Category: Making Money

  • 492 Visitors From Reddit, Zero Buyers: Count Channels by Customers

    492 Visitors From Reddit, Zero Buyers: Count Channels by Customers

    Two founders published their numbers a week apart, and both make the same mistake visible. Yusuke Matsuba’s landing-page tool had 492 visitors from Reddit and zero paying customers from that channel, while 1,539 visitors from X produced 20. Khoa Nguyen’s twelve desktop apps earned $10k in ten months, and $4.8k of it — 56% — came from the one app he actually marketed.

    The fix is not more traffic. It is counting buyers per channel instead of visitors per channel, and asking every new signup where they came from while there are still few enough answers to read.

    The short version

    • Traffic is not demand. Reddit sent 492 visitors and produced 61 signups and no buyers. X sent 1,539 visitors and produced 20 buyers.
    • Your analytics mislabels the channel that convinced people. Of the 49 people who said Reddit, only 19 had Reddit as their first recorded visit. 18 showed up as Direct and 8 as Google.
    • The app you market is the app that sells. Eleven of twelve apps had a landing page and a changelog and earned about $2.2k between them.
    • Shipping releases is not selling. Seventeen releases a month came with orders of 27, then 23, then 13. Two weeks off, and orders rose.
    • Add one question at signup and one table a month. Compare the answers of people who paid, not people who signed up.

    What 55 customers actually looked like

    LandingBoost, Matsuba’s landing-page tool, has 55 paying customers and $1,459 in revenue since launch. Every sale was $49 or less. He sorted those customers by their first recorded visit and got this:

    Channel Visitors Signups Paying Revenue
    X 1,539 — 20 $582
    Google 682 — 10 $218
    Reddit 492 61 0 $0
    Untraceable — — 16 —

    Reddit produced the second-largest audience, more signups than either paid channel, and no money. Those two paid channels account for $800 of the $1,459, and 16 customers could not be traced at all. That last number matters: if nearly three in ten buyers have no known source, your channel report is already fiction.

    Matsuba is careful about the sample — “the numbers are small, so this is my case, not a rule.” The shape shows up in almost every small-site analytics account I have seen, mine included.

    Your analytics cannot see the channel that closed the sale

    Since 25 July his signup form has asked one question: where did you hear about us. 146 of 324 people answered. Forty-nine said Reddit and none of them paid. Twenty-seven said X and three paid. Nine said Google and none paid.

    Now compare that with the analytics. Of the 49 who credited Reddit, only 19 actually arrived from Reddit first. Eighteen were logged as Direct. Eight were logged as Google. Someone reads a thread, closes the tab, searches the product name three days later, and the analytics credits the search — not the thread.

    The same distortion runs the other way, too. Before each payment, 25 of the 55 customers had visited from X at least once. Eleven had visited from Google. None had visited from Reddit.

    That is the argument for a question instead of a dashboard. Attribution tools record the last machine-readable touch; a person records what they remember being convinced by, which is what should decide next month’s spend.

    Shipping releases is not marketing

    The second post is a revenue split. StoicSoft, twelve paid desktop apps in ten months, just over $10k in revenue and about $8.6k after tax and refunds. 1DevTool earned $4.8k, or 56% of the total. Server Compass made $1.3k, bundles $1.5k, 1AIVault $434, 1MarketingTool $266, four apps together made $217, and four made nothing.

    Khoa Nguyen’s explanation is blunt: 1DevTool is the only app he ever marketed. The other eleven got a landing page and a changelog, then he went back to code. “Zero sales did not mean bad product. It meant never marketed.”

    The release numbers are worth sitting with. In August, September and October he shipped 17 releases of 1DevTool each month. Orders went 27, then 23, then 13. Then he travelled, shipped nothing for two weeks, and orders went up. Each release was pulled by 100 to 230 installs, which felt like progress and was not.

    Every launch decayed on its own schedule. Server Compass made $33 in its first month, peaked at $296 in April and was at $59 by September. 1AIVault did $138 in June, then $122, then $116, then $0 in September. 1MarketingTool was at zero in October. Nothing grew because nothing was being sold.

    Volume without intent is not an asset

    Across all his sites, Nguyen was getting about 900 Google landings a day. One third of that went to a free 3D model library on another site, which sold exactly one license, at a discount. His blog, eight posts a week, brought 6% of the traffic. People searching for free 3D models want free 3D models; they were never his buyers.

    This is also a bad month to judge anything from search volume alone. The November core update started on 10 November and still had no completion notice on Google’s status dashboard this morning, so the baseline is moving while you read the chart. I wrote about the timing and what to postpone on 13 November.

    For a tool site, the useful number is not sessions. It is sessions from queries where somebody already wanted the thing you charge for — smaller, narrower, more honest. The same logic sat behind the 88-page tool site I wrote about on 20 November: pages can be indexed and still never meet a buyer.

    The fix: one question and one table

    Both posts point at the same two habits. First, ask. Second, count buyers, not visitors. This is the per-channel version of the switch I made from traffic totals to completed actions yesterday: one number is a proxy, the other is a person doing something.

    Where did you hear about us? (one choice)
    [ ] X / Twitter
    [ ] Google search
    [ ] Reddit
    [ ] YouTube or TikTok
    [ ] A friend or colleague
    [ ] Other: __________ which post or thread? __________

    Make the last field free text. A fixed list tells you the platform; a sentence tells you the thread, the keyword, or the person who sent them. Matsuba’s own comment section landed on the same gap: his form could not distinguish his own Reddit posts from replies in someone else’s thread.

    Then keep one table, updated monthly. Analytics fills the first two columns; the buyer columns come from answers and checkout, not the dashboard.

    Channel Visitors Signups Buyers Revenue Notes
    X          
    Google          
    Reddit          
    Direct          
    Other / unknown          

    Leave a cell blank when you do not know. A blank is information; a guess makes the table useless. Nguyen learned the same lesson from the other side: in November his sales rose 1.8x with no release, and he could not explain it, because his checkout carried no referrer, his webhook saved nothing useful, and his server log kept 14 days. He added source tags to every buy link on 15 November, eight months after his first sale. It is a ten-line script. Do it before your first customer, not your hundred and fiftieth.

    What to do this week

    • Add the signup question today, with one free-text field. Do not wait for a redesign.
    • Email your last 20 buyers and ask the same question. With small numbers, a reply rate of even a third beats any dashboard.
    • Build the table with whatever you have, blanks allowed, and put one sentence beside each row explaining the change.
    • Pick one channel and market it for four weeks before adding a second. Nguyen’s four zero-revenue apps were not bad products; they were never sold.
    • Tag every buy link with a source parameter before your next launch.

    Neither founder needed more traffic. One needed to stop counting a channel that never paid, and the other needed to stop building apps he would not sell. The number to watch every month is the fourth column of that table: how many people paid, and from where. Everything else is a proxy that flatters you.

    Sources: Yusuke Matsuba’s channel breakdown (Indie Hackers, 11 November 2026) and Khoa Nguyen’s twelve-app retrospective (Indie Hackers, 18 November 2026).