Execution

The 90-Day Roadmap: Where Strategy Stops Being a Document

By Rishi Desai7 min read

Every failed strategy I've ever autopsied died the same way: not in the thinking, but in the sixty days after the offsite. The document was fine. The team even agreed with it. Then the quarter's urgencies arrived, the strategy went into a drawer, and by month three the company was back to running on instinct, now with a nicer-looking artifact to feel guilty about.

The 90-day roadmap is the anti-drawer device. It's the bridge between "what we decided" and "what anyone does on Tuesday," and it has exactly four load-bearing components.

1. Initiatives, not intentions

"Improve retention" is an intention. "Rebuild onboarding so 60% of new customers hit first value within 14 days" is an initiative. It has a definition of done. A good 90-day roadmap holds five to eight initiatives, never more. If everything is a priority, you've re-created the to-do list with better formatting. Each initiative must trace directly to one of the strategic choices on the one-pager; anything that can't name its parent choice gets cut, however attractive.

2. One owner per initiative, a name, not a team

"Marketing owns this" means nobody owns it. Every initiative gets a single human owner who reports on it weekly. In founder-led companies this surfaces an uncomfortable math problem immediately: there are more initiatives than credible owners. Good. Better to discover the capacity constraint at planning time than at the post-mortem. Cut the roadmap to fit the people, never the reverse.

3. Sequencing by dependency, not enthusiasm

Most roadmaps are secretly ordered by excitement. The correct order is by dependency and by information value: what must be true before other things can start, and what will teach you the most soonest. Fix the checkout before buying traffic. Reprice before scaling sales. Run the cheapest test of the riskiest assumption first. A roadmap that front-loads learning gets smarter every month; one that front-loads the fun stuff gets surprised in month three.

4. A weekly cadence with teeth

The roadmap lives or dies in a standing 45-minute weekly: each owner, two minutes, on track or not, the blocker if not, the decision needed from the room. Metrics first, discussion second, decisions logged. That's the whole liturgy. Two rules keep it honest: no status theater (the dashboard is pre-read, the meeting is for exceptions and decisions), and no silent slippage (an initiative can be consciously re-scoped or killed, but it cannot just quietly stop being mentioned. That's how strategies die of neglect rather than decision).

Founder action: Take your current priorities and subject them to the four tests: definition of done, single named owner, dependency-based sequence, and a weekly slot on the calendar where they're reviewed. Whatever fails a test isn't on a roadmap. It's on a wish list. The free 90-Day Roadmap Template walks the whole structure.

None of this is glamorous, and that's rather the point. Strategy gets the applause; cadence gets the results. The companies that outperform aren't the ones with the most brilliant documents, they're the ones where, ninety days later, the document and reality still describe the same company.

Build Your 90-Day Plan, Start Free Get the Roadmap Template

Roadmaps fail for four reasons, and only one is ambition

Capacity fiction. The plan assumes people have time they do not have. Nobody is at 0% utilisation when the quarter starts. They are already running the business. A roadmap that adds work without naming what it displaces is not a plan, it is a wish with dates. The test: for each owner, what did they stop doing to take this on? If the answer is nothing, the item will slip and everyone will act surprised.

Ownership diffusion. "Product and marketing" owns nothing. One name per item, and that name is a person with the authority to make the call, not a committee that will schedule a meeting about it.

Invisible dependencies. Item seven cannot start until item three finishes, and nobody drew that line. The plan looks parallel and executes serially, which is how a 90-day plan becomes a 150-day plan without anyone deciding it should.

No kill criteria. This is the big one and the most consistently omitted. Nothing on the roadmap has a defined condition under which it stops. So nothing stops, initiatives accumulate, and by quarter three the team is carrying the full weight of every bet it has ever made.

Kill criteria: written when judgment is still cheap

The reason to define kill criteria in week one is not process discipline. It is that your judgment about when to quit is at its most honest before you have invested anything.

By week eight you will have spent money, made public commitments, and told people this was going to work. The sunk cost is real and so is the reputational cost of reversing. Every incentive at that point pushes toward "give it one more month," which is how a failing initiative consumes two quarters instead of one.

A kill criterion written in week one is a message from a version of you with no ego in the outcome. Make it specific enough to be unambiguous: "If we have not booked six qualified demos from this channel by day 45, we stop and reallocate the budget to the partner motion." Not "if it is not working." Everything looks partly like it is working when you want it to.

Say the number out loud in the room where the roadmap is agreed. Getting a team to agree the exit condition in advance is a far easier conversation than proposing a shutdown mid-quarter.

What a 90-day roadmap should actually contain

Six columns. If it needs more, the plan is doing something other than planning.

Initiative, stated as an outcome, not an activity. "Reduce trial-to-paid drop-off at the integration step" rather than "improve onboarding."

Owner, one person, named.

Capacity check, what this owner has stopped doing, or the fraction of their time this genuinely represents.

Depends on, the item or decision that must land first. Blank is a valid answer and a good sign.

Proof metric, the single number that will move if this works, and its current value. Without a baseline you cannot demonstrate anything.

Kill criterion, the condition and date at which this stops.

Ninety days holds roughly five to seven initiatives for a team of twenty. If your list has fifteen, you have a to-do list again, and the quarter has already been decided: the two things with the loudest champions will get done and the rest will quietly slip.

What to do about it

Four things worth doing this week

Add the capacity column to your current plan. For every item, what did the owner stop doing? The blank rows are the ones that will slip.
Write kill criteria for everything, today. Condition plus date. Do it while nothing is invested and the judgment is uncontaminated.
Draw the dependency lines. Anything with more than two upstream dependencies is not a 90-day item.
Cut to seven. Then to five. The cut list is more informative about your strategy than the keep list.

Questions

On this topic

Why 90 days rather than a year?

Because 90 days is roughly the horizon over which a team can hold a plan in mind and over which assumptions stay roughly valid. Annual plans get written once and referenced never; the world moves and the document does not. Ninety days is long enough for something structural to land and short enough that the plan is still recognisable at the end of it.

What if priorities change mid-quarter?

Then change them, deliberately, in the review, with the displaced item explicitly killed or deferred rather than silently carried. The failure mode is not changing the plan; it is adding to it without removing anything, so the team ends the quarter carrying twelve commitments it agreed to five of.

How is this different from OKRs?

OKRs specify what you want to achieve. This specifies who is doing it, what they stopped doing to make room, what has to happen first, and when you quit. In practice most OKR implementations collapse because they answer the ambition question and leave the capacity and exit questions entirely open, which is where execution actually breaks.

Reading is cheaper than a consultant. Up to a point.

When it stops resolving the problem, thirty minutes of structured framing usually will. Free, and useful either way.

Book a Call More insights

The 90-Day Roadmap: Where Strategy Stops Being a Document

1. Initiatives, not intentions

2. One owner per initiative, a name, not a team

3. Sequencing by dependency, not enthusiasm

4. A weekly cadence with teeth

Roadmaps fail for four reasons, and only one is ambition

Kill criteria: written when judgment is still cheap

What a 90-day roadmap should actually contain

Four things worth doing this week

On this topic

Related

The 90-Day Roadmap Template

Operating Model & Transformation

The 4D Method, D3 and D4

Reading is cheaper than a consultant. Up to a point.