{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-09-12T15:53:04.117Z"},"content":[{"type":"documentation","id":"a0fcf1d6-fe8b-492b-9042-81ecb0747fa5","slug":"now-next-later-roadmap-guide","title":"The Now-Next-Later Roadmap: A Simpler Way to Plan Product","url":"https://www.koji.so/docs/now-next-later-roadmap-guide","summary":"A Now-Next-Later roadmap organizes product work into three horizons — Now (in progress, well-defined), Next (validated, not yet scoped), and Later (unvalidated opportunities) — instead of date-based timelines, matching certainty to time. Popularized by Janna Bastow of ProdPad, it fixes date-roadmap problems: dates become promises, output is rewarded over outcomes, and learning is resisted. Frame items as outcomes not features, keep Now short, review on a cadence, and require evidence to promote items between horizons. That evidence requirement is a research problem Koji solves with always-on AI interviews: discover problems for Later, validate before Next, confirm impact after Now. Six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) make demand countable so promotion rests on evidence, not the loudest opinion.","content":"**A Now-Next-Later roadmap organizes product work into three time horizons — Now (in progress), Next (up soon), and Later (under consideration) — instead of pinning features to specific dates.** It trades false precision for honest flexibility: near-term work is specific and committed, while later work stays deliberately fuzzy. Popularized by ProdPad co-founder Janna Bastow, the format has become the default for modern, outcome-driven product teams. The hardest question it raises is not *how* to draw the columns but *what earns a place in each one* — and that is a research question. Tools like Koji answer it by turning continuous customer interviews into the evidence that decides what moves from Later to Now.\n\nThis guide explains the three horizons, how to build and run the roadmap, what belongs in each column, and how to populate it with customer evidence instead of opinion.\n\n## Why date-based roadmaps fail\n\nThe classic Gantt-style roadmap — features stacked against quarters and months — looks reassuring and is almost always wrong. It implies a confidence about the future that no product team actually has. The problems are predictable:\n\n- **Dates become promises.** Stakeholders treat a Q3 estimate as a contract, and every slip erodes trust.\n- **It rewards output, not outcomes.** A timeline measures whether you shipped, not whether anything improved for customers.\n- **It resists learning.** When research reveals a better bet, a date-locked roadmap makes changing course feel like failure.\n\nThe Now-Next-Later format fixes this by matching certainty to time. You know a lot about this month and very little about next year, so your roadmap should say a lot about the first and stay vague about the second.\n\n## The three horizons\n\n**Now** — what the team is actively building. These items are well-defined, scoped, and tied to a clear problem or outcome. Confidence is high because the work is in flight. Keep this column short; a focused team has only a few things in \"Now.\"\n\n**Next** — what comes after the current work. These items are validated enough to believe in but not yet fully scoped. You know the problem is real; you are still shaping the solution. \"Next\" is your near-term intent, not a commitment to dates.\n\n**Later** — ideas and problems worth considering but not yet validated. This is where opportunities live before they earn investment. Items here are intentionally loose — often framed as problems or opportunities rather than features — and many will never graduate. That is the point: \"Later\" is a holding area for bets you have not yet justified.\n\nThe flow of the roadmap is the flow of evidence: an idea enters Later, accumulates research support, gets validated into Next, and is committed into Now.\n\n## How to build a Now-Next-Later roadmap\n\n### 1. Anchor each column to outcomes, not features\nFrame items as the problem to solve or the outcome to move (\"reduce onboarding drop-off\"), not the feature to ship (\"add a setup wizard\"). This keeps the team focused on impact and leaves room to discover the best solution.\n\n### 2. Keep \"Now\" ruthlessly short\nA team that has eight things in \"Now\" has nothing in \"Now.\" Limit work in progress so the column reflects reality.\n\n### 3. Make movement evidence-based\nThe discipline of the format is the promotion rule: nothing moves from Later to Next, or Next to Now, without evidence that it is worth it. This is where most roadmaps quietly fall back into opinion — and where research becomes the engine.\n\n### 4. Review on a cadence\nRevisit the board every sprint or month. Promote what new evidence supports, demote what has gone cold, and retire ideas that research has invalidated. The roadmap is a living document, not an annual artifact.\n\n## What earns a slot: research, not loudest opinion\n\nThe weakness of any roadmap is that columns fill with whoever argued hardest. Now-Next-Later only works if movement between horizons is earned with customer evidence. That means a continuous stream of insight feeding the board — exactly the problem AI interviews solve.\n\nWith Koji, you run always-on customer research that keeps the roadmap honest:\n\n- **Discover problems for \"Later\"** — continuous interviews surface unmet needs and friction you didn't know to ask about, populating the opportunity column with real demand.