Where did the hours go?

Last week I named the three stages of the trap that catches almost every advisory practice that figures out delivery. This week I want to stay with stage one – the capacity erosion – and walk through what we did before we did anything else, which was figure out where the problem actually was.

There is no magic bullet for this. The industry around us is selling plenty of them, and a fair amount of this newsletter exists to push back on that. The thing that actually moved the needle for us was real-world visibility into where our time and energy and attention were going – the kind of visibility we could act on, not the rough sense we'd describe to a colleague over coffee. Everything else had to start from there.

The breaking point wasn't a moment. It was a feeling that deepened over months. The volume of preparation work kept growing. The time available to deliver real value kept shrinking against it. The future I could see, if I extrapolated from where the curve was pointing, was one of grind rather than satisfaction. The present usually felt fine. It was the trajectory that was wrong. We were further along the curve than we'd articulated to ourselves, and we were about to reach for the kind of fix that wouldn't actually fix it.

Time tracking was the first move, and neither of us wanted to do it. In our corporate lives we'd both railed against the idea, especially given how much of the work we now did was fixed-price – tracking hours felt like a thing the wrong kind of practice did, for the wrong kind of reason. The reality is that habit breeds discipline, or the lack of it. What you measure is what you focus on. If you can't quantify a problem, you can't hope to fix it. So we tracked.

The first thing it surfaced wasn't about the work. It was about us. The instinct, when you start blocking time, is to block every available hour. To stop being human about it. To ignore the natural rhythms of energy that everyone has and pretend you don't have them, because acknowledging them feels like falling further behind. So we'd plan a day with no slack in it, and then add pressure when the day didn't survive contact with reality.

The variance was the second thing. I gave myself an hour for this. I'm now two hours in and no closer to the answer. Not on one task. Not on one client. Across multiple tasks and multiple clients in the same week, repeating. The plan wasn't the problem. The work underneath the plan was. Or sometimes, as it turned out, we were.

What time tracking didn't tell us was when in the day each task was landing, or whether the time was actually ours when we were spending it. Both of those turned out to matter at least as much as the hours themselves.

The same task, on different days, in different parts of the day, took wildly different amounts of time. Reading a P&L well, framing the right question for a client conversation, writing commentary that turns numbers into a recommendation – this is hard cognitive work. It has a time-of-day signature, and we'd been doing it whenever the calendar happened to leave a gap, which is the worst possible scheduling rule.

There's a study I keep coming back to, on parole decisions made by Israeli judges. Sixty-five percent of cases were granted parole first thing in the morning and just after meal breaks. Just before each break, the rate dropped close to zero. The researchers thought it was decision fatigue and that's been argued back and forth ever since, but the broader point holds: humans don't bring the same quality of attention to a hard problem at every hour of the day, and the variation is large. We found this out fast every time we tried to write client commentary at three in the afternoon. The science was just naming what we were already living.

The slogan version of the prescription is do hard work in the morning, and that wasn't the lesson we took. Kristina and I are wired differently, and our actual windows aren't the same. Once we'd each figured out our own and started moving the cognitive work into them, the variance shrank.

This one took us longest to see, and once we saw it, a lot of the rest made sense.

Multitasking is a myth. What feels like multitasking is rapid task-switching, and every switch carries a real cost – somewhere between ten and twenty minutes before your mind is actually doing the new thing, depending on which study you read. That two-hour task that was supposed to take an hour? Sometimes the system underneath it was broken. Sometimes the work was in the wrong window. And sometimes – more often than I'd like to admit – it was a one-hour task done four times because we got pulled away three times. A block of time isn't the same as a block of protected time. We weren't watching the second part, and it cost us.

Mood and emotion sat alongside all three of these. Not as a separate reading, but as something that pulled or pushed on the others. A difficult client conversation we hadn't processed made the next two hours of cognitive work harder, regardless of when in the day they happened. A small win made the same two hours easier. We didn't track it separately. We learned to let it inform how we read the rest.

The thing I want to say plainly, because it took me too long to figure out and almost no one writing about this will say it: the moment I most need to step away from the work is the moment I feel least able to. The ten-minute walk I should take when I'm overwhelmed is precisely the walk overwhelm tells me I can't afford. Every time I've ignored that signal, I've paid for it. That isn't a willpower problem. That's how the depletion presents itself. I had to learn to recognize the signal and go for the walk anyway.

The three readings, run together, in parallel, for long enough to stop being surprised by what we found – that was the diagnostic. And the diagnostic was what told us what to fix first.

When the readings were in front of us, the verdict separated cleanly. Some of what was eating our capacity was genuinely a tools and mechanical work problem. Some of it was a people problem – cognitive work landing at the wrong time, in unprotected blocks, on a calendar that didn't allow for the kind of humans we actually are. Both were present in our case. The proportions surprised us – more people-problem than we'd expected. And the order absolutely mattered.

People, process, tools – in that order. Any problem at this scale only gets solved with all three, but starting at the wrong end of the sequence is why most rebuilds fail. We couldn't process-and-tool our way out of a people problem. We could, sometimes, people our way out of something we'd assumed was a process or tools problem. The hours back from that move alone were real, and they cost us nothing to recover.

There were tools and processes later in this story – next week's issue is most of them – but it started here. The power of habit and the power of visibility deserve to sit alongside the power of automation when you're naming the things that change a practice. The industry talks about the third one constantly and the first two almost not at all, which is part of why so many practices buy the tool, miss the diagnosis, and end up no further forward.

I'll be honest – this was harder than it sounds. We were the people who most needed the diagnostic, and we were also the people who could least make space to run it. The first move was the hardest, because we had to use capacity we didn't yet have to recover capacity we needed. We had to start small. Track for a week, not four. Defend one window, not five. Take the walk that afternoon. The whole thing compounded; we didn't have to do all of it at once, and trying to would have killed it.

If you're going to take any of this seriously, you need a way to capture the time. Not reconstructed at the end of the day from memory, which is when most of the signal gets lost. We use Harvest. You can record time as it happens, or enter it after the fact when you've forgotten – which we do more often than I'd like to admit, and which is still better than nothing. You can build it out as deep or as shallow as the diagnostic needs: projects, clients, budgets within client engagements that flag when you're trending over the expected hours. The point of those alerts isn't necessarily to pull back. It's to provide the visibility, so you stop being surprised by the variance because you can see it forming in time to do something about it.

There's a free tier that's enough for solo work, and paid tiers if you need it across a team. If you want to try it, this link gives you 10% off your first month on a paid plan.

Full disclosure: if you sign up through that link, I get a one-time $10 credit. That's the extent of it.

Once the people layer is honest, the rest of stage one is process and tools – rebuilding the system underneath the work so the mechanical layer stops eating the time you've now scheduled properly. That's where data architecture comes in. That's where the maintenance trap shows up. And that's where the question of where AI does and doesn't belong has to be answered carefully. I'll walk through what we did and what we learned, and dangle a longer issue I want to write on the AI question specifically.

If you're sitting with the variance right now, that's the place this issue is meant to find you. And if you're starting to suspect the problem isn't only with the work, that's the other place.

Join the Waitlist • Schedule a Demo

Know someone navigating the compliance-to-advisory transition? Forward this email — or better yet, send them to baifokal.beehiiv.com to subscribe.get "good meeting, thanks" as they walk out the door, that one's for you. Hit reply and tell me about it — I'm genuinely curious how many of us have had that exact experience.