Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Playbooks

Customer Research Programme Playbook

An operating playbook for continuous customer research: decision intake, standing recruitment, interviews and surveys, synthesis, a repository and readouts.
Playbookintermediate12 min readUpdated

What this playbook covers and when to run it

The playbook covers research as an ongoing operation rather than a project: how requests for evidence arrive and are turned into decisions worth researching, how the method is chosen, how participants are found and their consent recorded, how interviews and surveys are scheduled, how findings are synthesised and stored, and how readouts feed planning. It does not teach any of those methods. The marketing research pillar sets out the per-study process and what research can and cannot answer, and each phase below hands off to the guide that owns its method.

Run it after the first study has worked. The pillar is clear that a team which has never run a study does not need a programme; it needs one decision, one question and one small method, done properly, so the habit forms. The programme is for what comes next: when several teams start asking for evidence, when the same customers are contacted by different people for different reasons, when findings are repeated because nobody could find the last ones, or when research is commissioned and then quietly ignored.

It sits beside other listening and testing motions rather than replacing them. The voice of the customer guide runs relationship and transactional surveys and closes the loop with customers; the conversion research sprint studies one funnel in a bounded burst; the experimentation programme playbook tests changes once research has found what is worth testing. This playbook is the standing research capacity those motions draw on. Like the rest of the playbooks library, it is a starting shape: a small team may run it with one researcher and a few hours a week, while a larger one splits it across a research function, analysts and several decision owners.

Roles and responsibilities

  • Research owner. Owns the programme: the intake, the decision log, the participant pool, the schedule of conversations, the repository and the quarterly readout. Decides with each decision owner which questions get research and which wait.
  • Decision owners. The people who will act on the findings, such as a product marketer, a pricing lead or a channel owner. Each brings a decision rather than a question, agrees what evidence would change it and reports back on what they decided.
  • Interviewers and moderators. Hold the conversations and sessions with a discussion guide, take notes in the agreed format and flag anything that needs the privacy contact. In small teams the research owner and decision owners take turns.
  • Analyst. Draws survey samples, analyses counts, checks coding and states what each finding can and cannot support. Owns the bases and limits that go into every readout.
  • Privacy contact. Whoever owns data protection. Decides the Lawful Basis for each use of participants' Personal Data, approves the participant information and retention, and is consulted when a study changes how data is used.
  • Sales and support colleagues. Hear customers every day, suggest participants and topics, and help reach business buyers. They introduce; they do not screen or interview their own accounts.

A RACI matrix helps write these roles down per phase. The commonest gap is between the research owner and the decision owners: research is delivered, and nobody is accountable for deciding anything with it. Name the decision owner on every intake entry before any fieldwork starts.

Cadence, inputs and outputs

The rhythm below is a suggested starting point. Adjust it to how many decisions need evidence, how quickly they move and how much time the team can give, not to the rhythm you would like to have.

CadenceOwnerInputsOutputs
MonthlyResearch owner, decision ownersNew requests, the decision log, what the repository already holdsDecisions accepted, merged, answered from existing evidence or deferred; briefs opened
ContinuouslyResearch owner, interviewersThe participant pool, the current themes, open briefsCustomer conversations held and recorded; notes in the repository
When a decision needs countsAnalyst, research ownerA brief that needs measurement, the customer list, a pretested questionnaireA survey wave with its sample, bases and limits recorded
QuarterlyResearch owner, decision ownersFindings since the last readout, the decision logA readout; decisions recorded against the findings that informed them
AnnuallyResearch owner, privacy contactThe repository, the participant pool, consent recordsOut-of-date findings archived, records past their retention removed, the pool refreshed

Each cadence feeds the next. The monthly intake decides what the conversations and survey waves look for; the conversations fill the repository; the repository and the decision log make the quarterly readout short to prepare; the readout feeds the quarterly planning playbook, so evidence reaches the plan at the moment choices are made rather than afterwards.

The playbook in phases

  1. Take in decisions, not questions

    Every request starts from the decision it serves. The research design guide works from the decision to the research questions, and asks whether primary research is needed at all; the research brief template is the intake form, with fields for the decision, the decision owner, the questions, ethics and data. Before any fieldwork, check what is already known: the repository first, then a pass of desk research with a source log, so interviews and surveys are spent only on what nobody can answer yet. Each request leaves the monthly intake with one of four outcomes: answered from existing evidence, merged with an open brief, accepted with a decision owner and a decision date, or deferred with a reason.

