AI cannot fix bad strategy

Most of the problems that show up in digital products aren't caused by poor execution. They sit much earlier: the wrong problem has been prioritised, the journey doesn't reflect how users actually behave, the structure doesn't support the task it's meant to enable. When that happens, no amount of optimisationglossary×OptimisationOptimisation is the process of improving a product or journey to increase performance, usability, or conversion.Open glossary term→ will fix it. AI just makes it happen faster.

Reaching for it is also a way of avoiding a harder conversation. Rewriting the content is something a team can start on Monday. Establishing that the journey should not exist in its current form requires reopening a decision somebody senior already signed off, and that has no obvious first step.

The speed makes it more tempting still. Optimisationglossary×OptimisationOptimisation is the process of improving a product or journey to increase performance, usability, or conversion.Open glossary term→ produces visible movement within days, which buys time and looks like a responseglossary×ResponseA response is the data or result returned by a server after receiving a request.Open glossary term→. The underlying problem is unchanged, but the pressure to address it has been released.

Content can be improved, but if it's answering the wrong questions, it will still miss the mark. Journeys can be streamlined, but if they're built on incorrect assumptions, they will still create frictionglossary×FrictionFriction refers to anything that slows users down or makes it harder for them to complete a task. It can be caused by poor design, unnecessary steps, unclear messaging, or technical issues.Open glossary term→. Interfacesglossary×InterfaceAn interface is the point of interaction between a user and a system, where inputs are made and outputs are received. It can be visual, physical, or conversational.Open glossary term→ can be refined, but if the underlying flow is flawed, the experience will still break down. Everything becomes more efficient. But not more effective.

The distinction shows up in what the metrics do. Efficiencyglossary×EfficiencyEfficiency measures how quickly and easily users can complete tasks once they are familiar with a system.Open glossary term→ work moves things like time on taskglossary×Time on TaskTime on task measures how long it takes users to complete a specific task.Open glossary term→ and completion rate within the existing journey. Effectiveness work changes whether that journey was the right one, and it usually shows up somewhere else entirely, in retentionglossary×RetentionRetention measures how well a product keeps users over time by continuing to deliver value. It is a key indicator of product success and long-term viability.Open glossary term→, in support volume, or in revenue per visitor.

AI is very good at making things look better. It can produce cleaner copy, more consistent structures, and more polished outputs. From the outside, it can look like the product has improved significantly. But underneath that, the same issues often remain. The strategyglossary×StrategyStrategy is a high-level plan that defines long-term goals and the approach to achieving them.Open glossary term→ hasn't changed. Metrics may shift slightly, engagementglossary×EngagementEngagement refers to how users interact with a product, content, or experience, including actions like clicks, time spent, and interactions.Open glossary term→ may improve in places, and there's a sense that things are moving in the right direction. But the core problem is still there, just hidden behind better execution. That only holds for so long, because users don't respond to polish. They respond to relevance.

This is worth watching for in reporting. If output volume is rising, quality scores are rising, and the commercial numbers are flat, that combination is not a measurement problem waiting to resolve itself. It usually means the work is improving something that was never what stood in the way.

Before AI is introduced, before optimisationglossary×OptimisationOptimisation is the process of improving a product or journey to increase performance, usability, or conversion.Open glossary term→ begins, before anything is generated, the direction needs to be clear. What problem is being solved? Why does it matter? How does the product support that in a way that makes sense for the user and the business? Without that clarityglossary×ClarityClarity is how easily users can understand what is happening and what they need to do.Open glossary term→, AI has nothing solid to work with.

Used properly, once the strategyglossary×StrategyStrategy is a high-level plan that defines long-term goals and the approach to achieving them.Open glossary term→ is right, AI can be incredibly effective. It can scale content, accelerate iterationglossary×IterationIteration is the process of repeatedly improving a product through cycles of testing, feedback, and refinement.Open glossary term→, and support refinementglossary×RefinementRefinement is the ongoing work of clarifying, breaking down, and sizing backlog items before they enter development. It is where ambiguity gets resolved while resolving it is still cheap.Open glossary term→ in a way that would otherwise take significantly more time. When the direction is clear, AI becomes a multiplier. When the direction is wrong, it simply reinforces the problem, making it harder to recognise and more expensive to fix later. AI is not a safety net. It will not step in and identify that the wrong problem is being solved. That still requires human judgement. The temptation is to use AI to improve what exists. The reality is that sometimes what exists needs to be challenged first.

A useful gate before any of it starts: name the decision the user is trying to make, and what they need in order to make it confidently. If that cannot be answered without referring to the current design, the strategyglossary×StrategyStrategy is a high-level plan that defines long-term goals and the approach to achieving them.Open glossary term→ work has not been done and no amount of generation will substitute for it.