Start from the problems to solve, not the features to ship. Each line says which number it should move, by how much, and how you'll know. A list of projects with dates is just a schedule.
The benchmark
Three kinds of requests compete for your capacity: growth projects, requests from other teams and site maintenance.
Decide the split before you fill in a single box, for example half for growth, a third for requests and the rest for maintenance. Adjust these proportions as needed.
What matters is setting them in advance, before the first urgent request sets them for you.
Almost every roadmap the authors review is missing one line: the reserve. In their own engagements, they keep a very simple table, with planned time in one column and time actually spent in the other. In the first half of 2026, across three of those accounts, actual time exceeded the plan by 13 to 20%. On a fourth, a single month absorbed 114 hours instead of the 25 planned: no one had planned for the unplanned, which is more or less its definition.
The authors plan roadmaps all year, and they miss by the same margins as the teams they work with.
So your reserve already exists. It's just invisible.
It draws itself down from the most fragile project, often a growth project, because it has no deadline imposed from outside.
Setting aside 15 to 20% of capacity means making visible a cost you're already paying, and choosing what it protects.
The rule
Nothing goes in without an expected effect and a way to measure it, and nothing goes in unless something comes out or gets pushed back.
The steps
- The allocation. Set it for the quarter, reserve included, and write it at the top of the document.
- A card for each idea. It has four fields: the problem, the target metric, the estimated effort and the person responsible.
- The ranking. Weigh the expected effect against the effort. The RICE method, described by Sean McBride at Intercom, adds a confidence factor meant to curb enthusiasm for appealing but poorly supported ideas. An idea with no data behind it then drops down the list on its own.
- The monthly review. Reread the list in light of the results and cut what didn't move anything, even if the project shipped on time.
- The line for leadership. Keep it ready for the next urgent request:
"To fit this project in this quarter, I suggest pushing back this one, which was meant to gain two points on checkout completion. Which of the two do you prefer?"
The template
Copy it, then rewrite it with your own lines. The figures in the table are made up for the example.
| Problem | Target metric | Expected effect | Effort | Confidence | Owner | Category |
|---|---|---|---|---|---|---|
| Mobile shoppers drop off at the delivery choice | Delivery-to-payment rate | +2 points | 2 weeks | Medium | Product | Growth |
| Customer service can't find modified orders | Ticket handling time | Cut in half | 1 week | High | Customer service | Requests |
| The reviews app slows down product pages | Mobile load time | One second less | 3 days | High | Tech | Maintenance |
| Unplanned work | None | None | 15 to 20% of capacity | None | Whole team | Reserve |
The pitfall
The most common case is the roadmap filled to 100%. The first incident, and there will be one, pushes everything back, and trust along with it.
A roadmap is also a tool for saying no: when the urgent request arrives, show the person behind it what you'd have to remove to make room.
The conversation changes, and sometimes the person asking chooses to wait.