  2. Choose the smallest method

    Depth or counts, chosen for the decision. The comparison of qualitative and quantitative research recommends sequencing them: explore, then measure, then return to explain what the measurement cannot. Most intake decisions can start with the standing conversations already scheduled; a survey wave is added only when the decision depends on how widely something holds. Longer methods are kept for questions that need them: a diary study when the behaviour unfolds over days or weeks and an interview would only reconstruct it afterwards.

  3. Keep recruitment and consent standing

    Recruitment is the slow part of research, so the programme keeps it warm. The recruiting guide builds the screener, the invitation on a research footing and the exclusions that stop the same voices recurring; the pool holds only people who agreed to be contacted for research, with what they agreed to and when. The tutorial on building a research participant pool sets that pool up, from the routes customers join by to rotation, a participation record per member and a review on a fixed date. The participant information and consent tutorial turns the principles the site's research guides describe from the MRS and ESOMAR codes into documents: participation is voluntary and informed, people are told who is collecting the data and why, answers are confidential unless people agree otherwise, and research is never a pretext for selling. The lawful basis, retention and wording belong to the privacy contact. When a survey wave is needed, drawing a sample from the customer list records the draw so the next wave can repeat it.

  4. Run interviews on a rhythm

    Conversations are scheduled whether or not a project is live, each with a discussion guide built around what the participant actually did. The customer interviews tutorial owns the method: past behaviour rather than opinions, guides that avoid leading, and notes taken in a consistent form. Between projects, the conversations follow the themes the last readout left open, and the jobs-to-be-done framework gives a stable frame for what to listen for across many conversations. Where the buyer is a business, researching B2B buyers covers whose experience answers the question and how to reach participants through account owners without letting them choose who is heard.

  5. Run surveys when counts are needed

    A survey wave starts only from a brief that needs measurement, and its questions come from what the conversations found. The survey and questionnaire design guide owns question types, wording, order, length and consent; pretesting the questionnaire catches confusing questions, missing options and broken routes before the survey reaches customers, because once it is in the field the answers cannot be asked again. The analyst writes down the population, the sample and the bases before fieldwork, so the readout can say whom the results describe. Questions of Sample Size are settled in the design guide, not by a rule of thumb in the programme.

  6. Synthesise and record

    Notes, transcripts and open answers become findings that can be found again. Coding open-text answers turns comments into themes counted against a stated base, with a second person checking the coding. Where a model helps at volume, AI for research synthesis keeps every theme traceable to evidence and has quotes and counts checked by a person, because a model is unreliable at exactly the parts that make synthesis worth doing. Each finding then goes into the repository that building a research repository sets up: one record per finding with its source, strength and date, with identifiable material kept in the study's restricted store rather than in the shared record.

  7. Read out and follow the decisions

    Findings are reported so they get used. Reporting research findings leads with the answer to the decision, states how strong each finding is and what the study cannot say, and ends with recommendations that have owners. The quarterly readout gathers what the period found, and the decision log records what each decision owner decided and which findings informed it. When a question outgrows the team, such as a large survey, a population the pool cannot reach or a need for independent fieldwork, commissioning a research agency covers what to buy, how to brief it and how to accept deliverables that show their method, bases and limits.

The research intake: what qualifies and what waits

The monthly intake is where the programme protects its time. Requests arrive as questions ("what do customers think of the new plan?"); the intake turns each into a decision, an owner and a date, or sends it back. The table sets out the usual outcomes. It is a working rule for triage, not a scoring model; use the impact and effort matrix if several accepted requests compete for the same weeks.

The requestOutcomeWhat happens next
Names a decision, an owner and when it will be madeAcceptBrief written from the template; method chosen; added to the conversation themes or a survey wave
Asks something the repository already answersAnswer from existing evidenceResearch owner sends the findings with their date and strength; request closed
Overlaps an open briefMergeQuestions added to the open brief; requester joins as a second decision owner
Has no decision behind it ("it would be interesting to know")DeferSent back with the question "what would you do differently with the answer?"
Needs a population or scale the team cannot reachCommission or deferBrief prepared for outside help, or deferred with a reason
Wants evidence for a decision already madeDeclineExplained in the decision log; offered a check after launch instead

Two rules keep the intake honest. A request without a named decision owner is not accepted, however interesting the question. And every accepted request has a date by which the decision will be made, because evidence that arrives after the decision is a record, not research. Segment questions are handled the same way: a request to rebuild a Buyer Persona or revisit Segmentation or the Ideal Customer Profile comes with the decision it will feed, or it waits.

