A customer interview is a conversation about what happened at a company before they bought. Not about your product, not about what they like, and not about what they wish existed.
It is the cheapest source of advantage a B2B company has, and almost nobody uses it. Twelve conversations of forty minutes produce material you cannot buy from any tool, because it describes your customers rather than the market in general.
Interviews answer questions CRM data cannot. What broke before they started looking. Who first said something had to be done. What they tried earlier and why they gave up. Who had to approve it. What they were afraid of at signature.
Three things come out of those answers at once: ICP criteria based on situation rather than industry, a list of buying signals to watch for, and the language customers use to describe the problem. That last one is usually the painful part, because it turns out the whole website is written in words nobody uses.
Not only the happy ones. A good set is around twelve conversations across four groups.
Customers who stayed longest and pay most. This is where the ideal customer profile comes from.
Customers who churned in the first year. The most uncomfortable group and the most valuable, because it shows who not to take.
Companies that talked to you and did not buy. They will tell you what tipped it, and it is usually not price, though everyone internally is convinced it was.
People on the buying committee who were not your champion. Finance, security, IT. They see the same purchase completely differently, and their objections are what stop deals at the end.
Twelve is the number at which new themes stop appearing. If you are hearing the same things by the ninth conversation, you can stop early.
The whole craft is asking about the past rather than about opinions. Opinions are invented on the spot, memories are not.
Take the person back to the moment before the purchase and work forward in order. What was happening at the company when you started thinking about this. What was the first moment someone said this could not go on. How were you coping without it. What made that stop being enough.
Then the buying process. How did you find us. Who else did you talk to. Who had to agree internally. What was hardest to defend inside the company. What would have happened if you had bought nothing.
Finally one question that usually yields the most: if our product disappeared tomorrow, what would you do. The answer shows whether you are necessary or merely convenient.
Do not ask whether they like the product. Out of politeness they will say yes.
Do not ask which features they would want. You will get a wish list disconnected from anything anyone would pay for.
Do not ask how much they would be willing to pay. People do not know this about themselves and will answer with a number that sounds reasonable in conversation.
And do not sell during the interview. One attempt at an upsell turns the conversation into a sales call and the rest of the answers are worthless.
Forty minutes, video, recorded with consent. The recording matters more than notes, because exact wording is the point and whoever takes notes always smooths it out.
One person leads, a second listens and does not interrupt. If the rep who owns the account runs it, the customer will be polite. Better that it is run by someone with no commercial relationship with them.
Silence is a tool. After an answer, wait three seconds instead of asking the next question. The second half of the answer, the part after the silence, is usually the true one.
Send the invitation as a request for help, not as research. Response rates rise noticeably when you say you want to understand how they made the decision and that the call is not commercial. A small return favour, a summary of all the conversations sent to participants, raises it further and costs nothing.
The material is useless while it sits in a folder. Processing it takes a day and produces four things.
A list of recurring situations before purchase. These are your fit criteria and at the same time a hint about what to watch for as a signal.
A customer vocabulary. The exact phrases they use for the problem. These go straight onto the site, into emails and into ads, unpolished.
A map of the buying committee. Who appears, at what point, and what blocks them.
A list of objections in the order they arrive. That is material for sales and the only way to stop the answers being improvised on calls.
Decide when to repeat it. Once a year is enough, sooner after a significant product change or entering a new segment.
Will customers agree? More agree than anyone internally expects, especially when the request is for help rather than for a survey.
Is a survey not enough? No. A survey collects answers to questions you already know to ask. An interview shows you what you did not know to ask.
Can interviews be outsourced? Yes, and sometimes better, because people are more candid with an outsider. But someone from the company has to listen live, otherwise all that is left is a report nobody believes.
Interviews produce the criteria for account scoring and the list of buying signals. They are also the raw material for mapping the buying process. The full model is in the guide to Account-Based Marketing for B2B SaaS.
Daniel Tupczyński runs MOAT Marketing, an Account-Based Marketing partner for B2B SaaS companies. MOAT builds and runs ABM systems alongside the client’s own sales and marketing team, from market research and prospecting into Tier 1 accounts through to insights for product and strategy, so the know-how stays in the company. He also runs marketing at a B2B SaaS company.