How to interview customers to validate your idea?
Why you should never ask customers, what their problem is, and other important aspects of the process.
When a VC asks you: “Did you talk to someone before building your product?” they refer to customer interviews. And almost every founder would give a positive answer to this question.
However, 8 out of 10 founders I mentor presume that “talk to someone” means they show the prototype of their product to their friend, get feedback that sounds like “Wow! Wonderful idea!” — and that’s the validation of their concept. Here’s a big surprise: this is not a validation and the prototype demo is NOT a customer interview.
These are my favourite quotes from Rob Fitzpatrick’s The Mom Test, a book that every startup founder should know by heart:
The conversations where you talk about our product only give us rope to hang ourselves
Whenever I hear a customer compliments me I hear alarms going ”you’ve messed it up”
People know what their problems are but they don’t necessarily know how to solve them.
If you go to Google and start researching the subject of customer interviews you’ll stumble upon fancy techniques like “Jobs-to-be-done framework”, “4 lists formula”, “Stages of awareness differentiation” etc.
My advice is: when you’re on ideation stage — forget all these fancy words. Just go and interview people that you consider to be your potential customers.
But follow those simple rules:
- NEVER ask a person what they think about your idea or product.
- NEVER ask them if they would pay someone for solving their problem. Instead, ask them to describe their existing process of solving the problem you think they have. Ask them to walk you through their thinking process and the way they look for a solution.
- If you’re building a solution to the problem based on the tools that already exist and that you presume customers are unhappy about — ask, what exactly are they unhappy about. What “work around” have they found to deal with inconveniences they see in the existing tools.
- NEVER ask a customer a straightforward question like “What are your existing problems?” Or “What do you feel stuck at?” People are horrible, indeed, with identifying their problems — if they were better at this , psychologists would not make a fortune helping people with finding out what their problems actually are. During customer interviews you have to be this psychologist: ask about current emotions, reactions and workflows. Never ask about future or probable actions.
If you need some more details what questions to ask, what tone to use and how to establish a connection between you and a participant to make sure they tell you truth, whole truth and nothing but truth — I have a little e-book with illustrations about customer interviews. It explains 12 rules of customer interviews, what you should do and what should not be done under any circumstances.
In gereral, be very polite and positive but don’t overeact, encouraging a participant with the words like “great”, “you did a good job”!
- Start with thanking a person for yheir time, introduce yourself and remind about the purpose of the interview.
2. Then proceed with getting a permission to record the session.
— Thank you for taking the time to talk to me today. I really appreciate it. Let me remind you that we’re doing research on X…... Our main goal here is to understand what you’re dealing with right now.
Before we get started, let me ask you if you have any questions for me?
Feel free to ask any questions that might come up.
Just one more thing: before we begin I want to make sure that it’s Ok if we record this interview.
3. Examples of the questions you will be asking:
— How and when did you start doing X (the process you’re thiking of improving with your solution)?
— What solutions you’re using right now? Can you remember the curcumstances, how did you choose them? What made you think it’s a right choice for you?
— Please, try to remember how did you find them
— What can you tell me about the existing solution that you’re using now?
4. Listen carefully and whenever you can ask to eleborate. When a participant says something about their experience or make a statement, ask “Why would you think so?” or “What made you think this way?”
Questions like this provide you with valuable insight on HOW your customer thinks and their implicit motives.
5. Prompts are a total NO. When a participant is silent for sometime, try not to offer prompting ways out — it will mess up the interview. Because a participant will stop thinking and will try to follow your lead. It’s easier, and demands no commitment, and they will get their incentive anyway — so why bother?!
6. Wrap up by thanking a customer for their time once more. If I’ve promised an incentive (like a discount code or a gift card) I usually spend the last 5 minutes making sure the participant receives it (by email, or in Zoom chat) and is comfortable with it.
The interviews are obviously done for analysis. Make sure you record and transcribe every single word (use Otter.io for this or any other transcribing tool you can get tour hands on).
Why do you need text notes? Because it’s easire to highlight, comment and tag text than audio files.
How many interviews do you need to get statistically signficant results?
Generally, it is recommended to have 20–30 interviews. However, with a strong negative pattern (when there’s no pain point and problem to solve) you can usually tell much earlier. If you are having 10 out of 10 participants that say the same thing — I’d say you don’t need more data.
It’s tougher if you get 50/50 or 60/40 split in oppinions. This basically means that you’re getting random results, and you can trust them as much as you trust a coin when you flip it. There might be two most probable reasons for this:
— either the participants pool is too small
— or you have identified the wrong segment — it’s too wide, customers show ambiguous behaviour, you have to narrow down the segment.
I recommend to start with the second presumption and go over your demographics again.
Who are your participants? Perhaps, you can narrow the segment down a bit more? If you’ve narrowed it down, ran 5 more interviews and still the pattern is very unclear, run 5 more. But if you’re getting closer to 30 interviews and your results are very ambiguous — it could mean only one thing. You’ve messed up. The segmentatuon. Or the interviews.
If you see the repeating patterns, tag them and mark the specific quotes participants have use to address to these issues — you will be using them on the landing page to talk with your customers in their own words.
After every interview try to place a customer perception of the problem you’re trying to solve, within the Eisenhower matrix.
If you’re not familiar with it — it’s a beautiful concept that is used to increase productivity but I like to use it for problem differentiation too. If you place all existing products within this matrix you’ll see that mostly, people pay for solving Important and Urgent problems.
With Not Important but Urgent it works more or less too. But lower right squire is a dump for effort and energy — you won’t be able to build a profitable business on it.
After a set of interviews and a thorough analysis you will be able to come up with a clear understanding of the following things:
— your customer segment (who has this problem you’re trying to solve and perceive it as important and urgent or at least one of those);
— who is NOT your customer — a segment you will not be targeting from the very beginning because they are unaware they have a problem, they are not looking for solutions, and never pay for those;
— what are the competitive alternatives for your product? Competitive alternatives are wider than competition because they also include an option of doing nothing or doing something manually. For instance, you offer a fancy SaaS for budgeting — all the problems that you’re offering to solve can be solved using Excel.
If you’re going to market it as a B2C solution you have to understand that thousands of people out there don’t do their family budget at all and feel very comfortable with it.