Illustrative example — not a real result

A payroll software company keeps talking to customers

Quobrant Payroll, an invented payroll software company for small employers, ran one study last year on why trial users did not convert, and the findings changed its onboarding. Within weeks, 7 new research requests arrived from 4 teams. The research owner starts a monthly intake: in the first session, 3 vague requests about "what customers think of pricing" become one decision, whether to add a plan for employers with fewer than 10 staff, owned by the pricing lead and due in 10 weeks. Two requests are answered from the earlier study's findings. A standing pool is built from customers who ticked a box agreeing to be contacted for research; 64 people join in the first quarter, each record noting what they agreed to and when. Interviews run every fortnight whether or not a project is live, 2 per session, on the themes the last readout left open. The pricing decision needs counts, so one survey wave goes to a random sample of 1,500 customers drawn from the list after a pretest with 5 people that removes 2 confusing questions; 210 respond, and the analyst reports the base on every chart. The first quarterly readout lists 9 findings in the repository and shows 3 decisions they changed, including the new plan. At the year-end review, 40 pool records past their agreed period are removed and 22 findings older than the current product are archived.

Measuring the playbook

The programme works when research changes decisions and the effort to get evidence keeps falling. Operating signals show whether the rhythm is holding: whether the monthly intake happens and closes each request with an outcome, whether conversations continue between projects, whether every survey wave was pretested and recorded its sample and bases, whether findings reach the repository soon after fieldwork, and whether the quarterly readout runs.

Outcome signals show whether the programme earns its keep: the decisions recorded in the log with the findings that informed them, requests answered from existing evidence instead of new fieldwork, and repeat studies avoided because the repository already held the answer. The single most telling measure is the share of readout findings that a decision owner can point to a decision they changed; the repository's record of the decisions each finding informed makes it easy to count. Report these in the team's existing quarterly pack, which the reporting cadence playbook sets up, rather than in a separate research report.

Watch the health of the participant pool too: whether consent records are current, whether the same people are being heard too often and whether the pool still reflects the customers the decisions concern.

Common mistakes

  • Taking questions instead of decisions. A request without a decision behind it produces interesting findings nobody acts on. Turn every request into a decision, an owner and a date, or send it back.
  • Restarting recruitment for every study. Finding people from scratch each time is the slowest way to research. Keep a standing pool with consent recorded, and refresh it.
  • Treating consent as a tick box. A pool built from people who never agreed to research contact, or whose consent records cannot be found, is a weak foundation. Keep the records with the pool and review them yearly.
  • Letting sales choose who is heard. Colleagues who introduce participants should not screen or interview their own accounts. Use the screener.
  • Surveying before listening. A questionnaire written before any conversations measures the team's assumptions. Explore first, then measure.
  • Findings that live in slide decks. A finding nobody can find is research that will be run again. Record each one in the repository with its source, strength and date.
  • Readouts without a decision log. Without a record of what was decided, the programme cannot show its value and will be the first thing cut. Ask each decision owner what changed.

Frequently asked questions

When is a team ready for a research programme rather than a single study?

When one study has been run properly and its findings changed a decision, and when requests for evidence keep arriving faster than anyone can answer them one at a time. A team that has never run a study is better served by one decision, one question and one small method done well. The programme is what keeps the habit going once that first study has shown its value.

Do customer conversations need to happen when no project is live?

This playbook suggests keeping a light, standing rhythm of conversations even between projects, because recruitment, scheduling and consent are the slow parts of research and a standing rhythm keeps them warm. The frequency is a starting point that the team adjusts to its capacity and to how quickly its decisions move. What matters more than the number is that each conversation is recorded where the next person will look.

Who decides how participants' personal data is handled?

Whoever owns data protection in the business decides the lawful basis, retention and the wording that participants see. The research owner's job is to make those decisions possible by setting out clearly what each study will do with what people say, and to keep the consent records current. Participation stays voluntary and informed whatever legal basis applies, as the professional research codes describe.

Terms used on this page

Frameworks

Where this page is used

Sources

  1. Market Research Society — MRS
  2. ESOMAR — ESOMAR

Related