The first 8 calls we ran on Carriva were terrible. Not because the prospects were difficult. Because we were. We would describe the product in detail at minute 2, ask "would you use this?" at minute 4, and get a polite "yes interesting" at minute 5. Six weeks later, none of those polite-yes prospects converted. We had run customer-discovery theater, not customer development. Then we re-read Rob Fitzpatrick's "The Mom Test" and rewrote our entire interview playbook. The next 12 calls landed differently. This is what we learned about Mom Test customer interviews SaaS founders should actually run.
The premise of the Mom Test, in one paragraph
The Mom Test is a small book by Rob Fitzpatrick about customer interviews. Its core argument: people lie to you, especially nice people, especially when you tell them about your idea. The fix is to never describe the idea, never ask hypotheticals, and always anchor on the prospect's actual past behavior. If you do those three things, you cannot get a falsely positive answer because your questions do not have a "right" answer to give you.
We had read the book two years ago. We had not internalized it.
The first 8 calls, in detail
Our first round of Carriva calls in late 2024 went like this:
- Pleasantries (3 minutes).
- "We are building a tool that audits a client's RIS document for errors and missing rights, with a free B2C detector for end-clients." (3 minutes of monologue, illustrated.)
- "Would you find this useful?" (the killer question).
- The prospect would say "yes, it sounds useful" (because we are a likeable founder and they were being polite).
- Pleasantries (3 minutes).
- End of call.
8 calls. 8 polite yeses. 0 conversions when we followed up.
The critical mistake: we never let the prospect tell us about their actual workflow. We were so eager to validate our idea that we trampled the only signal that mattered.
If you spend more than 90 seconds describing your product in a discovery call, you are not doing customer development. You are doing a sales pitch with a fake hat on.
The Mom Test rewrite
We rewrote the entire interview structure. Here is what the 12 follow-up calls looked like:
- Pleasantries (1 minute).
- "Walk me through what happened the last time a client brought in a RIS." (open-ended, anchored in actual past behavior).
- We listen. We take notes. We ask "and then what happened?" "How long did that take?" "What did you do when you found that?" "How often does that come up?"
- We ask about other tools. "What did you try before? What did you stop using? Why?"
- We ask about money. "What do you charge a client for an audit like the one you just described? How much does this kind of work bring you per month?"
- End of call.
We almost never described Carriva. If a prospect asked "so what does your product do?" we gave a 30-second answer and pivoted back to their workflow.
The result: dramatically more useful information. We learned about specific RIS error patterns we had not anticipated. We learned that some advisors keep an Excel macro for trimester counting that has not been updated since 2018. We learned that the most painful part of an audit is often not the audit itself but the conversation with the client where the advisor has to explain a 14-page document. That last insight reshaped our roadmap.
The questions that worked
Specific questions we now ask in every Carriva discovery call:
"What is the last thing you did for a client today?"
Anchors in concrete, recent reality. Avoids hypotheticals.
"How long did that take you?"
Surfaces the time cost. Time is the universal currency.
"What is the worst part of doing that?"
Lets the prospect identify their own pain instead of us projecting our solution onto them.
"Have you tried solving this before?"
Reveals competing solutions, including ones we did not know existed (Factorielles' Zoom Carrière, in our case).
"What did you stop using? Why?"
Past failures are signal. They tell you what does not work in this market.
"How much would saving 60 minutes a week be worth to you?"
A money question without being a price question. The answer ("about 200 EUR a month, maybe 300") tells you the willingness-to-pay range.
"Who else struggles with this?"
Generates referrals. The prospect is happy to give you names because you have not asked them to buy anything. We got 9 of our next 12 conversations from this question.
The questions we stopped asking
Equally important to know what to cut:
- "Would you find X useful?" Always yes. Useless.
- "Do you think a tool like this would help you?" Same.
- "How much would you pay for this?" The answer is always cheaper than reality. Anchors them on a low number.
- "What features should it have?" The answer is always more features than they will ever use.
If you ask these, you collect noise dressed as signal. We collected a lot of noise in our first round.
