Now / Next / Later: Honest Time Horizons Without the Chaos

A Now/Next/Later roadmap replaces calendar dates with three confidence-based horizons — Now (committed, scoped work), Next (a validated problem with a directional solution), and Later (a raw opportunity, no commitment). Precision decreases as the horizon extends, which is the whole point: it tells the truth about what you actually know, instead of dressing an eighteen-month guess up as a promise.

Quick answer: Now is committed and scoped, shipping this cycle. Next is a validated problem with a directional fix, not yet detailed. Later is a raw opportunity with no delivery promise — and it only moves up when evidence clears a bar you set in advance.

Why Precision Should Shrink as the Horizon Grows

A date-based roadmap assigns the same false precision to everything on it, whether the item ships in two weeks or gets attempted in two years. A Now/Next/Later roadmap sizes your confidence to match reality instead: tight and specific up close, deliberately fuzzy far out. That single change removes most of the trust-eroding surprises that turn roadmaps into political documents rather than planning tools.

Software estimation research backs this up directly. Steve McConnell's Cone of Uncertainty, from Software Estimation: Demystifying the Black Art, shows that early-stage estimates commonly land within roughly a 4x range in either direction, and that the band only narrows once real scoping work happens. A roadmap date set a year out isn't a forecast — it's a guess wearing a forecast's clothes, and everyone downstream treats it as more solid than it is.

Now/Next/Later doesn't remove uncertainty. It labels the uncertainty correctly instead of hiding it behind a date picker, which is a very different promise to make to a stakeholder.

The gap in precision is the entire mechanism, not a side effect. A Now item reads like "ship inline filters for the search results page this cycle." A Later item reads like "search feels slow for power users with large workspaces." One is an instruction; the other is a hypothesis worth watching. Collapsing that distinction — writing Later items as if they were Now items with a fuzzier date — is how roadmaps quietly turn back into wish lists dressed up in new column headers.

This isn't only a startup habit, either. GitLab publishes its own product direction pages in the open, and they use a similar logic: broad themes for bets several quarters out, sharp specifics for the current milestone. Seeing the same precision gradient hold up on a roadmap read by thousands of external users is a reasonable signal that it scales past a single team's internal planning doc.

Notice the last row on the right: horizon roadmaps have their own failure mode too — a graveyard Later list — and it deserves the same discipline you'd apply to a slipping date. We'll come back to that. If you're weighing roadmap formats more broadly before committing to this one, our roadmapping complete guide walks through when a date-based, outcome-based, or horizon-based approach actually fits a given team's stage and stakeholder pressure.

Building the Three Buckets: Entry and Exit Criteria

Each horizon needs its own written definition of "done enough to be here" and "done enough to leave." Skip that step and Now/Next/Later degrades into three renamed backlogs carrying the same fuzziness you were trying to escape. The table below is a baseline — tune the thresholds to your team's cadence, not the structure itself.

Now: Committed and Scoped

Now is the only horizon that should ever look like a traditional sprint board. Items here are scoped to the story or epic level, dependencies are cleared, and the effort input feeding RICE is a real estimate rather than a guess. Nothing sits in Now without a named owner accountable for shipping it or explicitly killing it.

Because Now items are close enough to ship, this is also where go-to-market planning starts running in parallel, not after the fact. Our GTM launch complete guide covers what needs to be true — positioning, enablement, support readiness — before something in Now actually goes live.

Next: Validated Problem, Directional Solution

Next holds problems you've already validated, paired with a rough idea of how you'd solve them — not a committed spec. The validation has to come from evidence: a JTBD interview pattern, a support-ticket cluster, or usage data, not a single stakeholder's conviction that this matters. Sizing is rough on purpose; a full scoping pass only happens once an item is promoted.

• The problem statement is specific enough that two PMs would describe it the same way.
• A solution direction exists, but the scope is intentionally still negotiable.
• A RICE estimate exists, built from real inputs, not placeholders.
• No date is attached, internally or externally — that's what Now is for.

