I previously worked on a discovery where a preferred solution already had strong support before the team had properly examined the problem.

There was a clear need because the existing approach was costly and difficult to sustain, while the proposed intervention offered a route forward. There was also pressure to move quickly.

This work could easily have become an investigation into how that intervention should be delivered. Implementation and feasibility still needed to be understood, but that would not answer whether the proposed solution was right.

We widened the inquiry, looking at the surrounding service, where effort and delay were being created and whether different parts of the problem required different responses.

The recommendation was a proportionate next step rather than full adoption of the original proposal. The preferred solution remained viable, but it was no longer treated as inevitable.

The discovery had inherited a preconceived answer. Unless we made that visible, we risked validating it rather than investigating the problem.

This can happen even when the interviews, analysis and options work are rigorous. Earlier decisions may already have shaped what the team is expected to examine.

As such, discovery had begun before the discovery team had arrived.

Before the first interview

Good discovery practice already recognises that teams work within constraints.

The GOV.UK Service Manual tells teams presented with a predefined solution to question it and reframe it as a problem before starting research. It also distinguishes between hard constraints, such as primary legislation, and softer constraints that may be open to change. Discovery can also conclude that the proposed work should not continue.

The guidance mainly focuses on constraints within the problem, including legislation, technology and existing operations. Other constraints shape what the team is able to investigate.

For example, funding may already have been approved for one type of intervention or a delivery date may already have been announced.

Some of those boundaries will be reasonable. A discovery cannot investigate everything. What matters is whether the team understands the boundaries around the work. It needs to know:

What can we question?
Which evidence will carry weight?
What can that evidence change?

The question has already shaped the research

The starting question shapes the work.

“How should we replace this platform?” directs attention towards replacement options, while “How should we improve this service?” leaves more space for changes to operations, policy, process or technology.

The wording influences which users are included, which evidence is collected and which interventions appear relevant.

Tversky and Kahneman’s 1981 paper on framing showed that the way a decision problem is presented can change people’s preferences. Their experiments examined decisions under risk rather than product discovery. It is still useful here because it shows how the wording of a problem affects how people approach it.

The boundary also extends beyond the written brief. Teams learn what can be questioned through the reactions they receive as the work progresses.

Someone says a decision has already been made, but nobody can identify where it was recorded or who made it. A non-digital option is ruled out because the money was approved for technology.

Nobody needs to formally instruct the team to narrow the inquiry; it adjusts in response to what the organisation appears willing to consider. Sometimes that reflects a genuine limitation. Sometimes an assumption becomes an immovable fact because questioning it would be difficult.

Which evidence carries weight?

Teams can inherit assumptions about what counts as credible evidence.

User research may carry weight until it challenges an executive commitment, while operational experience may be welcomed but treated as anecdotal. The difficulty comes when everybody knows, but nobody says, which evidence will win.

Teams may be asked to gather evidence without knowing what would be considered sufficient, who will interpret it or how conflicting findings will be weighed.

The Magenta Book, the UK government’s guidance on evaluation, recommends identifying learning goals, intended users and decision points early. Work can then be designed around the questions that need answering and when the evidence will be required.

The Magenta Book covers formal evaluation, although its guidance on connecting evidence to a known decision also applies here.

Before gathering evidence, the team should know who will interpret it and what they are likely to accept as persuasive. Without that clarity, research can be sound and still have little effect.

When findings meet authority

In an open discovery, the findings might redefine the problem, challenge the proposed intervention or show that the work should stop. In a more constrained piece of work, the technology, budget and deadline may already be fixed. The team can improve the implementation but cannot change the direction.

That work may still be useful, but the team should understand that it is improving an existing direction rather than openly investigating the problem.

The GOV.UK guidance on agile governance says teams should understand which decisions they have the authority to make, where that authority ends and who is accountable when a decision sits elsewhere. That clarity is often missing at the start.

A team can spend weeks gathering evidence, only to find at the final playback that nobody in the room can act on it. The sponsor may agree with the findings but have no route to revisit the funding. The researchers may have done good work but were simply given less room than they realised.

The organisation still appears to be choosing, even though the significant choices have already closed.

Boundaries are necessary

Discovery needs boundaries.

Herbert Simon’s work on organisational decision-making challenged the idea that people make choices with complete information and unlimited capacity. People within organisations make decisions with limited knowledge, attention and ability to understand every possible consequence.

A team still needs a clear purpose, a timescale and an idea of the decision it is there to support, but trouble starts when an old organisational choice is mistaken for an immovable fact. For example, a technology selected years ago can become the only viable platform, even when the reasons for selecting it are no longer understood.

Research into organisational path dependence describes how earlier decisions can narrow future options as investment, routines and expectations build around the existing direction. This can eventually create a form of lock-in and discovery can reinforce that pattern without anyone intending it to.

If each investigation accepts the same inherited choices, it only looks for improvements within the existing path. Those improvements justify further investment. By the time the next discovery begins, alternatives are even harder to consider.

Fixed, negotiable or assumed

Before research gathers pace, teams should classify significant constraints as fixed, negotiable or assumed.

Fixed constraints are outside the team’s current authority or cannot realistically change within the relevant period. Primary legislation may fall into this category.
Negotiable constraints are real but may change through evidence, agreement or escalation. Scope and internal deadlines often fit here.
Assumed constraints are treated as fixed without a clear current rationale, identifiable decision or accountable owner.

Assumed constraints are usually the ones worth examining first.

Teams should not challenge every constraint. Legislation, safety requirements or genuine policy decisions should not be treated as organisational failure. The value comes from spotting boundaries that nobody has consciously chosen.

Before collecting evidence, establish:

What has already been decided?
Which decisions can be revisited?
What evidence could justify that?
Who has the authority to act?

I would also ask:

Which finding would the organisation struggle to accept?

The answer may tell the team more about the discovery’s limits than the formal brief.

A more honest discovery

In the work I described, the useful intervention was widening the question rather than using a novel research method.

The proposed solution remained a credible approach, but it became one possible response rather than the assumed destination. That gave us room to examine the surrounding service, compare alternatives and recommend a proportionate next step.

We did not remove the boundaries around the work. We understood them better, which meant we could be clearer about what the discovery was able to examine and what its findings might change.

That is a better foundation than quietly inheriting a preconceived answer.

James Lilwall || Product Thinking

The Hidden Architecture of Discovery