How do you actually learn from your existing customers? (And I'm experimenting with something different to find out)"
Hello all I'm Vanessa a founder based in New Zealand, currently running a few startups, both at early stages.
A while back I was stuck on a problem I suspect a lot of founders here have hit.
We had customers. The product was working. But we were completely in the dark about where to find more of them. Our top of funnel was broken and we didn't know how to fix it — because we didn't really understand our customers' journey before they found us.
The issue wasn't that we had no feedback. We had fragments collected at different points along the way. The problem was it was scattered, inconsistent, and impossible to draw real conclusions from. Nothing that could tell us with any confidence: this is your person, this is where they were, this is what finally pushed them to go looking for a solution.
Surveys didn't cut it. Response rates were dismal and the answers were shallow. Scheduling calls took forever. And even when we did get responses, turning them into something we could actually act on felt like another full time job.
I've been researching this problem properly before building anything — and I want to hear from as many founders as possible first.
A few things I'm genuinely curious about:
- How do you currently collect meaningful feedback from existing customers?
- What are your honest response rates like — and what do you think causes that?
- When you do get feedback, how do you turn it into actual decisions?
- What's the most broken part of the whole process for you?
I'm building something early stage around this — a lightweight tool to help founders collect and make sense of customer feedback without the usual friction, but I want to make sure I'm solving a real problem before I go any further.
Here's where it gets a bit different — I'm experimenting with how I collect responses.
Instead of just asking you to type in the comments, I'm also offering the option to answer a few questions by voice, asynchronously, in your own time. No scheduling, no pressure — just hit record and talk. Your response gets transcribed instantly and your voice is never stored. It's as easy as leaving a voicemail but completely private.
Answer by voice here: app.speakit.tech/questionnaire
Alternatively as always drop your thoughts below.
Whichever feels easiest. I'll share everything I learn with the community once I've spoken to enough people.
I know a few early-stage founders who have existing customers and are struggling to collect meaningful, actionable customer feedback. They'd probably be willing to answer any questions you have if you want.
This hits — “feedback fragments but no clear picture” is exactly the problem.
I’ve seen behavior signals > surveys in most cases. Turning feedback into action is the hardest part.
Async voice is interesting — curious about response rates.
Also, if you’re exploring ideas — $19 puts it in real competition. Tokyo trip + $500 min.
Round just opened: tokyolore.com 🚀
Really glad that framing landed! it's the thing I keep hearing across every conversation I have with founders so it felt important to name it directly.
The behaviour signals point is sharp. You're right that what people do tells you more than what they say they do, and the best feedback probably combines both layers. Quantitative signals tell you where to look, qualitative conversation tells you why it's happening.
On turning feedback into action being the hardest part, that's actually the thing I'm most focused on. Most tools stop at collecting and organising. The gap I keep coming back to is what happens after? how do you go from a pile of transcripts and responses to knowing what to actually do next. That's the part that still feels really manual and painful for most founders I've spoken to.
On voice response rates; I don't have enough data yet to give you a meaningful answer, which is honestly part of what I'm trying to find out. If you're curious enough to try it yourself I'd love to know how it feels compared to typing — app.speakit.tech/questionnaire. Your reaction to the experience would be as valuable as the answers themselves.
Totally agree — “what happens after” is the real gap. Collecting feedback is easy, deciding what to do next is the hard part.
I’ll try the voice flow, sounds interesting. Curious to see how it compares to typing.
We’re actually exploring something similar around turning raw feedback into clear next actions — happy to share what we’re seeing if useful!
Yes! Would love to hear what’s working for you in that space as well. What’s your product?
Love that — happy to share!
We’re building Tokyolore — it’s a competition where founders submit ideas, get an AI-powered business breakdown, and compete for a growing prize pool (winner gets a Tokyo trip).
The idea is to surface real signals early — not just what people say, but what they’re willing to commit to. Still early, but seeing some interesting behavior already.
Happy to share more if you’re curious!
The problem framing here is sharp. Most feedback tools solve “how do I collect more responses” when the actual problem is “how do I make sense of what I already have.” The fragmentation across NPS, support tickets, onboarding surveys, and churn interviews that watsonfoglift described is something I’ve seen at every early-stage company I’ve worked with. Nobody has a single view of the customer journey. The async voice angle is an interesting wedge. Curious about one thing: when you transcribe the voice responses, are you planning to just hand over raw text, or will there be some AI layer that extracts themes and clusters similar feedback automatically? Because the collection side is only half the problem. The synthesis side is where most founders give up.
This is the sharpest framing of the problem I've heard yet and you've articulated something I've been circling around but hadn't quite nailed. The real problem isn't collection, it's synthesis. That's exactly it.
To answer your question directly, yes, the plan is absolutely an AI synthesis layer, not just raw transcripts. Raw text handed back to a founder is just a different flavour of the same problem. The value is in what you extract from it: themes, patterns, clusters of similar feedback, and critically, what you should actually do next based on what you're hearing. That last part, the action plan output, is the thing I'm most focused on getting right. Most tools stop at organising. I want Speak It to tell you what to do.
The reason I'm validating before building is partly because of exactly the question you're asking. The synthesis side is hard to get right and I want to understand what founders actually need from that layer before I commit to building it a particular way.
Given you've seen this problem at multiple early stage companies I'd genuinely love to pick your brain on what good synthesis output would look like for you. Would you be up for a quick 15 min conversation? Happy to share everything I'm learning in return.
The scattered fragments problem you describe is the real bottleneck, and most founders I've talked to hit the exact same wall. You end up with NPS scores, support tickets, onboarding surveys, and churn interviews in five different tools, and none of them connect into a coherent picture of who your best customers are or how they found you.
One thing that's worked surprisingly well for me is dogfooding -- literally using my own product the way a customer would, every week, and logging what's frustrating vs. what clicks. It sounds obvious, but most founders stop doing this once they have real users. The gap between what you think the experience is and what it actually is widens faster than you'd expect. And the insights you get from being your own power user are qualitatively different from what surveys surface -- surveys tell you what people think they want, but using the product tells you what's actually broken in the workflow.
On the response rate problem: I've noticed the biggest predictor isn't the format (survey vs. call vs. voice memo) -- it's timing. Catching someone within 10 minutes of a meaningful action in your product (their first aha moment, a failed export, a completed project) gets 5-10x the response rate vs. a generic how's it going email sent on a Thursday afternoon. The emotional context is still fresh.
The async voice approach is interesting though -- there's a real subset of people who'll talk for 3 minutes into a mic but won't type two sentences. Curious what your response rates look like compared to the text option once you have enough data.
Thank you so much for this — genuinely one of the most useful responses I could have hoped for.
The timing insight is something I hadn't fully considered and it's stuck with me since I read it. The idea that emotional context is still fresh within 10 minutes of a meaningful action makes complete sense — and it's a very different problem to solve than what I've been focused on. That's exactly the kind of thing I wouldn't have surfaced from a survey.
The dogfooding point also landed. You're right that most founders drift away from it once real users show up. The gap between perceived and actual experience widening faster than you'd expect is a really honest observation.
Curious — when you talk about the fragmentation problem across five different tools, how are you currently dealing with it? Are you manually pulling things together somewhere, or have you mostly accepted that the picture is always going to be incomplete?
On the voice vs text question — I don't have enough data yet to give you a meaningful answer but that's exactly what I'm hoping to learn. If you're ever curious enough to try the voice option I'd genuinely love to know which felt easier for you personally — app.speakit.tech/questionnaire — no obligation at all.