How is this different from general market research?

The subject is an AI product, so the research has to handle questions a general study does not: what the system can actually do today, how much review effort its output requires, and whether the buyer's objection is about capability, accountability or both. The method is ordinary primary research applied to that subject.

Can you work with an idea that is still vague?

Yes, and discovery is designed for exactly that. The starting material is usually several competing hypotheses with uneven evidence. The work separates an observed complaint from an internally generated idea, and produces a ranked register rather than a verdict on a single concept.

Will you tell us whether to build the product?

No. The work produces evidence, thresholds and an explicit account of what remains unknown. The investment decision stays with the people who carry its consequence. Where the evidence points clearly in one direction, the report says so plainly, with the conditions under which it would not hold.

What access would the research need?

Introductions or a recruitment route to practitioners and budget holders, the current hypotheses and their origin, and any existing usage, support or pipeline data. Participant time is usually the binding constraint, and scoping should treat it as one rather than assume it.

Can you size the market?

Not as a forecast. Research can establish who the buyer is, what they use today, what they pay for adjacent work and how they are reached. Turning that into a market size requires assumptions that belong to your team, and presenting the result as a projection would overstate what the evidence supports.

How does this connect to the AI work you do elsewhere?

Research often reveals that the blocking question is technical: whether the system is accurate enough, whether its output can be reviewed affordably, or whether a security review will stop the sale. Those lead to evaluation, security and governance engagements, which are scoped separately.

Capability family

AI product-market fit and market research

Find out what the market will actually pay to have solved.

AI product-market fit research connects a claimed opportunity to the people who own the work, the workflow it would change and the evidence that would confirm or refute the idea. Treat the proposition as a hypothesis until the research answers it.

Book a 30 minute readiness call

A proposition is a hypothesis until someone tests it

Illustrative scenario

Claimed problem
The difficulty the product assumes someone already has.
The people who own it
Practitioners who do the work and the budget holder who buys.
Current alternative
Manual effort, an incumbent tool, or doing nothing.
Decision threshold
The result that would change the plan, agreed in advance.

Conceptual model of the discipline; the engagement boundary requires separate agreement.

Discipline model

Evidence about the problem outranks enthusiasm for the solution

The relationships that define the discipline

Problem: Severity, frequency and who actually feels the cost.
Buyer: Who signs, who blocks and what each needs to believe.
Workflow: How the task runs today, including its informal steps.
Threshold: What result would confirm or refute the plan.

Use these connected areas to identify which research question matters now.

AI makes many things possible, which is exactly why the discipline matters. An idea that describes a new capability is not the same as a problem someone is paying to have solved. The work connects four things that teams usually examine separately: the severity of the problem, the person with the budget, the workflow as it really runs, and the threshold that would tell the team it was wrong.

Illustrative situation

Enthusiasm from people who cannot buy

A constructed situation and the question it raises

Illustrative scenario

The signal
Users are enthusiastic in every conversation.
The gap
None of them holds budget or influences the purchase.
The question
What would the person who signs need to believe?

Fictional example; no engagement, executed test or measured outcome is implied.

Research that only reaches users produces a reliable and misleading result. The people doing the work describe the difficulty accurately, and they are not the people who decide. A useful research design separates the two, records which findings came from which, and treats a buyer conversation that has not happened as an open question rather than an encouraging silence.

Working approach

State the threshold before the data arrives

From belief to a result that could change the plan

Surface the belief
Write down what the team currently assumes, and why.
Set the threshold
Define the result that supports it and the result that refutes it.
Observe
Collect evidence under a design fixed before collection began.

Proposed sequence; research access and participant time are agreed separately.

A threshold set after the data arrives is not a threshold. The approach makes each belief refutable first, then designs the observation: which population, which period, which instrument, and what would confound the measure. That discipline is what separates research from reassurance, and it is usually the part a team under time pressure skips.

Working artefact

A register the team can argue with

What a research output carries with it

Illustrative working format

Evidence basis: Who was asked, what was observed and what was not.
Interpretation: What the evidence supports, and where it stops.
Open question: What remains unanswered and how it could be answered.

Illustrative structure, not a completed engagement or an observed result.

The outputs are opportunity registers, workflow models, positioning statements, pilot designs and readiness reviews. What makes them useful is the same in each case: the evidence basis is visible, the interpretation is separable from the observation, and the unanswered questions are named rather than smoothed over. A team should be able to disagree with a conclusion by pointing at its evidence.

Scope and next decision

Research informs a decision; it does not make one

What this work does and does not establish

Included: Primary research, workflow study, proposition and pilot design.
Excluded: Market sizing as forecast, fundraising narrative, campaign delivery.
Retained: Your team keeps the investment and product decisions.

Bring the proposition, the decision you face and the evidence you already hold.

Research reports what a defined sample said and did. It does not establish market size, predict adoption or guarantee that a product will succeed. Where a finding rests on a small sample or a constrained recruitment route, the report says so. The value is a decision made with the uncertainty visible, rather than a number that hides it.

Explore the discipline

Specialist areas within this capability

Choose a question to explore its methods, boundaries and indicative outputs.

Search this capability family