5 Weeks of Freemium, and What the Business Is Actually For

5 weeks ago I tore the trial wall off Searcher OS. 2 weeks after the switch I published the teardown with early numbers and a promise to report back once the data had time to age.

This is that report. I pulled every number below from the production database this morning. One finding genuinely surprised me. And the bigger story sits underneath it.

Quick recap if you're new here: Searcher OS is an operating system for buying a small business. For most of a year the front door was a short free trial and a premium price. In late June I replaced it with a genuinely usable free tier and kept the paid product for serious, in-market buyers.

The scoreboard at week 5

• 165 people signed up in July, roughly 4x the monthly pace before the switch. (For contrast: one week in May, 3 people signed up.)
• Searcher OS crossed 100 paying customers this week.
• New paying customers, the full-price kind, are arriving at several times the old monthly rate. No trial deadline, no discount, no conversation with me.
• Cash collected is averaging about $7,000 a month over the past 5 months, and it's accelerating.
• Still $0 on ads and $0 on payroll. AI agents handle most of the day-to-day support and break/fix on the $200/mo tier.

(A group licensing motion is contributing meaningfully too. Different lesson, different post.)

So the switch worked. But the useful part of a checkpoint is holding what happened against what you predicted. That's where it gets interesting.

The theory behind the pivot

My July argument was a depth story. The people who get excited about Searcher OS are the ones who stick around a few weeks and hit the second layer, where the AI quietly starts running their workflow. A business search runs in cycles measured in weeks: new listings drop, NDAs go out, CIMs come back, deals die. A 7-day trial ends before the first cycle completes.

So I predicted slow-burn conversions. Free users would wander the product for weeks, hit the deep end somewhere around week 3, and upgrade when the depth sold itself.

What the data says instead

Since the switch, every new subscription has started within that person's first 8 days on the product. Most inside a week.

The week-3 upgrade I predicted hasn't happened once. 5 weeks in: zero.

I stared at that for a while. Here's my current read, held loosely because it's early.

When I drew the freemium line, I followed my own checklist rule: keep the aha moment in the free tier, put the urgency in the paid one. The capabilities behind the paywall are the ones active, in-market buyers lean on hardest. A buyer who shows up mid-deal, with a live pipeline and real momentum, hits that line almost immediately. They find the deep end in days because they arrive already knowing what they need.

Serious buyers bring their own urgency. The trial deadline was piling pressure on everyone else: the early-stage browsers, the maybe-next-quarter crowd, the people whose search hadn't warmed up yet. Under the old model those people expired and vanished. Under the new one they stay.

In July I wrote that if your product's depth shows up in week 3, you shouldn't make people decide in week 1. I still believe that. But 5 weeks of data added a clause I didn't see coming: take the deadline away, and the serious ones decide in week 1 anyway. Their urgency was theirs all along; the paid line just gives it somewhere to land.

The question that's still open

So what happens to the ones who stay? That's the affiliate bet, the piece I flagged as unproven when I made the switch: grow a big free base, then monetize the long tail through partnerships over time. It hasn't paid out yet.

But I want to be precise about what kind of question that is, because it changed shape for me over these 5 weeks. This was a one-way door, and I've already walked through it. Freemium is producing more paying customers in total volume than the trial ever did, so walking it back would mean rebuilding a wall that was costing me customers. That option is off the table no matter what the affiliate math ends up saying.

The live question is narrower: does affiliate revenue become a material mover, or does Searcher OS settle in as what it already is, a healthy, self-running SaaS throwing off a supplementary income stream while I work on other things? I honestly don't know. The free base is young and the partnerships are early. I'll report back when the math has had time to say something.

Both outcomes are fine. One is just bigger.

What the business is actually for

Here's the part I'd underline if you're reading this as a pricing case study: I can afford to hold that question loosely because the business runs either way.

AI agents handle most of the support queue and the break/fix work. The scrapers pull fresh listings off hundreds of broker sites every morning without me. The real test was July itself. I was gone most of the month, Colorado, then Yosemite, then Hawaii with the family, and the product onboarded users, answered tickets, and collected revenue the whole time. The month signups 4x'd is the month I was mostly out of the building. (I checked in more than I needed to, mostly out of habit.)

So my job has shrunk to a choice I get to make each morning: spend the day growing it, or let it compound quietly while I work on something else. Lately I mostly choose something else.

That's the thesis I left corporate to test. Income that arrives whether I showed up that day or not. The feeds keep declaring SaaS dead, and I get the argument: AI made software cheap to build, so moats look thin. But from inside one, a solo SaaS with $0 payroll, run by agents, growing on content and banked SEO, is the healthiest version of this business model I've ever operated. For a solo operator, AI collapsed the payroll line and left the revenue line alone. That math got better, not worse.

And there's a quieter return that took me longer to see: credibility. What I'm actually most interested in is the agents themselves. Searcher OS is my proof that they can run a real business, with paying customers, production stakes, and a support queue with no humans behind it. When I sit down with an operator who wants agents in their company, I can point at a live P&L instead of a slide deck.

Two things are growing on top of that proof right now. I'm exploring adventure retreats that combine my 2 favorite things, the outdoors and AI: the first one runs September 18 to 20 in South Lake Tahoe, 8 seats, hands-on agent building between trail time (details here). And I've started helping companies and individuals figure out where agents actually pay for themselves (the advisory page has the shape of that).

So that's the checkpoint. The front door is working, the affiliate question is open, and the agents are running the place either way. Ask me again in the fall.