6 months, 46K queries, 10 power users, $0 marketing spend

The difference between the use cases you expected and what those 10 power users actually built is pretty striking. Curious which one surprised you the most.

The 10 power users is the number I'd frame. 46K queries reads like traction,
but it's volume. Ten people who built things you never designed - a tax-aware
backtester, a compliance screener - that's the product telling you what it
actually is.

I see the same shape from my side. I build a desktop finance tool for
freelancers, different world, same pattern. The users who stick are the ones
bending the tool toward their own process. The ones asking for the generic
version leave, because the generic version is everywhere.

And your Twitter numbers match my experience almost to the digit. Months of
posting, a handful of signups. Meanwhile the long-tail pages I almost didn't
write bring most of the traffic now. "Best MCP server stock data" and
"offline invoicing software no subscription" are the same species of query:
nobody shares them, everybody searches them. Boring keywords with intent beat
interesting keywords without it.

I run free browser tools with no login too, so I’d push back a little on the idea that you need auth before you can measure this properly.

A session-scoped ID plus one well-chosen completion event gets you surprisingly far. The metric that actually changed my decisions wasn’t visits, it was the percentage of sessions that completed a full run of the tool. That separates people who came to actually do the job from people who just bounced in from a search result.

For your case, something like “sessions with 3+ queries” might be the equivalent. You can measure that today, using the traffic you already have, without putting a login wall in front of an SEO channel that’s clearly working.

Can you elaborate on the SEO part? How did you find those long-tail queries, and how do you format your blog posts?

The SEO results are the part that stood out most to me — slow at first, but once it started working it became the only channel that kept compounding. Also interesting that the real use cases ended up being much more advanced than you originally planned for. Great example of letting users show you what the product should become.

Disclosure: I run a lead-list tool (LeadGrid) with the same shape of problem, so read this as a self-interested view.

The 46,000 is a volume number with no denominator. I count searches server-side keyed by the query, which means my dashboard counts queries and not people - one person running twenty variations of the same search shows up as twenty of something. Worth knowing which of those yours is before the number gets quoted back at you.

On adding auth to measure properly: it will measure the population that agrees to sign up, which is not the population you have now. So it doesn't hand you the denominator for the 46,000 retroactively, and it quietly breaks the comparison going forward. If the denominator is the actual goal, a session or fingerprint estimate over the traffic you already have gets you closer and costs you nothing.

The other thing I would protect is the Bing Copilot path. Someone arriving mid-question from an AI answer is the least patient visitor you have, and that channel is the one you said compounds.

On the free tier cap decision, one thing that helped us before landing on a number: cost per query, not call count. We do local speech-to-text plus an LLM cleanup pass per request, and the LLM step dominates cost by a wide margin, so a flat N-free-calls cap would have been badly mispriced against actual compute; a heavier request can cost 10-20x a simple one. We ended up bucketing free-tier limits by estimated compute cost per request type instead of one flat count, which also made it much clearer where the paid line should sit. Also, on the auth-friction worry: we saw better conversion gating the signup behind a few free actions people could take with zero signup, rather than an upfront wall, so they experience the real thing before we ask for an email.

The 'none of these were use cases I designed for' line is the real insight here; that's usually the signal a product's actually found product-market fit vs. just working as intended. Curious about the 800 tool calls/day with only 10 power users; that's a huge amount of usage concentrated in a tiny group. Are you seeing any of them approaching you directly for a paid/higher-limit tier, or is it staying purely self-serve so far?

10 power users out of 46K queries is actually a stronger signal than it looks. Power users are the seed of the business — the question is whether you can find the pattern that makes them power users.

The highest-leverage move at this stage isn't marketing spend; it's interviewing those 10 people individually. What were they trying to do the first time they came back? What did they try before you? What would make them pay? Ten conversations will tell you which niche to double down on, and that niche becomes the targeting for whatever marketing you eventually do.

I see the same dynamic in the course/creator space (I run iLoquio): the creators who succeed almost never start with broad marketing. They find the 10 people who desperately need the thing, serve them unusually well, and let that density produce the case studies and word-of-mouth that make later marketing work.

Also worth noting: $0 marketing spend with real usage means your unit economics are currently clean. Whatever you add next, protect that — paid acquisition before you know why power users stay tends to just buy churn faster.

This is a great breakdown of the gap between measurement assumptions and real user behavior. You expected basic stock screening but users built 25-year tax-aware backtesting engines, insider-cluster detection systems, compliance screeners. That's the classic blind spot - your measurement system (what you thought users needed) was completely wrong, but usage patterns revealed what they actually valued.

The other insight that stands out: SEO compounds, everything else decayed to zero. Twitter had zero conversions despite 120 tweets. Substack outreach got polite replies, zero retention. But search traffic keeps growing. That's the difference between building on owned channels vs rented attention - distribution channels determine which feedback signals actually reach you.