Your roadmap and your opportunity solution tree disagree because they're maintained separately. Write one tree — outcomes, opportunities, solutions, assumptions — and project it two ways: the OST for discovery, the roadmap for delivery. Same file, so they can't drift.
Start typing — or load an example.
Teresa Torres's Opportunity Solution Tree keeps solutions tethered to customer needs and a desired outcome; a now/next/later roadmap communicates sequence and intent. Teams keep both — maintained separately — and they drift: roadmap items lose their why, tested-nowhere solutions quietly become commitments. Here both are projections of the same text, so drift is structurally impossible.
The roadmap columns are derived from discovery status, never declared: NOW holds what you're delivering, NEXT holds what you're still testing, and LATER holds opportunities — needs you've acknowledged but haven't committed a solution to. That takes Janna Bastow's rule that later items should be problem statements to its logical end, and it means the roadmap cannot claim more certainty than your discovery supports.
The projections also audit each other: solutions with no opportunity ancestry land in a no why lane; committed solutions whose assumptions are all untested get an untested bet badge; a broken assumption on something you're delivering is the loudest flag on the page. Sources: Torres, Continuous Discovery Habits; Bastow on now/next/later.