Two weeks of optimising the wrong thing
For the first half of April I was the busiest version of myself. New automations into the CRM. A second Substack cadence I’d been meaning to test. Three tool experiments I’d been putting off for weeks. I kept a running list and crossed things off and felt productive in the way that always feels like progress at the time.
Friday of week two I checked the pipeline and it looked the same as the Friday before. A few more followers. One conversation that went cold. Otherwise flat.
I would have kept going for another two weeks if my coach hadn’t asked me one question on Saturday morning that I could not answer.
The question I could not answer
“Who exactly are you selling to?”
I started talking and within about ten seconds I realised I was going to hand-wave. What came out was “founders, Hong Kong, ops-ish, anyone serious about AI.” I could feel the shape of the person in my head but I couldn’t describe them in a sentence without hedging in three different directions.
He waited a beat and then said, “so you’ve been writing content all week for someone you can’t actually name.”
That was the moment. Not a grand insight. More like finally saying out loud something I’d been dodging for a while. Every DM I’d sent that month had been aimed at a silhouette. Every landing page used the word “operators” because it felt safe. The problem with “operators” is it can mean an ex-McKinsey RevOps lead at a Series C and it can mean a guy running a three-person agency out of his flat. Those two people don’t read the same things, don’t reply to the same messages, and don’t pay the same prices.
The constraint nobody wants to look at
There’s an idea from manufacturing that I’ve been reading about for months without actually using. It comes from a guy named Eli Goldratt and the short version is this: the throughput of any system is governed by its slowest step, and improving any other step changes nothing.
If welding is the bottleneck, hiring more painters does nothing. If painting is the bottleneck, buying a faster welder does nothing. The system only moves when you find the actual constraint and work on it directly. Everything else is motion that produces output without producing throughput.
I’d been buying faster painters all month. More content. More tools. More little automations. All of it was downstream of a problem that sat one layer higher than anything I was working on.
A fuzzy customer makes every message you write land in a slightly wrong place. Three things start happening at once:
The content doesn’t quite fit anyone, because it has to fit everyone. So nothing in it feels specific enough to make any one reader feel seen.
The DMs get polite replies but no calls, because the opener is general enough to be ignored politely.
The proposals read like templates, because that’s effectively what they are. They were written for a composite customer who doesn’t exist.
None of those problems get solved by writing more content or sending more DMs or polishing the proposal template. They get solved a layer up, by naming the actual customer.
The subtraction exercise
The fix was not to add anything new. The fix was to subtract.
I opened a fresh page on Sunday morning and listed every type of person I’d told myself I was selling to over the last six months. Then I went through the list with five filters and crossed off anyone who didn’t survive all five.
Industry. Not “B2B.” A specific vertical I understand well enough to describe the operational pain on a Tuesday morning.
Company size. A 40-person firm and a 400-person firm have completely different decision cycles. Pick one band and write to it.
Role. The thing a Head of Operations cares about is not the thing a CTO cares about. The pitch breaks if the role is wrong.
The actual pain. Not a generic “save time” or “automate workflows.” A specific friction the role feels every week, in language they would actually use.
The cost of doing nothing. A real number, in their currency, by a real date. Without this last one there’s no reason for them to move on a call.
By Wednesday afternoon I had something I could actually say in a single sentence without making it up on the spot. It was narrower than I wanted it to be. There were three or four people on the list I’d been pitching last month who didn’t survive the filters, and I had to accept I was never closing them anyway.
What changed by Thursday
By Thursday afternoon I’d had a conversation with someone who actually fit the new definition, and it felt different from the last dozen. I wasn’t trying to convince them of anything. I was describing their Tuesday morning back to them and they were nodding along. That happened twice in four days. It hadn’t happened at all in the two weeks before.
A week of better calls is not proof that the problem is solved. Two good conversations could be coincidence. What it did do, though, was make it obvious that the thing slowing me down had nothing to do with any of the work I was doing in week one or week two. The constraint was a layer up from all of it, and I’d been leaving it alone because naming the customer forces you to commit to one person and close the door on everyone else you were secretly still hoping would buy.
What I keep thinking about
The part I keep coming back to is how long I let this go.
Two weeks of work that felt productive because it produced output. Output is easy to mistake for motion. The constraint doesn’t care how much output you generate around it. It sits there quietly governing the whole system until you walk upstream and look at it directly.
I’m writing this down mostly so I don’t forget it the next time a new tool looks shiny. The next time I catch myself adding things to the top of the funnel and watching the bottom of it stay flat, the first question I want to ask is whether I can still describe my customer in a single sentence without making it up as I go. If I can’t, the tools and the content and the automations are not the problem.
The question I’ve been dodging is.
If you read this and felt a small flinch when I described the Saturday call, that’s probably the constraint in your own pipeline this week. Reply and tell me who you’re selling to. If it takes longer than a sentence, you already have your answer.