Product strategy
The objective was right. The reason behind it was a rumour.
One strategy in the FlowDesk demo workspace, built over a quarter. Watch the six steps assemble, then read what happened at each one.
Everyone in the room believed it. Nobody could say where it came from.
Step 01 · July
A belief nobody could source
FlowDesk sells shift scheduling to mid-size logistics operators. 400 customers, mostly Benelux, all signed in the last three years.
The VP of Product opened Q4 planning with the number that had been following her since spring: 22% of new accounts were gone inside 90 days.
She wrote the objective in ten minutes. Under it she wrote the thing everyone in the room already believed — onboarding is too hard, so we will make onboarding easier.
The objective went in as assumed. Nothing was attached to it. One question came back: what evidence shows that onboarding is where these accounts are lost?
The honest answer was that a sales engineer had said it in a meeting in March, and it had been true ever since by repetition.
The ASSUMED badge does one thing. It makes a claim that hasn't earned its place visible as a claim, to everyone who opens the objective.
Steps 02–03 · two weeks
Nothing here was new. It had just never been in one place.
FlowDesk already owned all of it.
The shape of it surprised everyone. The accounts that churned had finished onboarding. They imported their rosters, published a first schedule, and stopped.
What separated the survivors was the second user. And the person who mattered turned out to be the dispatcher, while the contract had been signed by the ops manager.
The original objective stayed. Its stated cause did not.
Step 04
Three solutions, each wearing its assumption on its face
An assumption you write down can be tested. An assumption you hold can only be argued about.
Their CEO wanted the done-for-you import service — a person from FlowDesk sitting with each new account for a day. The most expensive option, and the one with the most conviction behind it.
The third came from a support engineer: a shared schedule link anyone could open without an account.
In most planning rooms, the most senior conviction settles that argument. Here each option carried a different belief, and one of those beliefs was cheap to check.
Done-for-you import service
Assumes the blocker is effort.
Invitation prompts inside onboarding
Assumes the ops manager knows who to invite.
Shared schedule link, no account needed
Assumes dispatchers want the schedule and the account wall keeps them out.
Step 05
The prediction was written down before the test ran
Which is the only thing that makes a result mean anything afterwards.
The import service survived as a parked solution, with its assumption still unproven and the evidence for it visible to anyone who opened it.
When the CEO asked about it in November, the answer took eleven seconds to find and required nobody to remember anything.
Written before the test · falsifiable
If fewer than half the accounts share the link within a week, the assumption is wrong and the solution dies.
Step 06 · six weeks
Status went where the engineers live. Intent stayed where it started.
Nobody was asked to migrate, and nobody had to write a status report twice.
The validated solution became an epic, then four features. Status lived in the tracker the team already used. Intent stayed linked to the opportunity, still linked to the objective, still carrying the 41 pieces of evidence and the experiment result.
The release shipped on 14 November and moved into assessment, because on 14 November nobody yet knew whether it had worked.
NINETY DAYS LATER
22% → 15%
Target was 12%. The objective did not close.
Seven points, attributable
The same cohort analysis that found the pattern could show which accounts gained a second user, and what happened to them.
Three points, still open
They went back into the graph as a question, with the two parked solutions sitting next to them.
January
She presented to the board without building a deck. She opened the objective, and the chain was already there: the number, the evidence that reframed it, the solutions that lost, the one that won, the test that decided it, the code that shipped, and the metric that moved.
Before you worry about the whole tree
FlowDesk ran one objective. That was the whole rollout.
No portfolio was mapped first. One number that mattered, one honest question about it, and the rest followed. Every level is optional until you need it.
One objective, marked assumed
Pick the number that follows you around. Attach nothing. Let the badge embarrass you.
One opportunity, evidenced
Pull in what you already own. Most teams find the reframe here.
The rest of the portfolio
Add teams and objectives once one chain has proved itself. The floor is low; the ceiling is the whole tree.
She presented to the board without building a deck.
She opened the objective, and the chain was already there: the number, the evidence that reframed it, the solutions that lost, the one that won, the test that decided it, the code that shipped, and the metric that moved.
Fair questions
What people ask before they commit a team
- We already have OKRs. Does this replace them?
No. Your objectives stay your objectives, and this attaches evidence to them. Marta's objective never changed. What changed was that its stated cause stopped being a rumour.
- Do we need all the levels?
No. The demo uses one objective, one opportunity and one experiment. Add a level when a claim needs somewhere to sit.
- Will my PMs actually maintain it?
Only if it's where the planning happens. If it becomes a second place to report status, it will rot the way every other one has. The design intent is that the board is the artefact they would have written in a doc anyway, and that it replaces the status report.
- Do engineers have to move tools?
No. Epics and features sync to the tracker they already use. Status lives there, intent lives here.
- What happens when an objective doesn't close?
It stays open, with the part you can attribute separated from the part you can't. In the demo, seven of the ten points moved and the team could say which seven. The other three went back into the graph as a question.
- How long before we see anything?
The honest answer is that it depends on how fast you can connect the documents you already have. For a ten-person team, that has typically taken about five days. The first useful output is usually the discovery that a belief you'd been planning around doesn't hold.
What are you planning around that nobody can source?
Bring one objective and the evidence you already have. We'll show you the same six steps on your own numbers.
FlowDesk is the Shorter Loop demo workspace. The six steps, the badges and the links between them is live in it — open it and check any number here.