A small business is priced as follows. A buyer takes the trailing earnings, usually three years of them, applies a multiple drawn from what comparable businesses fetched, and adjusts for the risks they can see in the books. Every input in that sentence looks backward. The multiple is a summary of yesterday’s transactions; the earnings are a record of yesterday’s work.
That arithmetic has a blind spot, and right now the blind spot is large. Whether software will do a business’s core work in five years is a fact about the future, and the trailing numbers carry no trace of it. A business can be priced generously by every backward-looking measure at the exact moment its forward-looking value has started to fall. The two do not reconcile until enough transactions have gone badly that the comparables themselves reprice, and by then the window has closed.
Where this bites first: an example from insurance
Independent personal-lines insurance agencies are a clean illustration, because the exposure is concentrated and the whole market is visible in public files. An agency’s product is distribution: it sits between households and carriers and earns a commission for the sitting. Direct writers have spent two decades pulling personal auto toward channels with no agent in them, and software is now doing to servicing what direct distribution did to sales. The commercial side of the same industry, where risks are odd-shaped and a human broker earns their fee in judgement, is a different business with a different future. The exposure is specifically the personal book, and most specifically nonstandard auto, the segment where the customer shops on price alone.
Meanwhile the agency rollups are still buying, and still pricing books off trailing commissions. Both things are true at once, and that is the window: the asset’s trailing numbers still price well while its forward risk is not yet in the comparables.
What the public record shows in Houston
Texas publishes its insurance licensing universe as open data: every agency license, every carrier appointment, every ownership association. We measured it in July 2026. In the Houston metro area there are 2,081 property-and-casualty agencies whose carrier appointments we could classify. Of those, 846 hold books where the majority of appointments are nonstandard auto, the most exposed segment there is, and 253 have owners whose license or ownership vintage is twenty years or older, which is the profile of an owner for whom exit timing is no longer an abstract question.
One further measurement, inexpensive to make and difficult to ignore: of the 150 highest-scoring agencies on our exposure-and-vintage ranking, roughly two-thirds have no findable website at all: storefronts that exist only on directory listings and social media pages. Read that both ways. There is no digital moat between those books and a price-comparison app, and there has been no owner reinvestment in years.
Source: Texas Department of Insurance licensing datasets (firm licenses, active carrier appointments, relationship records) published on the state open-data portal, data.texas.gov, measured July 2026. Appointment counts measure carrier breadth, not premium volume; they are a strong proxy for what a book is made of, and only a proxy.
2 in 3
of the 150 Houston agencies scoring highest on our exposure-and-vintage ranking have no findable website at all.
The medallion lesson
There is a recent, well-documented precedent for a licensed asset whose trailing economics stayed excellent right up until the repricing: the taxi medallion. Medallion prices set records in the early 2010s while ride-hailing was already operating in the same cities, because medallion buyers were pricing trailing lease income, not the thing that was about to happen to it. The owners who sold into those record prices did well. The lesson is not that anyone should have predicted the exact year. It is that for an asset facing structural displacement, the selling window closes before the operating business visibly breaks, because the price breaks first, and the price breaks the moment buyers stop trusting the trailing numbers.
The two honest responses
If you own a business in this position, an insurance book heavy in personal lines, or any business whose core work is priced per unit and increasingly automatable, there are exactly two responses that take the problem seriously, and they start from the same place.
- Re-mix, then hold. Shift the revenue toward the durable side (commercial lines, advisory work, whatever in your trade requires judgement and a relationship) and adopt the automation yourself, so the margin expansion shows up in your own books rather than a competitor’s. This is slow, it is real work, and done over two or three years it also happens to be exactly what raises a sale price later.
- Sell into the window. Not a panic sale but a prepared one, while the trailing three years still look the way buyers pay for, to one of the acquirers who values your customer relationships more than your operations. For the most exposed books, the buyer who makes sense is often one who intends to run the book with software from day one, which means they are paying for your renewals and your switching costs, and those are worth the most before attrition starts.
Both responses begin with the same unglamorous step: knowing what the business is actually worth today, from its real numbers, against real comparables, not a rule of thumb. Owners who have that number make the re-mix-or-sell decision calmly and early. Owners who do not usually make it late, in response to the first unsolicited offer, which is the one negotiating position worse than either choice made deliberately.
What we will not claim
We do not know the year the repricing arrives, for insurance books or for anyone else, and neither does anybody selling certainty about it. Displacement timelines are the most confidently mispredicted numbers in business. What is knowable is the direction, the mechanism, and, for licensed trades in Texas, the precise shape of your own exposure, measured from the state’s files rather than asserted. That is the part worth knowing this year instead of finding out from a comparable after the fact.