The 5 Ways to Get Customer Insight Without a Researcher

Research stopped being a job title on most teams and became a thing product managers do between other things. In User Interviews' 2025 State of User Research, 71% of organizations reported having people who do research who are not dedicated researchers, and a separate 2025 survey put the share of organizations allowing non-researchers to run studies at roughly 84%. Meanwhile dedicated UX research roles were cut harder than most functions through the 2024 and 2025 layoff waves. If you have no researcher, you are the normal case, not the deprived one.

The five ways to get customer insight without a researcher are: mine the corpus you already have before running anything new, borrow the sales and success calls already being recorded, run five-conversation cycles instead of studies, use the support queue as a standing panel, and separate the questions that genuinely need a specialist from the ones that do not. Four of the five cost you no recruiting and no calendar time, which is the actual constraint.

What you lose without a researcher, and what you do not

Worth being precise, because the answer is not "everything" and it is also not "nothing."

What you lose: study design for high-stakes generative questions, rigorous sampling, moderation skill on sensitive topics, and the discipline that keeps you from confirming what you already believed. That last one is real. Commentary on research democratization consistently warns that non-researchers produce confident conclusions from thin data, and the failure is usually sampling rather than analysis.

What you do not lose: access to what customers already said. Your company generates a continuous stream of unprompted customer language across support, calls, reviews, and community, and reading it well requires no recruiting, no incentives, and no study design. It is the largest research asset most teams own and the least used.

The practical consequence: without a researcher, shift your mix from generating new data toward reading existing data, and reserve the new-data budget for the small number of questions the corpus cannot answer.

The 5 ways to get customer insight without a researcher

1. Mine the corpus before you generate anything new

Before designing any study, ask whether your existing feedback already answers the question. Support tickets, app store and review-site entries, community threads, survey free-text, and churn reasons are all unprompted, longitudinal, and already collected.

The advantage over a study is not just cost. It is that nobody was primed by your question. A customer writing a ticket at the moment of frustration is describing the problem as they experienced it, not as they recall it three weeks later for an interviewer. What the corpus needs is grouping, which is where an adaptive taxonomy does the work a researcher would otherwise do by hand: turning thousands of records into themes without someone reading and coding them.

Answers well: what problems exist, how often, in which segments, described how.Answers poorly: why someone chose a competitor, what they would pay, how they would react to something that does not exist yet.

2. Borrow the calls that are already recorded

Your sales and success teams record conversations with customers and prospects every week. That is a large, continuously replenished body of customer conversation that nobody on the product side reads.

It is imperfect data. The questions were asked for commercial reasons, the rep steered the conversation, and prospects say things in evaluations they would not say elsewhere. But discovery calls contain exactly the material a generative study would seek: current workflow, what they tried before, why they are looking now. Ingest them alongside your other feedback rather than treating them as sales artifacts, and check what your existing tooling can already pull through customer feedback integrations.

3. Run five-conversation cycles instead of studies

When you do need new data, do not run a study. Talk to five customers about one question, weekly or biweekly, and stop.

The median B2B SaaS company reportedly went from around 6 customer research projects a year to 25 between 2025 and 2026, largely because the cost of a small cycle collapsed. Five conversations will not give you statistical confidence and are not meant to. They will tell you whether your framing of the problem survives contact with a customer, which is the question that actually blocks most decisions. Recruit from your own corpus: the accounts that filed the relevant complaints are already identified and already interested.

The discipline that matters: write the one question before the calls, ask about past behavior rather than future intent, and stop at five. Extending to twelve does not make it rigorous, it makes it slow.

4. Treat the support queue as a standing panel

Support talks to more customers in a week than you will interview in a year. Two moves make that useful rather than anecdotal.

First, get themes and volumes out of the queue systematically rather than asking support what they are hearing, since the answer to that question is always the most recent escalation. Second, add one question to the support workflow when you need something specific, so agents can gather it in the course of conversations already happening.

That second move is the underused one. A single question routed through support for two weeks yields more responses than a survey, from customers already engaged, at close to zero recruiting cost.

