Not subscribed? Sign up to get it in your inbox every week.

Most local businesses run on memory.

Whose turn it is to follow up. Did we send the quote yet, or did somebody just say we did? Has the deposit cleared? Has production started? Is the install scheduled? Did we collect the final check?

Two months ago I wrote that I was going back to operating one of these full-time. Here's what I've actually been doing: pulling the business out of people's heads and putting it in Slack.

When I made the pivot, I thought the job was going to be marketing. Could you run Meta against a sign shop? Could you make Google Ads land? Could you compound through SEO? The short answer is yes.

Meta has been the breakout — we've roughly doubled the business in two months, with storefront campaigns pulling something like 9x ROAS against a buyer who is almost always the same person: a small business owner with a lease, a blank wall, a logo somewhere on their phone, and a lot of anxiety about whether the storefront looks legitimate enough to open. Banners pull cheap leads with capped order values. Monument signs took our money and produced zero. Mixed bag. Net good.

But growth is not the part worth writing about.

The part worth writing about is what happens to a thirteen-step physical fulfillment chain when you double the volume hitting the front of it.

What happens is everything starts leaking. (Like when a permit wasn’t approved because the city emailed about corrections and the email went to a sales rep who doesn't check that inbox). Installs scheduled into a calendar nobody reads. The work does not disappear loudly. It disappears in the slack — small s — between humans.

The clearest version of this happened two weeks ago. Our install crew showed up at a medical clinic, mounted the letters, and got back in the truck before anyone noticed one of the letters was upside down. The doctor saw it before we did. We fixed it the next morning, ate the cost, kept the customer. But the file made it through twelve handoffs and broke at the thirteenth, and there was no system in the company that would have caught it.

That story is funny to tell. It's less funny when you realize an upside-down letter is just the failure you can see from the parking lot. Everything else — a quote nobody sent, a mockup parked in a designer's drafts folder for two weeks — leaks out without anyone noticing. The medical clinic was loud. Most of what we lose, we lose silently.

So I stopped trying to throw more demand at the problem and started trying to remove the handoffs.

The frame I've been working off: the business should not live in anyone's memory. It should live in Slack. Slack is the interface. The database underneath is the source of truth. Trello is the visual board, QuickBooks is the money, and the bots in the middle move work between them so nobody has to remember whose turn it is.

In practice, that's meant building a real operating system on top of Slack. A /mockup command for the design intake — fills out a structured brief, creates a Trello card, opens a thread where the rep drops the storefront photo and the logo file, and auto-attaches every file to the right card so the designer never has to leave Trello to find what the customer sent. A /production command that turns a paid QuickBooks invoice into the right number of physical production items on the warehouse Trello board, hides every dollar amount from anyone who doesn't need to see it, and opens a thread where the production files land. I built the first version of that one over a long weekend, mostly because I was tired of watching paid jobs sit in my brother's head for three days before anyone in the warehouse heard about them. Trello status changes posting back into the originating thread so the salesperson sees in real time when the designer marks the work complete. Revision requests handled with a button instead of three forwarded screenshots. Lead enrichment from public sources running in the background.

The honest part of this story is that not all of it worked.

We spent a full sprint building a daily lead brief that aggregated permits, signals, and public construction notices into a triaged Slack list every morning. It cost us about seven dollars a day in API calls. We shipped it. Then I watched myself not open it. Then I watched everyone else not open it. We pulled it after a few weeks and rebuilt it as a weekly Monday post into a Google Sheet that the operator manages by hand. Cost dropped 97%. Usage went up. The lesson, which I owe to the dumbest possible source — paying attention to my own behavior — is that for a one-rep team, a daily review surface is too much friction. You build for the rhythm the operator actually has, not the one the spec says they should have.

The whole project, two months in, is one ongoing audit of where work used to disappear.

None of this is exciting in the way Meta ads are exciting. Nobody is going to write a thread about a /production slash command. But the actual lever in a service business — the one I keep getting reminded of the hard way — is not more demand. It's fewer places for work to disappear between the humans.

Most local businesses don't break from a lack of demand. They break because the volume eventually outruns the memory of the people running them, and the gaps where work used to fall through quietly start swallowing whole jobs.

We're running ours hot enough now to see the gaps. So we're closing them.

More soon.