Every roadmap argument is really an argument about certainty. Stakeholders want dates because dates let them plan. Product teams resist dates because the work is uncertain and dated commitments become sticks. Both positions are reasonable, which is why the fight never ends.
The resolution is not refusing dates. It is being precise about which things you are certain of.
Three formats
| Format | Communicates | Costs you |
|---|---|---|
| Timeline / Gantt | Specific things on specific dates | Certainty you do not have; every slip reads as failure |
| Now / Next / Later | Sequence and rough horizon | Frustrates anyone who genuinely needs a date |
| Outcome-based | Problems you will solve and the metric that moves | Demands real trust; reads as evasive without it |
Now / Next / Later, done properly
The workhorse. Three columns with deliberately decreasing precision — which is honest, because your confidence genuinely decreases the further out you look.
- Now — in progress. Named, scoped, with a real confidence level. Roughly this quarter.
- Next — validated and queued, not started. Some scope will change. Roughly next quarter.
- Later — problems worth solving, solutions unknown. Deliberately vague. Six months plus.
The discipline that makes it work: items must move backwards sometimes. A roadmap where things only ever advance from Later to Now is a queue, and everyone will correctly infer that Later means "eventually guaranteed" rather than "maybe".
High / medium / low next to each item does more for stakeholder trust than any format change. It lets someone plan around your high-confidence items and treat the rest appropriately, instead of guessing which commitments are real.
Outcome-based roadmaps
Instead of "Bulk export, Q3", you commit to "Reduce time-to-first-report from 40 minutes to under 10, by end of Q3" — and keep the solution open.
This is the most intellectually honest format and the hardest to introduce. It requires an organisation that trusts the team to choose solutions, and a stakeholder group that cares more about the problem than about a specific feature they already imagined.
Do not attempt it during a trust deficit. If your last three commitments slipped, switching to outcome language reads as avoiding accountability. Rebuild credibility with a few well-hit Now commitments first, then change the format.
When someone genuinely needs a date
Sometimes the request is legitimate — a contract, a conference, a dependent team, a regulatory deadline. Refusing on principle is bad product management.
Give the date, with the shape of the uncertainty attached:
Highly confident by end of March for the core flow. The reporting piece depends on the data migration, which we will know about in two weeks — I will confirm or move it then.
That sentence gives them something to plan with, names the dependency, and sets a checkpoint. It is far better received than either a bare date you will miss or a refusal to answer.
When someone demands a feature by a date, ask what they need to do on that date. Frequently the underlying need is served by something much smaller — a manual process, a report you run for them once, an export. You cannot find that out while defending your roadmap.
Saying no without spending goodwill
Three reliable moves:
- Trade, do not refuse. "We can do that — it displaces the onboarding work. Which matters more this quarter?" This converts a rejection into their decision, and the answer is frequently "leave it".
- Route it to the tree. "What problem does that solve for them?" moves the conversation from solution back to opportunity, where you can compare it honestly against everything else.
- Say not-now, and mean it. With a date you will revisit. "Not now" that is actually "never" is discovered eventually, and it costs more than a clean no.
Cadence
- Update monthly. Quarterly is too slow to be trusted; weekly is churn.
- Announce movement explicitly. Silently deleting an item is how stakeholders learn to stop believing the document.
- Keep one version. The moment a separate roadmap exists for the board, both become fiction.
- Show what shipped. A "recently delivered" column is the cheapest credibility available and almost nobody includes it.
For deciding the order in the first place, see prioritisation frameworks. For proving the outcomes actually moved, see North Star metrics and the numbers that mislead.