Specific things we learned that we would not have predicted
After 20 calls (8 bad, 12 good), three insights stand out:
1. The free B2C detector is the marketing weapon
When we asked "what is the worst part of finding a new client?", several advisors said "they don't know they have a problem until I look at their RIS, and by then they are already at my office and not paying me to be there." A free B2C detector that lets a prospect check their own RIS for anomalies and then prompts them to book a meeting with their advisor of choice solves this. We had built it because we thought it would be useful. We had no idea it was the single most asked-for feature.
2. The "audit" framing wins, not "simulation"
Repeatedly the prospects pushed back on the word "simulation" because every retirement-advisory tool already markets itself as a simulator. They wanted an audit. The distinction is not pedantic. It changes the SEO, the sales pitch, and the product roadmap. We discussed this in our piece on why we built Carriva: Carriva is AUDIT (diagnostic of past), not SIMULATION (projection of future).
3. The willingness-to-pay is higher than we feared
When we stopped describing the product and started asking how much they charge for an audit themselves, the answers were 250 to 600 EUR per audit. Our 79 EUR per month per seat suddenly looked very reasonable in context. Before this, we had been afraid to charge more than 49 EUR. The interviews fixed our pricing courage. We covered the broader pricing thinking in our pricing AI features for SMBs post.
The mechanics of running these calls
Tactical notes from running 20 of these:
- 30 minutes is the right length. 60 minutes is too long; the prospect feels harassed. 15 is too short.
- Calendly + Zoom + a Google Doc for notes is the entire stack. We tried more sophisticated tools. The friction was not worth it.
- Don't record without explicit consent. Some prospects are uncomfortable, and you can take great notes by hand for a 30-minute call.
- Send a follow-up the same day with a 3-line summary of what they said. They will correct you if you misunderstood. They will also remember you fondly.
- Mark every prospect with a tag: "high pain", "low pain", "competitor user", "won't buy", "willing to test". Use these to prioritize follow-up.
We talked to 20 prospects in mid-April 2025 from a list of 534 we had built manually. The list-building took 8 hours of solo work. The calls took roughly 16 hours over 3 weeks. The product-changing insights came from those hours.
The trap of your own enthusiasm
The hardest discipline is not technique. It is restraint.
You have spent months building this thing. You believe in it. You want to share it. Every minute of silence in an interview feels uncomfortable. The temptation to fill that silence with "we built X to solve Y" is real and it is the killer.
Sit on your hands. Let the prospect talk. The signal is in their reality, not in your pitch. The prospect who buys is the one whose past behavior matched the problem you are solving, not the one who said "yeah sounds great" on a call.
Your idea is the trap. The prospect's reality is the gold. Stop trading the second for the first.
What we now do every week
A current commitment: at least one customer-development call per week, every week, even when we are too busy. Especially when we are too busy. The week we skip is the week we drift.
We use the calls to:
- Validate or kill upcoming features.
- Discover edge cases we did not anticipate.
- Get referrals.
- Stay grounded in the customer's reality, which has a way of drifting in the founder's mind when you spend too long staring at code.
This discipline is part of the broader operating cadence of the studio that we discussed in our shipping four SaaS in parallel piece. Customer development is not a phase. It is a habit.
When the Mom Test breaks down
Three failure modes worth flagging:
- You can't get prospects on a call. Solving this is upstream of the Mom Test. If your cold outreach is converting at 1%, fix that first. We covered the relevance of customer outreach to the broader vertical SaaS strategy.
- The prospects are already customers and they tell you what you want to hear. Existing customers have a built-in bias to please you. Discovery calls are best done with strangers, ideally introduced through a third party.
- The product is so far from launch that there is no real "past" to anchor on. In this case, you anchor on the past behavior they already have without your product. "Walk me through how you do X today" is always a valid question.
What is next
We are formalizing our Mom Test playbook into a tiny internal doc with the questions we ask, the questions we do not, and the templates for follow-up. The discipline is to use it every week.
If you have an idea right now and you are debating whether to start interviewing prospects, the answer is yes, today, this afternoon if possible. Email three people. Ask for 30 minutes. Anchor on their past. Listen. Take notes. Send a thank-you. Repeat next week. The compounding effect of this practice over 6 months is the difference between products that ship to indifference and products that ship to a waiting list. We learned the hard way. You do not have to.
