Customer Discovery Evidence: What Counts and How to Show It

Customer discovery evidence is anything a real customer said, did, or paid for that tests one of your assumptions. The strongest kind is behavior: a workaround they already use, money they already spent, a pilot they agreed to run. The weakest kind is praise for your idea. This guide shows you how to collect the strong kind, label it honestly, and put it in front of investors in a form they can’t easily pick apart.

Trydialectic

trydialectic.com

Stress-Test Your Investor Evidence

Dialectic helps founders uncover fragile assumptions and rehearse tough investor questions before presenting their pitch.

Explore Dialectic

TL;DR

Evidence is a customer’s past behavior or commitment. Compliments and “I’d definitely use that” are weak signals.
Interview about a recent event, then run experiments. Conversations give insight, and experiments give stronger proof.
Log every interview, pull direct quotes, and label each assumption validated, invalidated, or needs more information.
Show investors your strongest evidence first, with exact counts, clear categories, and a plain statement of what you haven’t validated.
Expect investors to attack the gaps. Rehearse the hard questions before the meeting.

What counts as customer discovery evidence
Strong evidence versus weak evidence
How to gather evidence through interviews
Turn conversations into experiments
Log it so it holds up
Find the weak spots in your business model
How to present customer discovery evidence to investors
Our perspective
Test your evidence before the investor does
FAQ
Sources

What counts as customer discovery evidence

Harvard Business School’s Rock Center guide describes customer discovery as building a clearer picture of your target market: who the customers are, their journey, the solutions they use today, their frustrations, and how they decide. Evidence is whatever you collect that supports or contradicts that picture.

The same guide names interviewing, surveying, and observing as the starting methods. A broader HBS background note also lists surveys, usability tests, market trials, split tests, Net Promoter Score, focus groups, and conjoint analysis as techniques for entrepreneurial research, and it explains that founders’ research goals differ from those of managers working in established markets.

The stakes are real. The Rock Center guide says a common cause of startup failure is not validating that the problem is real and that customers are looking for a solution. Evidence is how you find that out before you build, and before an investor asks.

Needs are situations, not feature requests

Pragmatic Institute defines a customer need as a valuable benefit a customer seeks in a particular context, and it says needs should be studied as situations and desired outcomes rather than requested features. HBS Online adds that needs can be functional, social, or emotional. A customer who says “add a dashboard” has handed you a solution. The evidence is the situation behind it: what they were trying to do, what got in the way, and what they did next.

Strong evidence versus weak evidence

Not every data point deserves the same weight. The UC San Diego guide Talking to Humans advises treating praise for your idea as weak evidence and looking for buying behavior. It also cautions against treating small-sample statistics as indisputable.

A practical way to sort what you have:

Signal	Strength	Why
“Great idea, I’d use it”	Weak	Costs the speaker nothing
Signed up for a waitlist with no qualification	Weak	Shows curiosity, not need
Described a recent painful event in detail	Moderate	Real situation, but no commitment
Uses a clumsy workaround today	Moderate to strong	They are already spending time on the problem
Spent money on a prior solution	Strong	Proves budget and urgency
Introduced you to the buyer or shared data	Strong	Costs them effort and social capital
Agreed to a paid pilot, deposit, or pre-order	Strongest	A real commitment

The table is our ordering, built from the principle above: the more it costs the customer to say yes, the more it tells you. Affiniti Venture Studio makes a similar point, advising founders to tally active workarounds, prior spending, time-sensitive pressure, access to buyers, data sharing, and willingness to run a pilot instead of counting positive reactions.

Category interest also isn’t product demand. NC IDEA warns that the existence of similar products shows interest in a category but not demand for your specific product. Its fix is to test whether a defined group will respond to a concrete call to action.

How to gather evidence through interviews

The Talking to Humans method starts with the riskiest assumptions. Then you recruit relevant candidates and build an interview plan around a recent event. Keep notes and look for patterns across conversations.

Ask about the past, not the future

Penn State’s Customer Discovery 101 workbook recommends neutral, open-ended questions about current behavior, with follow-ups like “why” and “tell me more”, rather than leading questions. A pitch disguised as a question produces polite answers and no evidence.

Affiniti’s list of strong questions covers a recent event, what the customer did, the cost of the workaround, how often it happens, what happens if they do nothing, what they’ve tried, and what would be required for a paid pilot. In plain form:

“Walk me through the last time this came up.”
“What did you do about it?”
“What did that cost you in time or money?”
“How often does it happen?”
“What have you tried before? Why did you stop?”
“If nothing changes, what happens?”
“Who else has to say yes before this gets paid for?”

Question 7 matters more than founders expect. It exposes the gap between the person who feels the pain and the person who signs the check.

How many interviews are enough

There is no magic number. NC IDEA puts quality first and suggests at least 15 interviews, noting you may need more than 20, or that patterns can show up after 5 or 10. Stop when new conversations stop surprising you, not when you hit a count. Always record how many people you spoke to and how you chose them, because an investor will ask.

Turn conversations into experiments

Interviews tell you what people say. Experiments tell you what they do. Giff Constable’s Testing with Humans draws the line clearly: conversations give insight and inspiration, while experiments give stronger proof on whether an assumption holds.

Talking to Humans suggests experiments such as an online store, ads, or A/B tests of copy, landing-page messaging, and prices. Each has a clear pass or fail result. If you are moving toward a product, the same logic applies to testing an MVP with real users: define what behavior would prove the assumption wrong before you build.

