Sunday. The week's over, the numbers are what they are, and here's what I've got. Three things I didn't know on Monday.

I've been doing a weekly review for years. Friday afternoon, hour blocked, look at what happened, adjust. It's one of the habits I'd have told you was solid. If you'd asked me to list the things I do consistently, it would've been near the top.

This week I went back through the calendar for the last six months and counted how many of them I actually did.

About sixty percent. Which isn't terrible on its own.

Then I looked at which ones I skipped.

Every single one of them came after a bad week.

Not most. All of them. The pattern was so clean it was almost funny. Good week, solid numbers, things landing, I sit down Friday and review with a coffee and feel like a professional. Rough week, something fell through, a client went quiet, I look at the calendar block on Friday afternoon and decide I've got something more urgent to do, and the something more urgent is always real enough that I don't have to notice what I'm doing.

So what I actually built wasn't a review habit. I built a celebration habit with a review habit's name on it. For six months I've been carefully documenting exactly the weeks that had the least to teach me, and skipping every single week where the lesson was sitting right there on the table.

What got me is that this was invisible from the inside. The habit felt real because on any given Friday I either did it or had a legitimate reason not to. You can't see a pattern from inside one instance of it. You can only see it by counting, after the fact, which is exactly the kind of thing nobody does with their own habits because we assume we already know how we behave.

The fix isn't more discipline, because discipline isn't the failure point. The failure point is that avoidance dresses up as prioritization and I can't tell them apart in the moment. So I moved the review to Monday morning, before anything else, when the week's not yet full of urgent things to hide behind. It's been two weeks. Ask me in six months whether that worked, and I'll count instead of guessing.

The larger lesson, and this one stings a bit, is that any habit you've never audited is a story you're telling yourself. I'd have sworn on this one. I'd have put it on a list of things I do well. The gap between what I believed about myself and what the calendar said was six months wide and I walked around inside it completely unaware.

What makes this worse than a plain consistency problem is the direction of the bias. If I'd skipped randomly, sixty percent coverage would still give me a representative sample. I'd have a decent picture of how things actually go. Instead I built a dataset with every hard week filtered out, then made decisions off it. That's not incomplete information. That's actively misleading information that I generated myself and then trusted because I generated it.

And I'd have defended it. If you'd asked me in June how the business was going, I'd have answered from that record, confidently, and I'd have been wrong in a specific and predictable direction. Rosier than reality, with receipts.

Go count something you think you're good at. Not how often you do it. Which times you skip. The pattern in the gaps tells you more than the pattern in the entries. It's a rough hour and it's worth it.

Sheldon sits by the door around the same time every afternoon. Yorkies are not large animals and he is not a subtle one, so this involves a fair amount of staring at me with his whole body. There is no ambiguity in the message. There has never been ambiguity in the message.

Here's what I noticed this week, and I'm aware of how this sounds.

He does it on the days we go and he does it on the days we don't.

Rain, a call that runs long, a knee that's being uncooperative, doesn't matter. Same spot, same time, same expression of total confidence that this is happening. And when it doesn't happen, he's disappointed for about four minutes and then he's asleep, and the next day he's back at the door at the same time with exactly the same conviction.

He has never once adjusted his behavior based on his hit rate.

I've been thinking about that all week, because I do the opposite constantly and I've always thought of it as being adaptive. I show up for things in proportion to how well they've been working. Content that gets a response, I make more of. Outreach that goes quiet, I quietly do less of. The reasoning feels sound. Follow the signal, don't waste effort where it isn't landing.

The problem is the timescale. Most of what I do has a feedback delay long enough that my adjustments are responding to noise. I pull back on something in week three because week two was quiet, and I'll never find out what week nine would have looked like, because I ended the experiment before it produced any information. Then I conclude it didn't work, which is technically true and completely useless, because what actually didn't work was my patience.

Sheldon's system is dumber than mine and better than mine, because his consistency isn't contingent on results he can't properly evaluate anyway.

I'm not going to pretend this means you should never adjust. Obviously you should. But I think the honest version is that adjustment needs to be scheduled, not reactive. Decide in advance how long you're running something and what would genuinely constitute a signal, then run it that long regardless of how it feels in week two. The dog gets to be reactive because his feedback loop is one day long. Mine isn't, and I've been pretending it is.

The other thing worth naming is what all that mid stream adjusting actually costs, which is more than the abandoned experiment. Every time you change the thing, you reset the clock on learning anything about it. Six months of constant tweaking doesn't give you six months of data. It gives you a dozen two week fragments, none of them long enough to mean anything, and a strong feeling that you've been working hard. The feeling is real. The learning isn't.

I think that's why the quietly consistent people tend to pull ahead in ways that look sudden from outside. They're not more talented. They just left things alone long enough to find out whether they worked.

Anyway. He got his walk. He was right to be confident.

I gave a client a recommendation about six weeks ago that I was genuinely confident in. Walked them through the reasoning, they went with it, and it didn't work. Not catastrophically. It just didn't do the thing I said it would do, and by this week that was clear to everyone including me.

My instinct, and I want to be honest about this because I think it's most people's instinct, was to manage it. Not to lie. To contextualize. Point at the parts that did move, note the external factors, frame it as a phase in a longer arc. All of that would have been defensible and some of it would have been true.

I didn't do that. I opened the call by saying I was wrong about it, here's specifically where my reasoning broke, and here's what I'd do differently.

The silence afterward was about two seconds long and felt like a week.

Then the conversation changed completely. And I mean the whole texture of it changed. They started telling me things they hadn't told me before. Constraints they'd been managing around, a piece of internal politics that had been shaping their decisions the entire time, a budget conversation they'd been having without me. Stuff that, frankly, would have changed my recommendation six weeks ago if I'd had it.

Which raises an uncomfortable question about why I didn't have it.

I think the answer is that I'd been presenting myself as someone who has this handled, and people don't hand you complicating information when you're projecting that you've already got the picture. Why would they? You've told them the picture is complete. Volunteering a complication just makes them the person who slowed things down.

By admitting the picture was wrong, I accidentally signaled that more information was welcome. And it turns out there was a pile of it sitting right there the whole time.

So the lesson isn't the warm one about vulnerability building trust, though that's probably also true. The lesson is more mechanical than that. Certainty is a posture that suppresses information flow. If you always seem to know, people stop telling you things, and then you know less, which makes you more likely to be wrong, which you'll then be tempted to cover with more certainty. That's a loop, and it tightens.

The way out is to be wrong out loud earlier and more often than feels comfortable. Not as a performance of humility. As a practical mechanism for getting people to tell you what they know.

I want to be careful not to turn this into a tidier lesson than it is, though. The recommendation still didn't work. Six weeks of their time and money went into something that didn't deliver, and no amount of productive conversation afterward gives that back. There's a version of this story that ends with everything working out and the mistake being secretly good, and that version is a lie. The mistake was just a mistake. What came after was salvage.

The useful part is what it says about the six weeks before. The information I needed was sitting in that room the entire time, and the thing keeping it from me wasn't their reticence. It was my posture. That's the part I actually have control over, and it's the part I'd been treating as an asset.

That relationship is in better shape now than it was before the mistake. I'm still turning over how I feel about that.

That's the week. Three things I didn't have on Monday. One of them came from an audit I didn't want to run, one of them came from a client call I didn't want to make, and one of them came from a nine pound dog who has never once doubted that the walk was happening.

See you Tuesday.

One step, one day. Grace over guilt. — Dan Kaufman