Building was the easy part

I have five products live on the internet. Caddie Live, Society Caddie, Do This Next, Adrift and Buy Me a Tee. All of them were built in evenings and weekends around a full-time job, and all of them work.

That last sentence used to feel like the hard part. It isn't. It stopped being the hard part somewhere around the second one, and I have spent the time since then slowly working out what the hard part actually is.

#Building is the bit I already knew how to do

I have been making digital products professionally for more than twenty years. Agencies, in-house teams, freelance. Ecommerce platforms, creator tools, internal systems, interactive installations big enough to walk around inside.

So when I sit down on a Saturday with a clear idea, the thing gets made. That is not a boast, it is just what two decades of practice buys you. I know how to scope something down, I know which decisions can wait, and I know what "finished enough to ship" looks like.

What I did not have was twenty years of practice at the other thing.

#The pattern I kept not noticing

Here is the honest sequence for most of them:

1. Have the idea, usually late, usually convinced it is obvious
2. Build it, faster than expected
3. Ship it
4. Sit in complete silence
5. Slowly work out that nobody was ever going to just turn up

Step four is the one that got me. Not because it was surprising the first time. Because it kept happening, and I kept treating each instance as bad luck rather than as a pattern with a cause.

I spent twenty years learning to build things. Nobody taught me how to make people care.

#What I got wrong

I thought the product was the argument. If it was good and it solved a real problem, that seemed like it should be sufficient. It is not even close. Being right is the easy half. Getting someone to find out you are right is a separate job with its own skills, and I had none of them.

I confused launching with marketing. A launch is an event. Marketing is a habit. I did a lot of the first thing and almost none of the second, then wondered why the graph went up for two days and then flattened.

I optimised the part I enjoyed. Given a free evening and a choice between writing a feature and writing about the product, I wrote the feature every single time. Adding another feature is the most productive-feeling way imaginable to avoid the actual problem.

#The numbers

Caddie Live went live in early June, so these are three-month numbers rather than a verdict. I had been building it since December, though. It started as an AI rules checker, became a live scoring app, and turned into what it is now somewhere in between. Rebuilding it twice was, of course, the easy part.

Six months building it. Three months learning how to get it in front of people. On the evidence, I have that ratio the wrong way round.

I spent €25 promoting Caddie Live on Instagram. It was seen 1,893 times and reached 1,514 people. Twenty-three of them clicked through to the site. Two signed up.

Two is not a channel. Two is an anecdote. And that is the actual lesson from the €25, which is not the one I expected: it was not enough money to learn anything. For a while afterwards I thought I had tested paid acquisition and found it wanting. I had not tested anything. I had bought roughly nineteen hundred impressions and a rounding error, then drawn a conclusion from it, which is a worse mistake than not running the ad at all.

None of those numbers are particularly impressive. Some of them are not even large enough to be meaningful. That's rather the point.

#The uncomfortable bit

I can account for about a dozen of the people who found Caddie Live.

Two came from €25 of Instagram ads. Eight or ten came from people I know. That is the end of the list. Everyone else arrived on their own, which sounds encouraging until you realise I have no idea how.

There are rounds on there from courses in several countries, so they are plainly not all friends of mine. Beyond that I could not tell you much.

Row after row in my dashboard's acquisition panel says "not recorded". The biggest country in the users-by-country panel is "Unknown". Thirty-six people have downloaded the iOS app and I cannot connect those downloads cleanly back to sign-ups. One person signed up to test the Android build, which I am choosing to read as a rounding error rather than a market signal.

The dozen I can explain did teach me one thing worth keeping. Knowing you does not make someone interested in your side project. A stranger looking for a way to score a society day has already decided it might be useful. A friend has to first decide whether the thing you made has anything to do with them. Your own network is the first place everyone reaches and the last place that behaves like distribution.

I spent twenty years learning to instrument products. Then I shipped my own and forgot to instrument the way people arrived.

That is probably the most useful thing the numbers have told me.

#The product I built about it

Eventually I did what I always do when a problem annoys me: I built something. Do This Next hands you the single marketing action worth doing next. Building it forced me to admit that at any given moment there is usually one obvious next action, and that the reason I was not doing it was never that I did not know what it was. It was that "market the product" is not an action. Uncomfortable to discover about yourself via a product you wrote for other people.

#Where this leaves me

Being able to build quickly used to feel like the advantage. I am no longer sure it is. The cost of making a plausible version of almost anything has collapsed. The scarce part has moved downstream: attention, trust and distribution.

So the plan for the next while is deliberately unglamorous. Fewer features. More conversations. Write regularly rather than in bursts. Pick one channel per product instead of a thin presence on five.

And close a gap in my own measurement that I had not properly admitted to. I can tell you what my users do in some detail, and I look at that every day, out of habit as much as diligence, because it is what I do for a living. What I cannot tell you, for most of them, is how they found me. I instrumented the product thoroughly and the way in not at all, which is a strange thing to have done for someone who thinks data is most of the job.

I will report back. Including the parts that do not work, which on current form will be most of them.