You probably spent time in a room last year mapping out your product’s next 3 to 5 years. Offsite somewhere, maybe. A Notion doc the length of a quarterly report, a Gantt chart that stretched to 2028. By month 4, a core assumption was already wrong.

This keeps happening because the planning model is borrowed from an industry where it made sense.

Where the 5-year roadmap came from

Roadmaps didn’t originate in software. They’re a construction artifact, useful when changing direction in year 2 means scrapping a half-built factory or renegotiating supplier contracts that took 18 months to sign.

Software doesn’t have those constraints. The marginal cost of pivoting is low. But startups imported the format anyway, mostly because investors asked for it. The roadmap became a confidence signal: we know where we’re going.

The vision part holds up fine. Knowing you want to be the default tool for SME procurement, or the fastest API for payment routing, is durable. The specific build sequence underneath that vision is usually fiction after about 6 months. But the fiction looks like a plan, so it stays.

What rolling 90-day themes actually means

A theme is a 90-day strategic focus area. Something like “reduce friction for first-time buyers” or “prove enterprise viability.” Leadership picks the direction. Product and engineering decide what to build under it.

Contrast that with a typical roadmap entry: “Build onboarding v3 with 4 tooltips and a progress bar by Q2.” One ages well. The other is usually wrong before it ships.

The full loop goes like this: pick a theme, ship work under it for 90 days, run a short retrospective, pick the next theme based on what you actually learned. The retrospective is where the model earns its value. If you can’t honestly say what didn’t work, you’re just doing quarterly roadmap planning with extra steps.

Why 90 days

Long enough to ship something meaningful and see how it performs. Short enough to correct course before you’ve committed a full engineering team to something you can’t recover from.

Quarterly cycles also align with how most finance teams and boards already think. You’re adopting a new planning cadence, but the reporting rhythm stays the same.

Themes vs. OKRs

OKRs (Objectives and Key Results) ask: what will we achieve, and how do we measure it? Themes ask: what problem are we focused on?

The difference matters when you’re 6 weeks into a quarter and discover your key result was the wrong metric. With OKRs, that discovery feels like failure. With a theme, it’s information. You were pointing at first-time buyer friction, here’s what you learned, here’s what the next 90 days should address. No failure narrative required.

What “evidence” looks like at the end of 90 days

You won’t have statistical significance after a single quarter.

Did conversion at the problematic step improve? Did support tickets about the onboarding flow drop? That’s enough to make a smarter bet on what to point at next.

The tradeoff you’re accepting

Your board and investors will ask for the roadmap. “We run on 90-day themes” is a harder conversation than pulling up a Gantt chart that goes to 2028. You’ll need to articulate what’s directionally true over 2 to 3 years without committing to specifics that’ll be wrong before you get there.

The other cost is team judgment. A rigid roadmap removes daily decision-making from the team. Everyone knows what’s next. Rolling themes require your engineers and PMs to make more calls about what’s worth building under the current direction. If your team doesn’t have that yet, the model produces noise.

Be honest about where you are. A well-run 5-year roadmap beats a badly-run theme process.

When this model works and when it doesn’t

Rolling themes make sense when your market is moving fast. In AI tooling or developer infrastructure, a competitor or a model release can make a roadmap milestone irrelevant before you’ve staffed it.

It tends to break down in regulated industries. Medical devices, fintech with heavy compliance requirements: those contexts genuinely need longer planning horizons because the regulatory calendar controls your timeline more than the market does.

And it requires honest retrospectives. Teams that run themes but can’t say “that didn’t work” at the end of 90 days end up with the same lock-in as a roadmap, just with quarterly decoration.

What to do next

Start with one theme. Pick a direction for the next quarter, stated as a problem to solve. Brief the team on what you’d consider evidence of progress at the end of 90 days. Let them figure out what to build.

At the end of the quarter, run a 2-hour retrospective: what worked, what should stop.

Most teams that try this once don’t go back. The ones who do usually find they had an investor communication problem that needed solving separately from the planning question.

If this kind of product thinking is what you want behind your next build, let’s talk.