Field Notes: What Last Week On the Road Taught Me About Writing Strategy Down
From a Veriff leadership offsite in Barcelona to the Product Academy Summit in Zurich — four ideas about why the discipline of putting it on paper still beats the discipline of talking about it
Last week was shaped by two main events. One was our own Product, Engineering and Design leadership offsite in Barcelona. The other was the Product Academy Summit in Zurich, where I was lucky enough to be Tanja Lau’s guest.
Different rooms, different agendas, but in each, a lot of new energy and little nuggets bringing fresh ideas.
This week’s issue is a bit of a mixed bag. Four short stories. One thread.
Why We Still Get on a Plane for This
Let’s start with Barcelona.
There’s a tempting argument, especially this year, that distributed teams with good tools don’t need to fly people to a hotel conference room to “align.” We have docs, we have Slack, we have AI summarizing everything for us. Why burn three days and a travel budget?
Here’s why: tension that gets summarized doesn’t get resolved. It gets postponed.
The value of an offsite isn’t the keynote slides or the team dinner — although those matter too. It’s that when Product, Engineering and Design leadership are in the same room for few days, the disagreements that normally live in the gaps between teams — the ones nobody quite has the standing or the occasion to raise — surface early, while they’re still cheap to resolve. A misalignment that costs an hour of friction in a workshop in Barcelona costs a quarter of friction if it’s left to surface in a roadmap review six months later.
Tension surfaced in a workshop is a discussion. Tension surfaced in Q3 is a crisis.
The format we used to get there is the part worth sharing.
The 6-Page Memo: A Forcing Function Disguised as Homework
Each area of the business came to the offsite having written a strategic memo. Max six pages. Structure was fixed for everyone:
Link to company goals
Current state
Target outcomes
Strategic bets
Risks and dependencies
That’s it. No fifty-slide decks, no “let me walk you through the context” preamble. Six pages, same skeleton, every area. Believe me when your area has 4+ teams, it requires a lot of neurone power to get it right.
What I liked about this format is that it does two things at once, and they pull in opposite directions in a useful way.
It forces crispness: Six pages and a fixed structure mean you cannot hide behind volume. If your “strategic bets” section is twelve bullet points, that’s not twelve bets — that’s a list, and lists are what you write when you haven’t decided yet. The format quietly punishes vagueness. You’re forced to choose the two or three bets you actually believe in, because there’s no room to gesture at everything.
It forces clarity for the reader, not just the writer: Because the structure is identical across every area, the format also acts as a translation layer. A leader reading the Platform team’s memo and the Identity Verification team’s memo can hold both in their head using the same mental scaffolding — goals, current state, target outcomes, bets, risks. No one has to learn a new way of reading just because the topic changed.
And here’s the part that surprised me a little: nobody felt like the format constrained them. If anything, everyone left feeling like their narrative finally sounded compelling and coherent (though all of us still have work for the readers to acknowledge they are coherent and compelling). That feeling comes because the format had quietly done the editing work for us. No one is happy presenting a strategy that sounds like a list of activities. Everyone is happy presenting a strategy that sounds like a story with a point.
Constraining the content, forcing it to be written in prose not bullet points yields better formed narrative.
From Barcelona to Zurich: The Product Academy Summit
The next day after our BCN offsite, I was in Zurich for the Product Academy Summit, as Tanja Lau’s guest. Three talks stuck with me and, in hindsight, all three are circling the same idea as the memo exercise: making your choices explicit is the actual work of strategy.
Tim Herbig and the Stupidity Flip
Tim Herbig shared a deceptively simple test for whether a strategy is actually a strategy: the Stupidity Flip.
Take your strategy statement, written as a short one-liner. Then write its literal opposite. If the opposite sounds obviously stupid, your original statement was NOT a real strategic choice. If the opposite sounds reasonable, maybe even sensible, then your original statement was a real choice between strategic options, and you chose to let a viable option go by to focus on another one.
“We want to grow” flips to “we don’t want to grow.” Nobody runs a company on “we don’t want to grow” — so “we want to grow” was never a strategy. It’s table stakes dressed up as direction.
