Skip to content
All articles
Innovation7 min read

Why Design Thinking Fails When It Starts with Ideation

A practical guide to framing evidence, needs and assumptions before a design-thinking team generates solutions.

For: Product managers, service designers and transformation teams

A room, sticky notes and a creative prompt can produce an impressive number of ideas. It can also produce a large number of solutions to the wrong problem.

The most common failure in Design Thinking is not a lack of creativity. It is premature certainty about:

  • who has the problem;
  • when the problem occurs;
  • what makes the situation difficult;
  • which outcome matters;
  • which solution category should be used.

When these assumptions remain hidden, an ideation workshop scales them.

The hidden solution inside the problem

Consider this brief:

How might we build an AI assistant that reduces support demand?

The wording makes three decisions before the work begins:

  1. AI is the mechanism.
  2. An assistant is the product form.
  3. Reducing contacts is the right outcome.

Customer evidence might show that contacts rise because an existing workflow hides who must act next. A new assistant could add another place to check while leaving the hand-off unchanged.

A better starting frame might be:

How might we help account administrators see and complete the next required action when a support case changes owner?

This question remains specific, but it permits interface, workflow, notification, policy and service options.

Separate four things teams often mix

Before ideation, create four evidence columns.

TypeExampleWhat it is not
ObservationSix of eight administrators opened chat after a case changed teamsA general truth about all customers
InterpretationThe ownership change may reduce confidenceA verified cause
NeedKnow who must act and what completion looks likeA product feature
ConstraintRegulated messages need an audit recordA reason to stop exploring

The distinctions matter because the team should be able to challenge an interpretation without discarding an observation.

Research recent behaviour

General preference questions produce general answers:

  • “I want it to be easy.”
  • “Notifications are important.”
  • “AI could be useful.”

Ask for a recent event instead:

  • Tell me about the last time the case changed owner.
  • What did you expect to happen next?
  • What did you see?
  • What did you do?
  • At what point did you contact support?
  • What would have made you confident enough not to contact anyone?

Recent-event questions create a sequence that can be observed, compared and tested.

A useful frame has tension

Strong frames are neither vague nor narrow.

Too broad:

How might we improve customer experience?

Too narrow:

How might we add a red notification badge?

Useful:

How might we help an administrator stay confident about progress when a case crosses team boundaries?

The frame contains a person, important progress and a circumstance. It does not contain the final mechanism.

Prototype the uncertainty, not the whole product

After ideation, ask which assumption could make each concept fail.

If the uncertainty is whether people understand a new sequence, a paper flow or clickable prototype may be enough. If the uncertainty is whether a human service creates trust, a role-play or concierge pilot is more credible. If the uncertainty is economic, an interface prototype is insufficient.

Prototype fidelity should match the learning question.

Mini-case: the discharge app

A day clinic sees frequent calls after minor procedures. A manager proposes a mobile app.

Research finds:

  • patients fold the instruction sheet before the nurse finishes;
  • family caregivers are often absent;
  • medication timing and symptom uncertainty drive calls;
  • some patients have limited mobile access.

The design opportunity is not “build an app.” It is to help a patient and caregiver recognise the correct next action under stress, after leaving the clinic.

Concepts could include:

  • a personalised large-print action card;
  • a caregiver hand-off message;
  • a timed check-in;
  • a teach-back conversation;
  • a digital guide.

The app becomes one option, not the definition of the problem.

A pre-ideation checklist

Before generating ideas, confirm:

  • the person and situation are specific;
  • observations come from real behaviour or artefacts;
  • interpretations are marked as interpretations;
  • the desired progress is clear;
  • the question contains no preferred feature;
  • constraints are visible;
  • the next prototype will test a named uncertainty.

Creativity becomes more useful when the frame is evidence-based.

Practical next step

Take one feature request from your current roadmap. Remove the feature, identify the observed event behind it and rewrite the request as:

How might we help [person] achieve [progress] when [circumstance]?

Then generate at least three mechanisms before evaluating any one of them.

References

  1. Stanford d.school. “Design Thinking Bootleg.” Official resource (opens in a new tab).
  2. IDEO.org. The Field Guide to Human-Centered Design. Official resource (opens in a new tab).

Continue with the tool

Use the Design Thinking guide to frame a challenge, prototype a critical assumption and run a behavioural test.