Later: Opportunity, Not Obligation

Later should read like a list of themes and opportunities, not a list of features waiting their turn. If an item in Later already reads like a spec, you've smuggled Next-level precision into a bucket that's supposed to stay deliberately vague. The only real obligation attached to a Later item is to re-examine it on a cadence — never to deliver it.

Keeping Later at the opportunity level is easier when your team is already comfortable writing roadmap items as outcomes rather than features. Our piece on outcome-based roadmaps vs. feature lists covers the phrasing shift that makes this natural instead of forced.

The Promotion Rule: What Earns a Jump from Later to Next

An item earns promotion from Later to Next only when it clears three tests at once: a validated problem, a rough sizing estimate, and a RICE score that beats your current lowest-ranked Next item. Skip any one test and you're promoting on vibes — the exact failure mode a horizon roadmap is supposed to prevent.

1. Validated problem. Evidence has to come from customer interviews, support-ticket clustering, sales-loss reasons, or usage data — not one loud stakeholder. If the signal is a growth or retention metric, cross-check the pattern against what's covered in our growth and retention complete guide before you trust it as a promotion trigger.
2. Rough sizing spike. A technical lead spends an hour, not a sprint, roughing out effort so the RICE "effort" input is a real number instead of a guess wearing a number's clothes.
3. RICE beats the floor. The item's Reach × Impact × Confidence ÷ Effort score has to outrank your weakest current Next item — otherwise you're expanding Next, not re-ranking it.

If the item touches pricing or packaging, add a fourth gate before it moves: run it past whoever owns pricing strategy. A mispriced promotion can quietly undo the RICE math entirely, and our pricing and monetization complete guide covers how those interactions typically break.

Gut feel still has a role in this rule, but only as a tiebreaker among items that already cleared all three gates — never as a fourth path around them. The moment "I just think this matters" becomes sufficient on its own, Next starts filling with opinions instead of evidence, and the whole format loses its honesty advantage over a date-based roadmap.

Write the three gates down somewhere every stakeholder can see, not just in a PM's head. A rule that only lives as tribal knowledge gets re-litigated every time someone's favorite idea stalls in Later, and you end up defending a decision from memory instead of pointing at a criterion everyone already agreed to.

The Graveyard Anti-Pattern: When Later Stops Being a Horizon

Later becomes a graveyard when it's treated as a place ideas get filed rather than a bucket things get judged in — no size limit, no expiration date, and no owner accountable for pruning it. The fix isn't deleting Later; it's giving it the same discipline Now already has: a cap, a review cadence, and an explicit kill step.

The bloat happens for predictable reasons:

• Adding is free. Dropping a line into Later costs nothing, so it becomes the path of least resistance for every idea nobody wants to fully reject.
• Nobody owns removal. Now has a delivery lead who feels the pain of an unshipped item; Later often has no one whose job is to say no.
• Killing feels political. Archiving someone's idea reads as a rejection of them, so teams quietly let items rot instead of closing them out loud.

Basecamp's Shape Up team, led by Ryan Singer, takes an even more aggressive stance: they don't maintain a persistent backlog at all. Ideas get reconsidered from scratch at each six-week betting table, and anything not actively re-chosen simply isn't sitting on a list somewhere accumulating guilt. Few teams need to go that far, but the instinct underneath it is correct — an idea nobody re-chooses is not a commitment, it's clutter.

A workable middle ground for most teams:

• Cap Later at roughly twice the size of Next, adjusted for team capacity — a guideline to tune, not a law.
• Require a one-line kill reason before anything is archived. "Not now" isn't a reason; "conflicts with this quarter's focus on retention" is.
• Put a named review owner and a fixed cadence — quarterly is common — on the calendar, so Later gets triaged on purpose instead of by accident.

Janna Bastow, who popularized the Now/Next/Later format at ProdPad, has been explicit that it only works if all three columns get revisited together. A Later list that silently accumulates while Now and Next get all the attention defeats the entire design — you've just rebuilt the backlog you were trying to escape, with better labels on it.

