steve blank, get out of the building, talk to customers, customer interviews, the mom test, explorative experiments…. customer interviews are a crucial part of the early-stage venture

confronting your hypotheses to the real world, to the ‘real’ customer

i’ve always found the process fascinating, yet time-consuming and sometimes frustrating (sometimes you really really wanna ask follow-up questions days/weeks later, we’ll try to solve that partially)
in this playbook i’ve been using, i want to share some tips, tools and prompts on getting bigger and faster bang for your buck. it starts out pretty simple… and then gets interested.

disclaimer: nothing revolutionary here, there’s probably someone out there much smarter doing this in a better way, and it’s of course not one-size-fits all

let’s get into it

1. select 2-3 ideal customer profiles (icps)

who’s your ideal customer? early adopter?
who has the pain, knows they have the pain, will pay to solve the pain, will switch to your solution to solve the pain and isn’t happy with their current solution?
stop trying to talk to everyone. get specific.
b2c or b2b? socio-demographic traits, psychographic traits

get as specific as possible

already lost? check this link out and keep diving deeper

2. create a structured interview (5-whys, jtbd, mom test)

i got lazy, i simply prompted it with claude and chatgpt, i preferred chatgpt’s output

want to build your own interview? use this prompt:

prompt:
I have a [business idea or product concept], but I’m not completely certain about my value proposition. I want to talk to 5–10 potential customers to pinpoint a real, recurring problem they face and see if my [idea/concept] could solve it.

I want a compact, bullet-point style interview flow that references The Mom Test, the 5 Whys, and Jobs To Be Done frameworks. I do have a rough demo or concept to show if needed. Please make it straightforward and quick to follow. The flow should cover:

- Warm-up questions (role, context, prior experiences)
- Past behavior / context (last time they faced the problem)
- Pain exploration (primary challenges, motivations, using the 5 Whys)
- Existing solutions & gaps (what they do now and why it’s not perfect)
- Desired outcome (their “magic wand” solution)
- (Optional) Quick demo or explanation of my concept (collect immediate reactions)
- Wrap-up (urgency, willingness to pay, who else I should talk to)

Please include tips on avoiding bias and on pulling out concrete, real-world examples (per The Mom Test), plus how to frame questions around Jobs To Be Done. Keep the final answer under 250 words, short and direct.

here’s what mine looks like:

3. 🔥 run synthetic interviews (before real ones)

i’ve talked about synthetic interviews and synthetic data before in this playbook
long story short: synthetic data and customer interviews are created using ai, they replicate human interactions. they are artificially generated data, digital clones of people, so you’re basically talking to digital twins of your ideal customers

how to run these? i’ll recommend 3 tools here from lo-fi/low cost to higher-fi/higher cost

1. cheap and fast options = this ‘user persona generator’ (it’s a public customgpt), it’s surprisingly good

2. slightly more expensive but much richer results = syntheticusers

3. bonus = get feedback on your landing page and/or website before you go live using wevo

   here are some results of the first 2 merged together (check out the quotes)

you can also run follow up questions… get some deeper answers to key questions

(💡note…you’re already building up your pitch and/or landing page here without even knowing it):

So tell me…what does the ideal solution look like?

how much would you pay?

what would convince you to pay?

what would block you from paying or switching?

tip: do try out syntheticusers and their free 7-day trial for much deeper insights and back and forth (i’m not affiliated)

4. real interviews (part 1): find icps online

i find that synthetic interviews alone don’t do the trick. they lack a few tiny nuances, sentences and secrets/details (although you’ll be amazed at how much their findings will correlate with real interviews). therefore it’s good to cross-reference and validate with real interviews

to do that, you need to find your icp and interview them.
this playbook isn’t about customer acquisition channels (i’ll write about this later) but i will give you some angles.
by the way….

so good to start testing it now.
if you can’t find real people to even do an interview with you…you’ll struggle to ever convert anyone to your solution.

/

your network: i’m hoping you have a bit of what they call founder-market fit which will allow you to have a bit of credibility/authority in the space, as well as a network to tap into.
if you’re still struggling, i suggest this tool happenstance (don’t ask me about data privacy pls) where you basically connect your gmail + linkedin + twitter

and can then freely search'/navigate your online network using natural language

here’s a preview of results:

i’m paying for a month or two of this while i do my customer research

if you liked this and are focusing on b2b or b2c2b customers, you’ll also like juicebox which makes boolean searches on linkedin sooooo much easier

results

and edit filters as you see fit

this is in my case…but you might be targeting a b2c persona or a very specific persona that can’t be found on linkedin

so for those i’ll let you take the next steps…here’s a bit of inspiration

→ youtube and linkedin are the most gender-balanced, so versatile for everything from tutorials to thought leadership.
→ pinterest is 74% women. where ideas become purchases: home decor, fashion, weddings, lifestyle brands thrive here.
→ reddit is 57% men. communities run deep, and passion drives engagement. think gaming, tech, finance, niche subcultures (my fav = r/breadstapledtotrees). also reddit strong for seo/geo purposed
→ tiktok and instagram lean slightly female. both platforms are perfect for capturing attention with aspirational or entertaining content.
→ x (formerly twitter) and reddit skews male, fast-paced discourse, bold ideas, politics and specific interests.

