A board's forecast page has always answered one question well: when does this whole backlog clear. What it's never answered is the question that actually drives a client conversation about priorities — of the five epics on this board, in what order do they land, and when does each one actually finish given everything ahead of it. Roadmap, in the Board menu, answers that directly, as one timeline instead of five separate forecasts you'd have to mentally stack yourself.
What it does
Open it and every epic with at least one remaining ticket already has a band on the timeline — nothing to click first. They're ordered by Value against how much is left to do, so the epic worth the most for the least remaining effort naturally sorts to the top. Each band runs from wherever the epic ahead of it is expected to land out to its own date — the very first epic's band starts today, since nothing's queued ahead of it yet.
Three buttons above the timeline — Aggressive (P50), Commercial (P85), Safe (P95) — switch which scenario the whole page reads at, and Aggressive is where it opens. This isn't a second line drawn over the first: clicking Safe re-derives every band from scratch at P95, so the second epic's date, the third's, and everything after it are each that epic's own P95 — never a mix of scenarios stacked down the list.
The part worth understanding before you read it
An epic's own forecast page asks a different question: "if this epic alone had the whole team, starting today." That's a fair question on its own — but it's not one two epics on the same board can both answer "yes" to at once, which is exactly why Roadmap doesn't use it for the timeline. Instead, a Roadmap band assumes the team works the list in order, one epic at a time, at its normal pace: the second epic's band only starts once the first one's has probably finished, the third's only after the first two's.
That has a real consequence worth knowing before you read a band's width as an epic's own duration: two epics can show an identical worst case on their own forecast pages — each assuming it alone gets the whole team, starting today — and still end up with very different band widths on the Roadmap, because it's rare for both epics to hit their own worst case in the very same simulated run. Read a band's right-hand edge as that epic's real finish date given everything queued ahead of it; for how long the epic takes on its own, its own forecast page is still the right place to look, and every row links straight there ("View forecast," or "Run forecast" if nobody's triggered one for that epic yet).
One more thing that falls out of the same logic: reordering an epic's Value can shift every date after it, because it changes that epic's place in the queue. That's expected, not a bug — the whole page recomputes downstream of wherever an epic now sits.
Reordering it yourself
Value and remaining effort decide the order by default, but "Reorder epics" — next to "Back to board," once there are at least two epics to put in order — lets you override it directly: drag epics into whatever sequence the team will actually work them, and save. That order then wins outright for the page, exactly as if you'd changed the Value field on every epic to match, and every date and band recomputes against it. It's a Roadmap-only setting — it doesn't move a ticket on the board itself, so getting a ticket into To Do to actually start that epic's work is still a separate, manual step. An epic added after you've saved an order sorts back in by Value/Effort, after everything that already has a saved position, until you fold it into a reorder of your own.
What doesn't get a band
An epic with no remaining tickets falls back to its own manual due date instead, shown as a plain marker. An epic with neither remaining tickets nor a due date drops into an "Unscheduled" list below the timeline — so nothing goes quietly missing from its own roadmap just because it has no forecastable work left.
And if nobody on the board has ever run a forecast at all — board-level or for any individual epic — there's simply no throughput history yet to compute a single date from, and the page says so plainly. Running any forecast anywhere on the board, the dashboard's own "Run forecast" or any epic's button, is all it takes for Roadmap to start computing dates from that point on.