I answered the phone. That was the problem.

The client was calling from Heathrow.

I don't remember what he said first, only that I was elbow-deep in a data migration, moving every user off our first product and into the new one we'd just built to replace it. A batch had failed for the third time on a profile I couldn't reproduce locally, and I'd finally sat down with every query pinned, every log tail open, ready to actually solve it. I picked up. Of course I picked up. He was stuck at the border. His lawyer wasn't answering. He needed to know if the document he had on him was going to work.

I don't remember what I said back, either. I remember I helped, and I remember shutting the laptop when the call ended because there was no getting back into that migration shit that afternoon. I opened it again the next morning and had to rebuild the whole thing in my head from scratch. That took me another hour of just sitting there rereading my own code, trying to remember what I'd been doing.

That call was correct to answer. Every single call like it was correct to answer. That's the trap I want to talk about.

We build software for law firms. Our customers' customers (the actual people trying to hire a lawyer or already working with one) call in a lot. Most of the calls are what you'd expect. When is my lawyer going to call me back? Am I eligible for this visa? Can I still qualify if my situation is X? A smaller number of them are like the Heathrow one: someone in the middle of a real problem, right now, who needs somebody to talk to in the next four minutes.

The problem, and I mean this as a description, not a complaint, is that from a ringing phone you can't tell which is which. So you pick up all of them.

Initially, we were the ones picking up. Me. My co-founder. Whoever was closest to a laptop, we got good at it. That was part of the problem too. We were fast, accurate, knew the product cold, and could answer eligibility questions better than most paralegals. Every rational analysis of any individual call said we were the right people to handle it.

I sat down one weekend and actually did the math, mostly because I was annoyed at how little I'd shipped that month. Somewhere around 35% of our working time was going to intake and the tail of intake: the follow-up messages, the "let me get back to you," the second call because the first one didn't answer their real question, the recovery time after each one where you sit and stare at your terminal for eleven minutes before you can write a line of code that isn't wrong. A third of our week. Gone.

A typical week looked like this. Monday I'd sit down with a plan: this week the migration would finally finish, or we'd ship the feature we'd been promising firms for the last three demos. By Wednesday afternoon the plan was in the bin. I'd spent Monday and Tuesday between calls, and the two blocks of building time I'd carved out had both been eaten by a customer question I'd sworn I'd answer "in five minutes." Thursday I'd rebuild the plan, angrier. Friday I'd ship maybe half of what I said I would, and I'd start Monday already behind.

The stuff we weren't shipping while that was happening was our outreach product and quality improvements to our products. I mention those two specifically because they're the ones that most directly move revenue for the firms we serve, which means they're the ones that most directly move revenue for us. The intake work wasn't just eating our time in the abstract. It was eating the exact hours that would have paid for the intake work five times over if we'd put them somewhere else.

I know how this sounds. Every founder eventually writes some version of this, about being too busy doing admin to build the important thing. I hesitated to add mine to the pile. But I think the standard version of it misses what actually makes the trap hard, which is the thing I said above: from a ringing phone you cannot tell which one is the airport call and which one is the fourth eligibility question of the day. So you can't triage. You can't say "I'll only take the important ones." There's no way to know without picking up. And once you've picked up, you've paid the interrupt cost whether the call was important or not.

That's what I want other founders to actually take away from this. Not "delegate more." Everyone knows that. What I want you to see is that some kinds of work are structurally incompatible with the thing you're supposed to be doing, and the incompatibility isn't about the hours; it's about the shape. An hour of intake doesn't cost you an hour of building. It costs you an hour of intake, plus however long it takes to reload the entire context of what you were doing back into your head: every open file, every terminal tab, the seven small things about how a bit of code works that you have to remember for the next line to make sense. For me that was usually longer than the call itself.

There's real research behind this. Gloria Mark at UC Irvine has spent decades measuring what it costs to get interrupted at work, and her number is roughly 23 minutes to fully refocus after being pulled away. That's the median. Mine felt longer. It usually was.

Two calls in a morning meant no product work in that morning. Not less. None.

What we did

I'll be quick because it isn't the point. We built our own intake. It handles the questions we'd already answered a hundred times ("when will my lawyer call," eligibility triage, status updates), because we were the ones who taught it. The stuff that actually needs a human (someone at the border, a lawyer locked out of their account at 11 pm, anything with a real deadline) routes straight to us, without us sitting through the fourteen other calls that came in the same hour.

I think this part of the story is boring on purpose. Every founder's version of this fix is different. Some of you should build. Some should buy a tool that already exists. Some should hire the person you've been putting off hiring because you kept telling yourself you could handle it a little longer. What you build or buy is much less interesting than the decision to stop being the person who picks up.

The 35% didn't come back to us as free time. That's not how it works, and I don't want to pretend otherwise. What came back was the shape of our weeks. Mornings became long enough to actually finish something. I stopped closing the laptop after every call. Not because the calls stopped, but because when they came in, I knew someone would handle the routine ones and I'd only see the actual emergencies. That's a different kind of interruption. You can plan around it.

We shipped the outreach flow we'd been trying to ship for months. We started actually improving other products: the kind of work that needs you sitting with a dataset for two hours noticing patterns, which is not something you can do in the four-minute gaps between calls. Honestly, it was a bit embarrassing. Like, oh. So this is what having a normal amount of building time feels like.

The part I want you to sit with

If you're a founder reading this, you probably already know which category of work is your version of intake. It's the thing you've been doing yourself for too long because you're good at it and nobody else can do it as well, and each instance takes twelve minutes, so it never feels worth the meeting to figure out how to hand it off. It's the thing that's on your calendar in tiny slivers between the real work and eating half of the real work anyway because of the interrupt tax.

You know what it is. I don't have to guess.

The reason it's hard to fix isn't that you don't know what to do about it. It's that fixing it means taking a week or a month of building capacity and spending it on the fix, and the fix doesn't look like product, and it doesn't feel like shipping, and there's a voice in your head (I had this voice for months) that says we'll get to it next quarter; the calls aren't that bad this week.

They will be that bad next week. They will be that bad the quarter after. The math will not improve on its own. Nothing about your growth is going to make intake shrink relative to your building time; if anything, more customers means more calls, so the tax grows with the business until you decide to cut it.

I'm not going to tell you what to build or buy or hire, because I don't know your situation and I've read too many founder posts that pretend the author's specific answer is anyone else's answer. What I'll tell you is that if you're doing this work yourself and you can't remember the last morning you had three uninterrupted hours in a row, the fix is overdue. Not "later this year" overdue. Now.

I wish we'd done it a year earlier. That's the whole thing. If it saves one of you a year, it was worth writing down.