Reducing uncertainty starts with a question, not a solution: which problem is valuable enough to deserve priority? Pick the wrong one and even the best solution is wrong.

Peter linked back to our June webinar, where Yaron Flint said a good challenge has a named owner in the business. Peter added one thing: a kill criterion that this owner agreed to before you started. Then the problem has someone who will actually decide about it.

This applies to new and more disruptive work. For incremental improvements to your core product you usually have the data already, and the uncertainty is lower.

4Tie it to the strategy

This rule matters most in large companies. You can have a portfolio, short loops and a well-defined problem, but if the work doesn't move one of the company's goals, it will most likely be cancelled, if not this quarter then the next. Exploration is allowed and needed. It just has to pay into a goal the company has set.

Before you test anything, Peter suggests two questions. Which company goal does this move? And who at the top really cares about that goal? Strategy fit is your license to operate.

5Make the top own the decision

Innovation teams need room, autonomy and time. What Peter sees more often is the opposite: projects get killed too early or never get the resources they need in the first place.

Some decisions can't be delegated, not to the innovation team, not to a consultant and not to an AI. You can't make the decision for your boss. But you can set it up so your boss makes it once, up front, instead of at some point later.

The decision contract

To put rule five into practice, Peter shared a simple tool: a one-page decision contract you agree on with your sponsor before the project starts. It covers:

If your sponsor won't agree to it, that's an answer too. The project probably doesn't have the backing it needs, and it's better to know that now than in month six or nine.

Put together, that's the system: spread your bets, learn in short loops, start with the problem and its owner, tie it to the strategy and make the top own the decision. Set up like this, a wrong call costs you one project, not the whole program.

From the discussion

Before opening the floor, Peter asked which uncertainty is the most expensive for the audience right now. The answers: customer value, technology and scaling, and delivery.

Will the startup's solution scale across our plants?

One attendee runs venture clienting in manufacturing. A startup may offer the best solution in a pilot, but can it roll out across every factory? Peter agreed that this step carries a lot of uncertainty.

How much proof is enough?

Another attendee finds it hardest to prove value. Ten customer interviews might point one way, but you only really know with 20, 30 or 50 proof points. Peter's suggestion was to go beyond interviews and test in the market, with real people using a product, a prototype or a feature. Both agreed that doing lean startup consistently is hard, mostly because you need people's time and attention again and again.

What if everyone wants one more round of analysis?

Frederic asked what an innovation manager should do when "let's do one more round of analysis" is the easiest thing to say. Peter's answer: ask which decision actually depends on that analysis, and how the result would change it. If nobody can answer, the analysis isn't helping. It's a way to delay.

Launch earlier than feels comfortable

Frederic shared a recent lesson from GlassDollar. A product went out later and more polished than it needed to. The product was good, but the market didn't care, and launching earlier would have taught the same lesson sooner. Peter tied it back to the two-way door: if you can come back through the same door, try it. Run several experiments at once and a failed test is just a step toward the one worth doubling down on.

When do you stop looking for a better solution?

There's always another startup that promises to do it better. So when do you commit? You only learn by testing with real customers, Peter said. When in doubt, go out with an 80 or 90% solution and improve from there, rather than bring nothing to market at all.

Full disclosure, as Peter said at the start of the session: he is an angel investor in GlassDollar and sits on its advisory board. You can find him at pbo.vc.