A simple experiment design:

Assumption: “Operations managers at mid-size clinics will pay to automate scheduling reports.”
Test: A landing page with a price and a “start a paid pilot” button, sent to 100 managers you’ve screened.
Pass line, set in advance: For example, ten click through and three book a call about pricing.
Result: Pass, fail, or inconclusive, written down the same day.

Setting the pass line before you see the result keeps you honest. The numbers above are an example, not a benchmark.

Log it so it holds up

Evidence you can’t retrieve isn’t evidence. NC IDEA recommends recording interviews and pulling direct customer quotes, then deciding whether each quote supports, contradicts, or leaves ambiguous your hypothesis, persona, and customer journey.

Penn State’s workbook goes one step further. It asks you to organize observations into themes and sub-themes and then mark each assumption validated, invalidated, or requiring more information, while recording new assumptions and next steps. Validation is provisional. Every result should produce a new question.

A log row can be as simple as:

Assumption	Type	Evidence	Status	Next step
Buyers have budget authority	Viability	4 of 12 said a director must approve	Needs more info	Interview directors
Problem occurs weekly	Desirability	9 of 12 described an event in the past month	Validated	Quantify cost
Users will upload their own data	Feasibility	2 of 12 would share data	Invalidated	Test a sample-data version

Find the weak spots in your business model

Evidence is most valuable when it contradicts you. Penn State’s workbook suggests listing your assumptions, classifying them as desirability, feasibility, or viability, and prioritizing by uncertainty and the impact of being wrong.

Affiniti notes that discovery can reveal that users and buyers differ, that a problem lacks frequency or urgency, that a workaround is good enough, or that buyers lack authority or budget. Those findings point to decisions: build, reposition, narrow, or pause.

These are the same weaknesses a skeptical investor goes looking for. Find them first and you can fix them or explain them. Wait for the meeting and you’ll be improvising.

How to present customer discovery evidence to investors

A slide full of enthusiasm reads as inexperience. A slide that separates signals by strength reads as judgment. Our recommendations, drawing on the research above:

Lead with the strongest signal. Paid pilots, deposits, and signed letters of intent come before interview counts.
Use exact numbers and define the sample. “11 of 14 clinic operations managers at 50 to 200 bed facilities” beats “most customers.”
Label categories honestly. Keep paying customers, beta users, waitlist signups, LOIs, and pilots in separate buckets. Don’t add them into one “customers” figure.
Say what the evidence means. One sentence of interpretation per data point.
Add one or two direct quotes that show the problem in the customer’s own words.
State what you haven’t validated. Investors trust a founder who has mapped their own blind spots.

If you have a live product with few customers, show repeated use. TechCrunch’s guidance recommends a daily-active-user histogram over the most recent 28 days to identify power users, because repeat usage is more persuasive than one-time activation.

Watch for the usual overreach: treating a 20-person sample as a market size, counting friendly introductions as a representative sample, and presenting a pilot you’re still negotiating as a customer.

Our perspective

Most founders we talk to have more evidence than they realize and trust it less than they should in the wrong places. They walk into a meeting proud of the interviews that went well and quiet about the ones that didn’t.

Investors do the opposite. A partner reads your deck looking for the assumption that carries the most weight and has the least support. Maybe the buyer isn’t the user. Maybe the problem is real but not urgent. Maybe the workaround is good enough. If your evidence log already names those risks, you can answer calmly. If it doesn’t, the question lands cold.

That is why we built Dialectic around adversarial, partner-style questioning rather than supportive feedback. Friendly review tells you what is working. We are designed to surface fragile assumptions, blind spots, and the tactical pivots that could turn them into strengths. Your discovery evidence is the raw material. The stress test shows which parts of it will hold under pressure and which parts need another round of interviews first.

Rehearsal matters as much as the audit. Knowing your numbers is different from defending them when someone interrupts. Practice against tough, realistic questions is what turns a good log into a confident answer.

D

Test your evidence before the investor does

If you’ve gathered interviews, experiment results, and pilot commitments, the next step is finding out how they read to a skeptic. Dialectic takes your pitch deck or memo and questions it the way a seasoned VC partner would. You get fragility and risk scoring, a structured list of weaknesses and blind spots, and actionable feedback. A real-time boardroom rehearsal grades your defense of the hardest questions.

The Tactical Audit is $25 one-off per audit, good for a single deck before a specific meeting. Founder Pro is $49 per month for ongoing iteration, and Venture & Studio 5 Seats is $149 per month for teams. You can upload your deck and see the plans. To see what the output looks like first, read the Nexus take-rate teardown or the ScribeAI sample teardown.

FAQ

Why is it important to discover customer needs?

It tells you whether the problem is real before you spend time and money building. Harvard Business School’s startup guide links startup failure to skipping validation that customers are looking for a solution. It also shows you who the buyer is, what they use now, and how they decide.

What is the difference between customer discovery interviews and experiments?

Interviews give you insight into the customer’s situation and language. Experiments, such as a priced landing page or an A/B test, show what people do when something is at stake. Use interviews to form assumptions and experiments to test the riskiest ones.

Is a waitlist good enough evidence for investors?

A waitlist shows some interest, but an unqualified list is a weak signal. It becomes stronger when you can say who is on it, how they were screened, and how many took a further step such as booking a call or paying a deposit. Show it in its own labeled category.

How many customer interviews do I need?

NC IDEA suggests at least 15 interviews, and some products need over 20. Others show clear patterns after 5 or 10. Keep going until answers stop surprising you, and record your sample so investors can judge it.