How these pages work
Every diagnosis here begins with the paperwork the study should already have. A response problem starts with the completion funnel from invited to finished; a representativeness problem starts by writing the population down and comparing the people who answered with the people who were invited; an interview problem starts by reading two transcripts and the screener; a problem with findings nobody used starts from the decision the study was meant to inform; a challenged market size starts by writing each layer's definition so that someone else could apply it; and a study running late starts by setting the decision date beside the plan, stage by stage. Once the fault is visible, a diagnosis table lists what you can see, the check that confirms each cause and the fix. Each cause then gets a short section with the fix, a way to verify it and the habit that prevents it, followed by a routine that keeps the next study clear of the same fault.
The pages assume the methods are taught elsewhere and link to them. Starting from the decision and writing the brief is in the research design guide; question wording, order and length are in the survey and questionnaire design guide; finding the right people and writing a ScreenerThe short questionnaire that confirms each research candidate fits a study, asking about recent behaviour without revealing which answers qualify. is in the recruiting tutorial; and choosing between counting and understanding is in the qualitative versus quantitative comparison.
Find the page for your symptom
Choose the row that matches what you are seeing. Diagnosis pages in this section are also listed beneath this introduction automatically; the last three rows send questions that belong somewhere else to the page that owns them.
| What you are seeing | Start with | Background |
|---|---|---|
| A survey in the field has far fewer completed answers than planned | Survey response rate low | Survey and questionnaire design |
| A reader disputes that the results describe the customers or the market | Survey results not representative | Drawing a survey sample from a customer list |
| A study was received politely and changed nothing | Research findings not acted on | Reporting research findings so they get used |
| A round of interviews produced praise and stories but nothing a decision can use | Customer interviews not producing insight | Customer interviews, step by step |
| A board, investor or finance director has questioned the market size | Market size estimate challenged | Market sizing |
| A study will report after the decision it was meant to inform | Research taking too long | Research design and the brief |
| You need to know whether a survey score predicts what customers do | Testing whether survey scores predict churn | Voice of the customer |
| A pile of open-text answers needs turning into something countable | Coding open-text survey answers | Survey analysis plan template |
| The study is too large or specialised to run in-house | Commissioning a research agency or freelancer | Research design and the brief |
Design, fieldwork or reporting: decide which stage failed
Nearly every symptom in this section traces to one of three stages, and naming the stage first saves most of the diagnosis. A design failure happened before anyone was asked anything: no decision named, a customer list treated as the market, groups the decision compares never represented on purpose, questions about opinions and the future instead of specific past behaviour, a market size built from a single chain of percentages. A fieldwork failure happened while the study was running: an invitation that gave no reason to answer, a frame full of dead addresses and people asked last month, one question that lost most of the people who reached it, participants recruited from whoever replied first, a group session where one voice led. A reporting failure happened after the data was in: no synthesis step and no CodeframeThe defined set of codes open-text answers are tagged with, built from a sample, applied consistently, checked by a second coder and counted against a base., interviews read as counts, heavy weights letting a handful of respondents drive a figure, a report with the answer buried, recommendations with no owner, findings left in a deck.
The test is to ask where the evidence first went wrong. If the population, the sample or the questions were wrong before fieldwork, no amount of analysis will rescue the study, and the fix belongs in the next brief. If the right people were asked the right questions and too few or the wrong ones answered, it is fieldwork, and the completion funnel says where. If the data is sound and nobody used it, it is reporting. A small sample sizeThe number of users or sessions an experiment needs to detect a given effect reliably. is a reporting question as often as a design one: the figure may be fine as long as the report says what it can and cannot support.
The professional codes the site's research guides describe from the Market Research Society set the standard these pages hold to at every stage: participation is voluntary and informed, and findings are reported with their limits.
When to run the programme instead
A diagnosis page fixes one study. When several of these symptoms arrive together, when every study starts from scratch, or when the same question keeps being researched because nobody can find the last answer, a standing programme is the better tool. The customer research programme playbook joins decision intake, method choice, a recruitment pool with consent kept current, regular conversations, survey waves, synthesis into a repository and readouts that track decisions; the survey launch checklist is the gate between a written questionnaire and a live invitation, run again before every wave; and the guide to building a research repository keeps findings where the next brief will find them. The customer research section and the methods section place the guides these pages draw on side by side.
