The 4 types of healthcare market research

Exploratory research

Exploratory research is what you run when you cannot yet write a sensible answer option. It is deliberately open: expert calls, in-depth interviews, small focus groups and observational work, with a guide that allows the respondent to reframe the question. In healthcare it is the stage that surfaces unmet need, the real sequence of a treatment decision, the vocabulary clinicians actually use, and the constraint nobody in the brief had thought of.

Its output is hypotheses, not estimates. A useful exploratory readout gives you a decision journey, a shortlist of drivers and barriers worth measuring, and the language to measure them in. Its characteristic failure is being quoted as though it were a count — 'six of ten interviewees said' is a sentence about ten people, not about a market.

Descriptive research

Descriptive research measures what is happening among whom. In healthcare that means awareness and perception trackers, prescribing and treatment-pattern studies, chart or case reviews, patient-journey quantification, and segmentation. It needs a sample large enough to support the cuts you will actually want — by specialty, practice setting, region, patient volume or payer mix — which is usually the thing under-specified in a brief.

The discipline of descriptive work is repeatability. If you intend to re-measure after launch, the questionnaire, sample definition and fielding window have to be stable enough that a later difference means a real change rather than a methodology change. Decide at wave one whether this is a snapshot or the first point on a trend line, because retrofitting a tracker is expensive.

Causal research

Causal research isolates cause and effect by designing a controlled comparison. Respondents are shown systematically varied alternatives, and the analysis infers which attributes moved the choice. The healthcare workhorses are discrete choice and conjoint analysis for product and formulary trade-offs, MaxDiff for prioritising claims or attributes, monadic and sequential message testing, and Van Westendorp or Gabor-Granger for price sensitivity.

These designs are the only ones that answer 'what should we change?' with a defensible number, which is why they are worth the extra sample and the extra design time. They are also the most sensitive to earlier stages: a conjoint built on an attribute list invented in a workshop will produce precise weights for the wrong attributes. Build the attribute list from exploratory fieldwork, and pre-test the exercise before you field it.

Predictive research

Predictive research converts evidence into a forward view: patient-flow and demand models, uptake and share forecasts, budget impact models, and scenario analysis around competitor entry, guideline change or coverage decisions. Inputs come from epidemiology and secondary data for the denominator, descriptive research for current behaviour, and causal research for how behaviour responds to the variables you control.

The honest form of a forecast is a range with its drivers exposed, not a single number. Stated-intent questions systematically overstate adoption, so predictive work should discount them against observed behaviour or choice-based results, and should state plainly which assumptions the output is most sensitive to. A forecast whose sensitivities are hidden is a decision waiting to be blamed on the model.

Qualitative and quantitative: a different axis

Qualitative and quantitative is not a fifth type — it is a separate dimension that cuts across all four. Qualitative work prioritises depth, context and why; quantitative work prioritises measurement, comparability and how many. Exploratory research is nearly always qualitative, descriptive and predictive nearly always quantitative, and causal research is quantitative by construction even though its content comes from qualitative groundwork.

Mixed-method designs are the norm in healthcare precisely because the audiences are small. Ten interviews with transplant nephrologists plus a short quantitative pulse across the wider specialty will often beat either approach alone, and costs less than trying to force a statistically powered sample out of a population that does not contain one.

Primary and secondary research

The other axis is where the evidence comes from. Secondary research uses what already exists — literature, trial registries, treatment guidelines, epidemiology, prescription and claims data, tender awards, formulary listings, company filings. Primary research creates new evidence by going to stakeholders directly.

The practical rule is to exhaust secondary sources before spending primary budget, then design primary work only around the gaps that remain. In healthcare this discipline pays twice: it reduces the number of expensive respondent hours you buy, and it means the hours you do buy are spent on judgement and intent rather than on facts a respondent would have to guess at anyway.

How to sequence the four types

A typical launch programme moves through all four. Secondary research and exploratory interviews establish the landscape and the hypotheses. A descriptive survey sizes the behaviours that matter and segments the audience. A causal study — conjoint, MaxDiff or message testing — identifies which levers move the decision and where the price corridor sits. A predictive model assembles the result into a forecast with scenarios.

Sequence matters more than scale. Each stage exists to reduce what the next one has to test, so skipping forward rarely saves money; it just moves the cost into rework. If budget forces a choice, run less at each stage rather than skipping a stage — a short exploratory wave before a trimmed survey beats a large survey built on assumptions.

Matching type to budget and timeline

Cost in healthcare research is dominated by respondent access, not analysis, so the type you choose matters less to the budget than the audience you choose. A handful of expert calls with practising oncologists can be arranged in days. An exploratory qualitative wave typically takes two to four weeks of fieldwork. Descriptive and causal quantitative studies usually need four to eight weeks, longer for multi-country designs or rare specialties.

When the decision date is immovable, the right move is usually to narrow the question rather than the method. Three well-chosen expert calls that resolve one assumption are worth more than a rushed survey that half-answers five. Keep a clear note of what you did not test, so the next wave starts from the real gap.

• Days: targeted expert calls to pressure-test one assumption.
• Two to four weeks: exploratory qualitative wave, or a short descriptive pulse in a reachable specialty.
• Four to eight weeks: powered descriptive or causal quantitative study in a single market.
• Eight weeks and beyond: multi-country designs, payer and HTA audiences, rare specialties, or a tracker's first wave.