by Format-3

Share article

The traditional product roadmap is not struggling. It is finished. Not because agile evangelists finally won the argument, but because the environment it was designed for no longer exists.

When AI capabilities evolve fast enough that a feature finalised six months ago risks being obsolete by launch, a document that locks in twelve months of delivery commitments is not a plan. It is a liability.

Pendo’s analysis of hundreds of software subscriptions found that around 80% of shipped features are rarely or never used. Those features were not built at random. They were prioritised through exactly the kind of roadmap-driven process most organisations still defend. The roadmap did not fail because teams lacked discipline. It failed because the signals it was built on were already fossilised by the time anyone shipped anything.

Why the traditional product roadmap is obsolete in an AI-driven world

Fixed delivery plans carry three structural flaws that AI-driven markets have turned from inconveniences into genuine liabilities.

• The re-plan tax. Everyone in the room knows the roadmap is wrong. Changing it requires executive sign-off, stakeholder communications, and a cascade of revised dependencies. So teams execute against a plan they know is wrong because the cost of changing it exceeds the cost of being wrong.
• Fossilised signals. The roadmap you publish today reflects customer conversations from last month and competitive data from last quarter. By the time it reaches engineering, it is already a fossil.
• The illusion of certainty. A confident twelve-month roadmap does not reflect superior foresight. It rewards what one practitioner calls “conviction theatre”: the PM who performs certainty most convincingly wins the most trust, regardless of whether they are correct.

The American Planning Association’s 2026 work on futures literacy frames this precisely: the future will not be shaped by certainty. It requires curiosity, and the courage, as Erich Fromm put it, to let go of certainties entirely.

Why fixed delivery plans fall short and what adaptive decision systems offer instead

The deeper problem is not the roadmap document itself. It is the operating model the roadmap enforces: output-focused, feature-contracted, and structurally resistant to learning.

Statistic callout: Around 80% of software features shipped are rarely or never used. Teams built them because a roadmap said to, not because evidence demanded it.

Adaptive decision systems replace that operating model with three interlocking shifts.

• Problem maps over feature contracts. Instead of committing to a feature list twelve months out, teams define the highest-value problems the business needs to solve, ranked by likely impact on revenue, margin, or customer outcomes. The roadmap told teams what to build. A problem map tells them what to fix, and trusts them to determine the best path within clear constraints.
• Bet portfolios updated weekly. A one-page live bet portfolio, refreshed every Monday, lists active bets with their hypothesis, the signal being watched, the kill condition, and the next decision date. It also publishes recently killed bets and what the team learned. That transparency builds more institutional trust than ten successful launches quietly shipped against a frozen roadmap.
• AI accelerator micro-squads. Rather than large cross-functional programmes with fixed scope, micro-squads of one PM, one designer, and one engineer are assigned a specific pain point and given full autonomy to prototype, test, and either scale or kill a solution within days. No backlog. No spec. A problem and a deadline.

The executive role transforms in parallel. Leadership defines measurable outcomes and guardrails around security, compliance, and budget, then grants squads autonomy over solution design. Shifting from output KPIs to outcome-based OKRs is not a measurement preference. It is what makes the whole model function. When AI can generate outputs almost instantly, tracking outputs tells leaders progressively less about whether anything of value is actually being created.

Funding follows the same logic. Traditional planning funds projects with fixed scope and timelines. A better model funds a small team against a clearly defined problem, gives it a limited period to show traction, and re-ups only if outcomes are moving. That keeps accountability high while preserving the flexibility that static project plans structurally cannot offer.

Practical recommendations for building an adaptive product organisation

The transition is not primarily a technical challenge. It is an emotional and cultural one, and the practical steps reflect that.

Replace one output metric with an outcome metric this quarter. Pick the most-watched dashboard in your organisation and ask what business outcome it actually predicts. If the answer is “none,” replace it. Shipping velocity means nothing when AI makes shipping trivially easy.

Write a one-page problem map for your team. List your top five problems, the metric each affects, and the squad that owns experimenting on it this quarter. Keep it visible. Update it when the evidence changes, not when the calendar says to.