A real strategic bet flips into something that sounds genuinely uncomfortable, maybe even wrong, to a chunk of your stakeholders. “We will develop Product X with internal capabilities” flips to “We will acquire a start-up that developed Product X adn scale it” — and that flip sounds plausible, which is exactly why the original is a real bet. Someone, somewhere, will be unhappy about it. That’s the signal.
A strategy choice is one for which the opposite is not stupid on its face. — Roger Martin
This is the one-liner version of what the 6-page memo does at the document level. Both are constraints designed to expose whether you’ve actually chosen something — or just described a direction everyone already agreed with before you opened your mouth.
Dominique Jost on How Todoist Builds With AI — Together
Dominique Jost walked through how Todoist evolved their internal ways of working with AI, and it’s one of the more concrete “how do we actually operationalize this” stories I’ve heard.
Phase one: they opened the gates. Anyone at Todoist could pick up an AI coding assistant and push a PR. Democratic, fast, very 2025.
Phase two: they noticed the problem that democratic access always creates — people were vibe-coding in every direction. Lots of energy, lots of PRs, but scattered across too many topics with no shared sense of *which problems mattered most*. Motion without direction.
Their fix wasn’t to restrict who could build. It was to curate what got built. They created a library of projects — a backlog of problems the company actually wanted solved — and if you wanted to spend your AI-assisted time vibe-coding something, you picked from that library. Same access, same tools, same speed. But now every contribution, however informal, was progress toward a list someone had deliberately curated.
Curating what gets built matters as much as deciding who gets to build it. So PM jobs are safe!!!
Stephanie Hürland: When Customers and Your Data Model Speak Different Languages
The last one is my favorite, because it’s a live example of something I’ve written about before.
Stephanie Hürland talked about how ZDF Spark ships AI-based projects fast — but the story that stuck with me was one she ran for Adidas: an AI assistant to help customers find the right shoe for their needs.
They shipped a first version. And quickly ran into a wall that had nothing to do with the AI itself: the way customers ask for shoes didn’t match how Adidas internally defines a shoe’s atomic characteristics. Customers describe what they need in terms of activities, feelings, contexts — “something for long runs that won’t wreck my knees,” “I need running shoes for mix terrain asphalt/grass”… Adidas’s internal supply-side data model describes shoes in terms of cushioning specs, drop height, upper material, last shape — the vocabulary of design and manufacturing, not the vocabulary of a person standing in a store trying to solve a problem.
The result: a demand signal, expressed in customer language, that was hard to match against a supply catalog encoded in a completely different language.
This is the Mental Model Gap I’ve written about before — the distance between how a product team thinks about what it offers and how customers think about what they need. What makes the Adidas case interesting is what it implies about AI-native interfaces: when the front door to your product is natural language, customers stop having to translate their intent into your taxonomy. They just say what they want. And suddenly, the gap between demand expressed naturally and supply encoded internally becomes visible — often for the first time — because there’s no UI of dropdowns and filters quietly forcing the translation on the customer’s behalf anymore.
What I loved is the fix wasn’t more AI. It was going back to people working in physical Adidas stores and asking them: how do you actually talk to customers? How do you figure out which shoe to recommend when someone describes a problem, not a spec? That conversation became an input into rebuilding the data architecture itself — so the supply side could finally be described in terms that matched how demand was actually expressed.
When the interface becomes natural language, the gap between how customers ask and how you’ve encoded your product stops being hidden. It becomes the project.
Conclusion
That’s a strange thing to say in 2026, when AI can generate strategy decks, one-liners, and code in seconds. But maybe that’s exactly why it matters more now, not less. The bottleneck was never the writing. It was — and still is — the deciding.
PS: It’s Monday — are you on top of last week’s performance?
If you are skipping your Weekly Business Review, lost in dashboards, or unsure why metrics move, you’re not alone.
📘 The Power of Analytics is your guide to choosing the right metrics and building a system to truly and easily understand product performance and customer behavior.
👉 Stop wasting your time and grab your copy here.