I used to think I knew what users wanted.
Then I launched Lucuma, our AI CV screening tool for recruiters. Within the first month, I had a hundred pieces of feedback sitting in my inbox, Slack DMs, and support tickets.
About 60% of it contradicted what I thought people needed.
This post is about what those first hundred users actually said, what surprised me, and the specific changes I made because of it.
The Feedback I Expected But Still Hurt
Early user feedback SaaS founders don't talk about: the part where people tell you your onboarding is confusing.
I thought Lucuma's setup was simple. Upload CVs, set your criteria, let the AI score them. Three steps.
Turns out three steps is still too many when someone doesn't understand why they need step two.
Users kept skipping the criteria setup. They wanted to jump straight to results. Then they'd email asking why the scores didn't match their expectations.
I added a single tooltip explaining that the AI needs to know what matters to you before it can screen accurately. Skipped setups dropped by half.
The lesson: users will always take the path of least resistance. If that path breaks the product, you need guard rails, not more features.
What Users Said They Wanted vs What They Actually Used
This one stung.
I spent two weeks building a feature that let recruiters export screening results to five different formats. CSV, Excel, JSON, PDF, plain text.
Usage after launch: 94% picked CSV. The rest picked Excel.
Nobody used JSON. Nobody used PDF. Nobody used plain text.
I had asked users what export formats they wanted in a survey. They said all of them. But when it came time to actually use the product, they defaulted to what they already knew.
I stripped out three of the five formats in the next release. Support requests about exports dropped to near zero because there were fewer options to get confused about.
The lesson: watch what users do, not just what they say they want. Early user feedback SaaS data is most useful when it's behavioural, not hypothetical.
The Feature Request That Appeared Twelve Times
Twelve different users asked for bulk actions.
Not in the same words. Some said "batch processing". Others said "select multiple CVs at once". One person just wrote "why can't I do more than one?"
But they all wanted the same thing: the ability to move multiple candidates through their pipeline without clicking each one individually.
I had deprioritised it because I thought recruiters would want to review every CV carefully. Turns out they do review carefully, but only after an initial filter. Bulk actions let them clear the obvious nos faster so they could spend more time on the maybes.
I built it. Time spent in the app per session went up by 40% because users weren't rage-clicking through obvious rejects anymore.
The lesson: if the same request shows up in different words, it's not a nice-to-have. It's a workflow gap.
The Complaint That Made Me Rethink Pricing
One user emailed: "Why is the Solo plan £99 when I only process 20 CVs a month?"
Fair question. Our pricing was based on features, not usage. Solo, Agency, Enterprise. Flat monthly rates.
I looked at the data. About 30% of Solo users were processing fewer than 50 CVs a month. They were paying for capacity they didn't use.
I didn't change the pricing structure, but I did add a 14-day free trial and made sure the sales page was clearer about who each plan was for. If you're screening 20 CVs a month, you probably don't need AI scoring. You need a spreadsheet.
The lesson: early user feedback SaaS pricing isn't just about what you charge. It's about whether people feel like they're getting value for what they pay. If they don't, they'll churn quietly.
What Nobody Told Me Directly But I Learned Anyway
Some feedback doesn't come through support tickets.
I noticed users were taking screenshots of their screening results and pasting them into emails instead of using the share link feature I'd built.
I asked one user why. She said the share link opened a new page and her hiring manager didn't want to click through. She just wanted to paste the results directly into an email thread.
I added a "Copy Results" button that formatted everything as plain text. Usage of the share link feature dropped, but users stopped screenshotting. The workflow got simpler.
The lesson: silent friction is still friction. Early user feedback SaaS insights come from watching where people work around your product, not just where they complain about it.
The Two Requests I Ignored On Purpose
Not all feedback is good feedback.
Two users asked me to add LinkedIn integration so Lucuma could pull candidate profiles automatically.
It sounded useful. But it would have taken three weeks to build, required ongoing API maintenance, and introduced a dependency on LinkedIn's rate limits.
I said no. Politely, with an explanation. Both users are still customers.
The lesson: early user feedback SaaS development is about making trade-offs. Some requests are feature creep disguised as user needs. Your job is to know the difference.
What I'd Do Differently Next Time
I'd start collecting feedback earlier.
I launched Lucuma with a closed beta of five users. I thought I needed to polish it first. Big mistake. Those five users would have told me about the onboarding confusion, the export format waste, and the bulk actions gap before I spent weeks building the wrong things.
Next time I'll get feedback at 60% done, not 90%.
I'd also track feedback in a single place from day one. I lost early insights because some feedback was in email, some in Slack, some in Typeform responses. I didn't start tagging and categorising until user fifty, which meant I had to go back and re-read everything.
What Comes Next
I'm applying these lessons to everything we build at Marvanova now. PostLift, dunlin Scout, SEN Letters UK. They all start with rough beta versions and real user feedback before we commit to features.
The first hundred pieces of feedback taught me more than any SaaS course or startup blog ever did.
If you're building something and you're not talking to users yet, start today. Not next week. Today.
Got questions about early user feedback or how we handle it at Marvanova? Email me at hello@marvanova.com.