When companies want to understand customers, they often start with people who have bought: what they like, how satisfied they are and whether they would recommend the product. These answers matter, but they do not fully address a different question: why did someone show interest, speak to the team and even request a quotation, yet ultimately decide not to buy? Customer insight for growth needs to include people who left the buying process.
Buyers and non-buyers answer different questions
Existing customers help explain the value they accepted and the experience after purchase. However, they have already overcome at least some barriers to becoming customers. Listening only to them provides insufficient evidence about the barriers keeping others outside. A registration process that current users consider acceptable may still be difficult for a first-time visitor.
This article proposes a practical approach: study the decision not to buy as a research question, rather than treating it solely as an unsuccessful sales outcome. This does not mean that every person who declines should be persuaded. The aim is to distinguish poor-fit prospects from suitable customers facing a barrier that the company could address.
Do not label every case “no need”
Before interviewing, separate cases by their actual journey: people who stopped before a conversation, people who received a proposal but postponed a decision, people who chose another supplier and people who continued with their existing approach. Former customers who left deserve a separate group because they have post-purchase experience that non-buyers lack.
For each case, record the stage reached, the task the person was trying to accomplish and the final decision. A sales record marked “lost on price” is only a starting point. Was that the customer’s explanation, the salesperson’s interpretation or a conclusion supported by evidence? Without that distinction, research may simply reproduce the team’s existing assumptions.
Ask about a specific decision, then compare it with behavioural evidence
Instead of asking “Would you buy if the price were lower?”, start with “What prompted you to consider this solution most recently?”. Reconstruct the process: which alternatives were considered, who participated, what information was missing, when the purchase stopped and what the customer did next. These are exploratory interview prompts, not a validated measurement scale.
Accounts of past events still require caution. Nielsen Norman Group distinguishes self-reported thoughts and feelings from observed actions, and notes that memory and question wording can affect responses. A retrospective account should therefore not be treated as an exact behavioural record.[1]
Where customers consent and the company is authorised to use the information, compare the account with conversation timelines, proposal versions or the actual stage at which the person stopped. Avoid collecting personal information that does not serve the research question. The purpose is to understand the decision, not to prove the customer wrong.
An example: is price really the problem?
A hypothetical illustration, not a research finding or a Dr.Hani client case. A software provider notices that some small businesses stop after receiving a quotation. Sales records describe the reason as “high cost”. Further conversations suggest that some prospects worry about staff time for migrating data and the absence of an implementation owner.
A possible hypothesis is that, for businesses with limited implementation capacity, perceived switching costs form part of the price of adopting the solution. A subscription discount might not remove the main obstacle. This suggests testing a proposal that includes migration support and a clearly identified implementation contact. However, that approach may not help customers who genuinely lack the budget or need for the product. Competing explanations must remain open; it would be premature to conclude that price is never the problem.
Turn the explanation into a testable hypothesis
A useful insight record specifies the customer group, situation, barrier, available evidence and remaining uncertainty. Only then should the team choose an action to test. In the example above, the company could compare its usual proposal with one offering implementation support, provided there are enough suitable cases and a credible basis for comparison.
Define outcomes before the test, such as the proportion starting to use the product after receiving a proposal, time to first use and support cost. A simple before-and-after comparison cannot attribute every change to the new proposal: lead sources, timing and sales practices may also differ. Initial interviews generate hypotheses; they do not establish the percentage of the market experiencing the same barrier.
Listen to non-buyers to understand the limits of the promise
Researching non-buyers is not about finding a clever message that convinces everyone. It helps identify promises that lack sufficient supporting evidence, difficult steps in the process and groups the company should stop pursuing. Sometimes the appropriate action is to change a process, sometimes to provide clearer information, and sometimes to accept that the product is not yet suitable for that group.
Customer insight needs perspectives that existing data may not capture. Before the next marketing meeting, ask: among the people we are listening to, has anyone decided not to choose us?
References
[1] Page Laubheimer (2024), Attitudinal vs. Behavioral Research in UX, Nielsen Norman Group. This source supports the distinction between self-reported and behavioural data. The practical approach and scenario in this article are the author’s recommendations and illustration.
Illustrative photo: Vitaly Gariev / Unsplash, used under the Unsplash License. The photograph does not depict research participants.