\n- **Validate before \"Next\"** — before promoting an idea, send an AI interviewer to test it with the people who'd use it. Koji runs voice or text conversations 24/7, probes each answer with follow-ups, and analyzes every transcript automatically.\n- **Confirm impact after \"Now\"** — once something ships, interview users to check the outcome actually moved.\n\nBecause the interviews are automated and continuous, the evidence layer keeps pace with the roadmap instead of lagging a quarter behind.\n\n## Structured questions to score what moves\n\nWhen you are deciding what graduates from Later to Next, mix narrative with countable signal. Koji's six structured question types make demand measurable:\n\n- **open_ended** — the unmet need in the customer's words, with AI probing\n- **scale** — how severe the problem is, or how valuable a proposed solution would be\n- **single_choice** — which of several problems hurts most\n- **multiple_choice** — which scenarios trigger the need\n- **ranking** — order competing opportunities by importance\n- **yes_no** — would they use the proposed solution at all\n\nThat gives you both a demand score and the verbatim reasons, so promotion decisions rest on evidence rather than the loudest voice in the room. See the [structured questions guide](/docs/structured-questions-guide) to design them, and pair this with a [weighted scoring model](/docs/weighted-scoring-model-guide) when you need to rank several validated candidates.\n\n## Communicating the roadmap to stakeholders\n\nNow-Next-Later is also a stakeholder-management tool. When an executive asks \"when will X ship?\", the honest answer is often \"it's in Later — here's the evidence we'd need to see to promote it.\" That reframes the conversation from dates to confidence, and invites stakeholders into the prioritization logic instead of holding you to a guess. Pair the board with the research behind each item and the roadmap sells itself.\n\n## Common mistakes\n\n- **Turning Later into a graveyard** — if nothing ever graduates, the column is just a backlog. Review and prune it.\n- **Over-specifying Later** — detailed specs for unvalidated ideas waste effort and create false commitment.\n- **Promoting on opinion** — without evidence-based movement, the format is just three colored columns.\n- **Letting \"Now\" sprawl** — too much in progress means nothing actually gets finished.\n- **Treating it as static** — the value is in the regular, evidence-driven review.\n\n## A worked example\n\nImagine a Now-Next-Later board for a B2B analytics product. **Now** holds \"cut time-to-first-dashboard\" — in active development, scoped, tied to an activation metric. **Next** holds \"shared team workspaces,\" validated through a dozen AI interviews where teams described workarounds for collaboration but not yet fully designed. **Later** holds a loose cluster: \"Slack alerts,\" \"white-label reports,\" \"mobile view\" — ideas raised by sales and support that nobody has tested. Over the next month, continuous interviews reveal that white-label reporting is a repeated dealbreaker in lost deals, so it accumulates enough evidence to graduate into Next, while \"mobile view\" goes cold and is demoted. Nothing moved because someone argued for it; each promotion was earned by customer evidence. That is the format working as intended — a roadmap that updates itself as the research stream brings new signal in.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — score opportunities with the six question types\n- [Research-Driven Roadmap Prioritization](/docs/research-driven-roadmap-prioritization) — the prioritization methods that decide what moves\n- [Weighted Scoring Model](/docs/weighted-scoring-model-guide) — rank validated candidates objectively\n- [Kano Model](/docs/kano-model) — classify features by how they affect satisfaction\n- [Continuous Discovery & User Research](/docs/continuous-discovery-user-research) — keep the evidence stream feeding the board\n- [Impact Mapping Guide](/docs/impact-mapping-guide) — connect roadmap items to business outcomes","category":"product-management","lastModified":"2026-07-27T03:19:42.141406+00:00","metaTitle":"Now-Next-Later Roadmap: A Simpler Way to Plan Product | Koji","metaDescription":"The Now-Next-Later roadmap replaces date-based timelines with three flexible horizons. Learn how to build one, what belongs in each column, and how Koji's customer research decides what earns a slot.","keywords":["now-next-later roadmap","now next later","product roadmap framework","agile roadmap","outcome-based roadmap","roadmap horizons","prodpad roadmap","flexible product roadmap"],"aiSummary":"A Now-Next-Later roadmap organizes product work into three horizons — Now (in progress, well-defined), Next (validated, not yet scoped), and Later (unvalidated opportunities) — instead of date-based timelines, matching certainty to time. Popularized by Janna Bastow of ProdPad, it fixes date-roadmap problems: dates become promises, output is rewarded over outcomes, and learning is resisted. Frame items as outcomes not features, keep Now short, review on a cadence, and require evidence to promote items between horizons. That evidence requirement is a research problem Koji solves with always-on AI interviews: discover problems for Later, validate before Next, confirm impact after Now. Six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) make demand countable so promotion rests on evidence, not the loudest opinion."}],"pagination":{"total":1,"returned":1,"offset":0}}