if all else fails…stick this in perplexity 

prompt: i’m looking for [insert persona type], where and which platforms can i find them on? act like a market research expert, be concise, gimme links pls

4. real interviews (part 2): record & centralize notes

this part is pretty simple but crucial, make sure you’re recording the interviews, we need the exact words

tips on tools to use?
granola: it’s a stealth notetaker granola is a little meeting notes companion that runs in the background (but of course disclose that you’re recording! no reason not to)
fireflies and tl;dv (most commonly used) both record and transcribe sessions, with tl;dv also hosted in the eu for stricter compliance. those 2 have the best multi-language. meetgeek also stands out if you need multiple language options or want easy export. explore and select at least one of these.

i don’t know how we ever lived without these?

that’s it… now go interview people… see you in a week

5. centralize

once that is done, centralize it all and store the notes and the exact transcripts (both synthetic and real)

6. time to ‘rag’ it up

here comes the fun part

i was originally doing this using mindstudio and also looked into haystack agents… then realized i was bringing a toothpick to a gun fight and that i was missing a key feature…so i started using notebooklm instead

notebooklm is a tool from google you’re probably heard about. it’s absolute gold. i can’t live without it. it allows you to synthsize documents, get briefings, ask questions and 🍒cherry on top of it all… create audio podcast episodes based on your source information

(disclaimer: there are still some hallucinations, as with any llm… be careful)

so load up all your notes and transcripts carefully named by source persona and type of persona and add them as sources into notebooklm (note you’re limited to 50)

i’ve named them by type and persona (note that you can select/deselect sources if for example you want to compare synthethic results with real interviews)

what now? start querying the data, uncovering trends, ask questions, get all the insights you want on your riskiest assumptions

tip: here’s a list of risky assumptions you might want to validate/invalidate

desirability
- customers recognize the problem: assuming that the market acknowledges the problem you’re solving as significant and pressing.
- willingness to pay: believing customers are willing to pay for the solution without solid validation.
- value proposition clarity: assuming the value proposition is clearly understood and compelling to the target audience without feedback.
- brand loyalty transfer: expecting that customers loyal to current solutions will easily switch to a new product.
- market size overestimation: overestimating the accessible market size without considering realistic market penetration rates.

feasibility
- technology readiness: assuming existing technology can be adapted to your needs without substantial development.
- scalability assumption: believing the solution can be scaled easily without significant adjustments or resource increases.
- regulatory navigation: underestimating the complexity and time required to navigate regulatory environments.
- resource availability: assuming that the necessary resources, both human and material, are readily available and affordable.
- timeline realism: setting unrealistic timelines for development and launch based on optimistic scenarios.

viability
- cost estimation accuracy: assuming the initial cost estimates will hold true without considering potential overruns.
- revenue projections: projecting future revenues without accounting for market dynamics and competitive responses.
- partnership dependencies: relying heavily on partnerships or third-party collaborations which may not materialize as planned.
- customer retention: assuming that customer acquisition automatically leads to retention without strategies to engage and satisfy.
- tech adoption curve: presuming a rapid adoption by users without considering learning curves or resistance to new technology.

from here you can

→ chat with the insights
→ find exact phrases that matter

→ get briefing documents you can easily share around with teammates

→ 🔥 turn findings into podcast episodes (great to share with team)

probably my favourite feature, especially for shareability, you can turn your findings into an actual podcast episode. i can’t tell you how useful this is once you want to share your findings with your colleagues (who maybe don’t understand the importance of customer interviews)

great thing is you can also customize the episode to your needs

small tip… start to extract the most powerful customer sentences, that we’ll then use for our validation experiments (landing pages and outreach campaigns)

7. rank risky assumptions & design experiments

the rest is textbook validation (which text book you might ask? this one)

rank your riskiest assumptions on a 2 by 2 matrix by how critical and/or how unknown the assumptions are

and start with the top right ↗️

not sure what experiments to run? here are some ideas?

smoke test: put up a basic page or ad that pretends the product exists, track clicks or sign-ups

trigger test: push a short prompt or email or ad campaign that asks people to act now, measure immediate responses

landing page test: build a quick page about the offer, see if visitors give an email or request more details

dry wallet test: list a product with a “buy” button, see if people attempt to pay even though you don’t charge

concierge mvp: manually perform all tasks behind the scenes, gather feedback before automating

wizard of oz mvp: create a front-end that looks automated but is run by humans in the back-end

fake door experiment: offer a feature link in your product that doesn’t exist yet, see how many click

pre-sell campaign: announce and allow pre-orders for something not yet built, gauge buying intent

social media ad test: run a quick ad that highlights your idea, measure click-through or sign-up rates

paper prototype test: sketch the solution on paper, watch how users interact and react

clickable mockup test: create a simple clickable design, let people navigate and gather early impressions

explainer video pilot: produce a short video describing the product, track views and sign-ups for interest

pop-up store pilot: set up a temporary physical booth or store, watch foot traffic and purchasing behavior

hope you liked this!

next time we might get into prototyping and validation 🙂 

⬛️

ps: some previous and relevant playbooks/posts you might have missed