Sales told a customer Q3. Support told a customer whose ticket has been open since January. The board saw it in a slide with a green checkmark next to it. Then engineering picks it up in April, finds the API it depends on doesn't do what everyone assumed, and quietly moves it to Q4. No meeting. No announcement. Just a date that changed on a doc almost nobody rereads once it's been agreed to once.
That's the moment a roadmap stops working, and it's rarely the date slipping that causes the damage. It's who found out, and how.
A wishlist can change for free
Treat the roadmap like a wishlist and every item on it is just a hope, revisable on a whim, because nobody outside the room where it changed is depending on the old version. Nobody built a sales conversation around it. Nobody staffed a launch around it. The cost of changing your mind is zero, so change happens quietly and often, and the roadmap becomes something everyone nods at in the quarterly meeting and nobody actually trusts between meetings.
A treaty can't
A treaty is different, and not because the words in it are fancier. It's different because other parties made decisions based on it — decisions that cost something to unwind. Break a treaty term and there's a process: you go back to the table, you say what changed, you say why, and everyone affected finds out at the same time, from the same conversation, not from a stale slide or a customer asking why the thing they were promised isn't there.
That's what a roadmap actually is, whether anyone calls it that or not. Sales quoted a date to a prospect. A customer built their own rollout plan around a promised integration. The board reported a number to investors based on what's supposedly shipping. Every one of those is a commitment made in reliance on the roadmap being true. Moving the date without renegotiating doesn't make the commitment go away — it just means the person holding it finds out the hard way, later, from someone who isn't you.
What renegotiating actually looks like
It doesn't require a formal ceremony every time a ticket moves. Internal sequencing shifts inside a sprint — that's implementation detail, nobody outside engineering needs a memo. The line is whether someone outside the team made an external commitment based on the old version. If sales quoted the date, if a customer is waiting on it, if it showed up in a board deck — that's a treaty term, and changing it means telling the people who relied on it, before they find out on their own. A two-line message — "this is moving from Q3 to Q4, here's why, here's what it means for you" — costs almost nothing and buys back the trust that silence spends.
The real cost of treating it as a wishlist
Do this enough times and the roadmap stops being useful even as a wishlist. Sales stops quoting it, because they've been burned. The board starts asking for weekly check-ins instead of trusting the quarterly one, because the quarterly one has been wrong too often without warning. Customers start asking for it in writing, with penalties, because a verbal date stopped meaning anything. Every one of those is the organization building its own workaround for a roadmap it no longer believes — which is a slower, more expensive version of just keeping the original promise honest in the first place.
A wishlist gets to be wrong for free. A treaty has to be renegotiated out loud, in front of whoever it affects. The dates on a roadmap will move — that's not the failure. The failure is when they move and nobody who was counting on them gets told.