Agencies sell the methods they own
Every research agency has infrastructure it needs to keep busy: a panel, a tracking product, qualitative facilities, a survey platform. Proposals are shaped by that gravity, so a panel company will propose a survey for a question that needed fifteen depth interviews, and a qualitative house will propose focus groups for a question that needed transaction data. Start by writing the decision the research must inform, in one sentence, and make each agency defend its method against that sentence, including a straight account of what the method cannot see. An agency that recommends less research than you asked for, or a method it does not sell, is displaying the judgement you are actually trying to buy.
Ask where the respondents really come from
The sample is where research quality is won or lost, and it is the part buyers examine least. Online panels contain professional respondents who complete surveys for income and have learned to pass screeners; business-to-business samples claiming senior decision-makers are frequently populated by people who once held the title or aspire to it. Put the practical questions early: where do these respondents come from, what is the incidence rate for our screening criteria and does the quoted price genuinely allow for it, which fraud and attention checks run during fieldwork, and can we see anonymised verbatims while fieldwork is live rather than only the final charts? An agency confident in its sample answers precisely. Vagueness here predicts exactly the finding you will be unable to trust later.
Insight versus reporting: test the difference before paying for either
- Ask who will personally analyse the data and write the debrief, and how many other projects that person is running; senior-led pitches with junior-run delivery are the industry's standard economics.
- Ask to see an anonymised debrief from a past project, and check whether it ends in implications and choices or in a restatement of the charts.
- Ask for an example where the findings contradicted what the client hoped to hear, and what happened next.
- Agree upfront which decision the work informs, so the debrief can be judged against it rather than against completeness.
- Reserve time in the contract for a working session after the debrief, where the analysts and your decision-makers interrogate the findings together; the second conversation is usually the valuable one.
The in-house alternative deserves an honest comparison
For a meaningful share of commissioned research, the alternative was never considered: your own team, talking to fifteen customers with a disciplined guide, would answer the question in three weeks at a fraction of the cost, with the advantage that the people who hear the answers are the people who will act on them. The trade-offs are real. Internal interviewers hear what they hope for, customers are polite to the people they buy from, and some questions (pricing, competitive perception, brand) need a distance or an anonymity that only a third party provides. The test is contamination: where your relationship to the respondent distorts the answer, buy independence. Where it does not, ask whether you are commissioning research or outsourcing the discomfort of hearing customers directly.
What experienced research buyers check before signing
Operators who have commissioned research for years, and lived with the decisions built on it, keep a short list. Match the seniority of the analyst to the difficulty of the question, not to the size of the budget. Never let the final report be the first time the findings meet a decision-maker; interim readouts catch misdirected fieldwork while it can still be corrected. Treat a finding that confirms every prior belief with the same suspicion as one that contradicts them all. And hold one question central throughout: will this work change what we do? Research that cannot change the decision, whatever it finds, is reassurance purchased at research prices, and reassurance is available cheaper.