Introduce a start / stop / continue cadence. A quarterly review that forces the stop conversation is often the most valuable planning ritual an organisation can adopt. When nothing ever stops, the team is running on autopilot rather than strategy.

Measure token ROI as an operating discipline. As AI adoption scales, token costs become one of the most volatile line items in the operating budget. Teams that monitor the value generated per unit of AI usage can make rational investment decisions rather than discovering waste in the quarterly finance review.

Fund teams, not projects. Carve out a micro-squad, assign it a single friction point, and isolate it from the standard ticketing queue. Give it a quarter. Measure outcomes, not deliverables.

How to manage the cultural shift away from fixed roadmaps

The hardest part of this transition is emotional. A roadmap that claims certainty feels safer than a portfolio that admits bets. That feeling is the problem.

Organisations that have made this shift successfully tend to do one thing differently: they make uncertainty visible rather than performing confidence. Publishing a bet portfolio, including the recently killed section, signals to every stakeholder that the organisation is learning rather than pretending. Hiding killed projects, by contrast, creates the kind of institutional distrust that no successful launch can repair.

The futures literacy framing from the American Planning Association applies directly here. Imagination is a skill that combines creativity and facts. Building it into planning culture means rewarding teams for learning fast, not for predicting correctly. That is a different performance standard, and it requires explicit leadership modelling. If the CDO still presents a Gantt chart at the board meeting, the organisation will keep building Gantt charts regardless of what the operating model says.

For product leaders navigating this shift, Format-3’s work on adaptive product strategy and the broader question of why most digital products should not exist offers a grounding perspective on where to focus effort.

Organisations that have already abandoned traditional roadmaps

The shift is not theoretical. Forward-thinking enterprise IT organisations are already deploying micro-squads in place of programme teams, assigning them specific internal friction points and measuring outcomes within days rather than quarters. The pattern is consistent: a small team, a defined problem, full autonomy within guardrails, and a kill condition written before the work begins.

The product management community has documented the same pattern independently. Practitioners who replaced their roadmaps with live bet portfolios report that within a quarter, the portfolio feels like the real plan and the roadmap feels like a public relations document. The evidence accumulates quickly once you run both in parallel.

Which tools enable continuous discovery and real-time experimentation

The tooling question matters less than the operating model, but the right tools make the model sustainable. Continuous discovery requires infrastructure for running experiments, capturing signals, and making decisions at the speed the market now demands.

Teams building adaptive product workflows in 2026 typically combine lightweight documentation tools such as Notion or Linear for live bet portfolios, product analytics platforms for signal monitoring, and AI-assisted prototyping environments for rapid hypothesis testing. The discipline of applies equally to product teams: the goal is generating fresh evidence, not recycling existing assumptions.

The death of the roadmap, as Adam Goodman put it, is the death of bad planning. What replaces it is not chaos. It is honesty about uncertainty, combined with the organisational discipline to act on what you are actually learning rather than what you committed to believing last quarter.

Key takeaways

Adaptive product organisations outperform roadmap-driven ones by replacing fixed delivery commitments with evidence-based bets, outcome metrics, and funded teams empowered to learn fast.

Roadmaps fail structurally | Details: Around 80% of shipped features go unused because roadmaps are built on fossilised signals, not live customer evidence.

Replace feature contracts with problem maps | Details: Define the top problems by business impact and let squads determine the solution within clear guardrails.

Publish your bets, including the kills | Details: A weekly bet portfolio with a “recently killed” section builds more trust than a confident roadmap that never admits error.

Fund teams, not projects | Details: Assign a micro-squad to a single friction point for one quarter and measure outcomes, not deliverables.

The shift is cultural before it is technical | Details: Leadership must model uncertainty openly; if executives still present Gantt charts, the organisation will keep building them.

By Format-3

Press & Media

More thoughts

Thought leadership creates value, builds knowledge and takes a stand, bridging the gap between traditional and digital platforms