Running the Cadence: Who Owns the Ritual and When

Now/Next/Later fails without a repeating ritual that moves items and enforces the criteria above — it's a discipline, not a document you fill out once a quarter and forget. Different horizons need different review frequencies, and someone has to be accountable for convening each one.

• Now gets reviewed inside your existing delivery cadence — standup, sprint review — because it's operational, not strategic.
• Next gets revisited every two to four weeks, ideally alongside your prioritization ritual, so RICE scores stay current rather than stale.
• Later gets triaged quarterly, tied to planning cycles, with the promotion and kill rules from above applied on purpose rather than left to whoever remembers.

At a company level, this cadence is usually a product operations responsibility: deciding who convenes each review, how decisions get logged, and how the roadmap gets communicated externally without quietly turning into a promise again. Our product operations complete guide covers how to build that rhythm without bolting on another meeting nobody wants to attend.

Keeping the Ranking Honest as Priorities Shift

The hardest part of Now/Next/Later isn't drawing three columns — it's re-ranking them honestly every time new evidence shows up, instead of leaving last quarter's order untouched because re-sorting a spreadsheet is tedious. Static tools make that re-ranking feel like extra work, which is exactly why stale roadmaps happen even inside teams using the right framework.

This is the specific friction Prodinja's Prioritization workspace is built to remove: it can re-rank items live as you drag them across Now, Next, and Later, so each horizon reflects current RICE inputs rather than whatever position an item happened to land in three sprints ago. It doesn't replace the judgment calls above — the entry criteria, the promotion gates, the kill decisions are still yours to make. It just removes the excuse teams reach for when the list is static and re-sorting it feels like too much friction to bother with.

Key Takeaways

• Precision should shrink with distance. Now is specific and scoped; Later is deliberately vague — that gap is honest, not sloppy.
• Now means committed and scoped, with a named owner and an exit of "shipped" or "explicitly cut," never "quietly slipped."
• Next means a validated problem with a directional fix, sized roughly, with no date attached internally or externally.
• Later means an opportunity, not an obligation — it should read like themes and problems, not a feature backlog wearing a new label.
• Promotion needs evidence to clear three gates at once: a validated problem, a rough sizing spike, and a RICE score that beats the current Next floor.
• Later needs a cap, a cadence, and a kill step, or it quietly becomes the graveyard version of the backlog you were trying to leave behind.
• The ritual matters more than the columns. Review each horizon on its own rhythm, or the format decays into three static lists within a quarter.

Frequently Asked Questions

Is a Now/Next/Later roadmap just three swimlanes on a Kanban board?

No — a Kanban board tracks workflow state (to-do, doing, done) for work that's already committed, while Now/Next/Later tracks confidence about whether something should be committed at all. You can run Kanban inside your Now column for execution, but Next and Later are about evidence and validation, not task status.

How many items should be in each horizon?

Now should hold roughly what your team can actually ship in the current cycle — err smaller, since anything left over just becomes silent slippage. Next typically runs a bit larger than Now since not every validated problem gets promoted immediately, and Later should be capped, not left to grow indefinitely, at roughly twice the size of Next as a starting guideline.

What do you tell a stakeholder who demands a ship date for a Later item?

Tell them the truth: it's an opportunity you're tracking, not a commitment, and it hasn't cleared the evidence gates that would earn it a spot in Next, let alone Now. Offer to show them exactly what would need to be true — the validated-problem, sizing, and RICE-ranking criteria — for it to move up, which is a more credible answer than a fabricated date.

Does Now/Next/Later still work if leadership insists on OKRs with dates?

Yes — sync your Now horizon to the current OKR cycle, since that's the work you're genuinely committing to inside a dated period. Leave Next and Later undated even under an OKR structure; they represent problems and directions you're tracking toward future cycles, not commitments you're making inside this one.

What earns a jump from Next back down to Later?

An item drops back to Later when new evidence contradicts the original validation — usage data reverses, a bigger opportunity outranks it on RICE, or the sizing spike reveals effort far beyond what the score assumed. Demotion should get the same one-line-reason discipline as a Later kill, so the move reads as a decision, not a disappearance.