5. Know which questions actually need a specialist

Triage honestly. Most research demand is not specialist work, and pretending all of it is leaves you doing none of it.

Fine to do yourself: validating a problem exists, understanding current workflow, sizing how many customers hit an issue, testing whether a specific flow confuses people.

Get help: pricing and willingness to pay, positioning and messaging research, anything about people who are not your customers, and anything where the answer determines a large irreversible investment.

The failure is not doing research without a researcher. It is doing the second category as though it were the first, then acting on it with confidence. If your team has no researcher and a question falls in the second bucket, buy a small engagement rather than approximating it.

Why the constraint is synthesis, not collection

Teams without researchers usually diagnose their problem as lack of access to customers. That is rarely true. More than 40% of companies do not talk to users during development, per McKinsey, but the teams reading this are typically drowning in customer contact and unable to convert it.

The bottleneck is synthesis. A researcher's scarcest skill is not running interviews, it is turning a large volume of messy qualitative input into a small number of defensible statements. That is what disappears when the role disappears, and it is also the part that has changed most, because grouping thousands of records into stable themes is now a tooling function rather than a human one.

Which reframes what to do about a missing researcher. The answer is not to interview more, it is to make the input you already receive readable, then spend your scarce new-data budget on the few questions the corpus cannot answer. Teams that do the reverse run more interviews than they can synthesize and end up with a folder of transcripts and no conclusions.

The honest risk in this model is sampling. Your corpus over-represents customers who contact you and says nothing about the people who evaluated you and left, or who never showed up. Every conclusion drawn from it carries that bias, and naming it in your writeups is the cheap substitute for the rigor a researcher would have supplied. The same limitation is described in why your usage data and your feedback disagree: both instruments are blind to the people who never engaged.

How to run this as a habit

Set a weekly reading slot. Twenty minutes on your themes and ten verbatims. This replaces the standing research rhythm and it is the highest-value habit on this list.

Get calls into the same corpus as tickets. One place, one taxonomy, or you will read one source and forget the other.

Keep a running question list. When something the corpus cannot answer comes up, write it down instead of researching it immediately. Batch these into five-conversation cycles.

Write conclusions with their limitations attached. One line naming who the data came from and who it excludes. This is what keeps thin evidence from hardening into fact.

Escalate the second-category questions. Pricing, positioning, and non-customers get outside help or explicit uncertainty, not a best guess in a deck.

The decision rule: weight unprompted existing feedback over newly generated data for questions about problems, and weight a specialist over your own judgment for questions about money, markets, and people who are not your customers.

FAQ

How do you do customer research without a researcher?

Start with the feedback you already collect, since support tickets, reviews, community posts, and recorded sales calls are unprompted and already exist. Then run small five-conversation cycles for the specific questions that corpus cannot answer, recruiting from the accounts that raised the relevant issue.

Is it a problem for product managers to run their own research?

It is now the norm, with roughly 71% of organizations reporting non-researchers doing research. The risk is sampling rather than technique: teams draw confident conclusions from whoever happened to respond. Naming who your evidence came from and who it excludes mitigates most of that.

How many customer interviews are enough?

For validating whether your framing of a problem holds, five is usually enough, and running twelve mostly adds delay. For questions requiring statistical confidence, pricing sensitivity, or market sizing, five is not enough and neither is twenty, which is where a specialist or a properly designed instrument is warranted.

How does Enterpret replace part of the research function?

Enterpret's adaptive taxonomy groups thousands of unprompted feedback records into stable themes without anyone hand-coding them, which is the synthesis work that disappears when a research role does. The customer context graph attaches account, segment, and revenue to each record, so a finding arrives with its size and its population attached rather than as an anecdote.

What questions should you not try to answer without a researcher?

Pricing and willingness to pay, positioning and messaging, anything about prospects who did not become customers, and anything where the answer commits you to a large irreversible investment. Attempting those with informal methods produces confident answers that are wrong in ways you cannot detect.

If you have more customer contact than you can synthesize, see how Enterpret's adaptive taxonomy turns an unread corpus into themes.

‍