Margin note
The entry multiple sets the return.
The Stickiness Trap
A business that never loses a customer can still lose you money if you paid too much for the privilege of owning it. In the vertical software sector, investors are prone to treating customer retention as an absolute shield against capital loss. They look at a company with high customer retention and assume that the durability of the revenue stream justifies any price.
This assumption is a dangerous confusion of product quality and investment value. Stickiness represents a quality of the product rather than a justification for an inflated purchase price. If you pay ten times revenue for a business that is growing at five percent a year, you have purchased a low-yield bond with terrible liquidity. When interest rates rise or market multiples contract, the valuation will compress regardless of how much the customers love the software. The business operations remain unchanged, but your investment returns are destroyed by the math of the entry price.
The Montreal Olympic Stadium is the same lesson in concrete. It is a useful structure that still stands and still hosts events, and it was a financial disaster anyway, because the capital deployed to build it was never going to be recovered at the price paid. Usefulness and value are separate questions, decided by separate people, and only one of them is settled at the moment you write the cheque.
The Arithmetic of the Return
The historical numbers from the private and public software markets show the consequences of ignoring valuation discipline. During the peak of the software bubble in 2021, the public median multiple for software-as-a-service businesses reached roughly seventeen times run-rate revenue — the SaaS Capital Index peaked at 16.9x in August 2021, on the same run-rate basis used below. Since that time, a significant re-rating has occurred.
The SaaS Capital Index held a public median near 7.0x run-rate revenue through early 2025, and then fell further. At the start of 2026, on the view that AI posed an existential threat to the software business model, the index re-rated sharply lower; SaaS Capital's own reading in April 2026 puts the median ARR multiple at decade-plus lows. Call the current public median somewhere in the low single digits. SaaS Capital's model for private companies predicts roughly 4.8x annual recurring revenue for bootstrapped firms and about 5.3x for equity-backed ones — figures worth holding loosely, because private marks lag public ones by two or three quarters and those numbers have not yet absorbed the public re-rating. The market has repriced the category twice now, and both times the direction was down.
The premium valuations are reserved for businesses that pair high growth with exceptional net revenue retention — call it net revenue retention of 120% or better alongside a Rule of 40 score somewhere around 50. The typical workforce training or compliance platform does not clear this bar. SaaS Capital's 2026 survey of private B2B software puts the median bootstrapped company in the $3M–$20M ARR band at 15% revenue growth and 103% net revenue retention. That is the shape of these businesses: stable, and ordinary. Paying a premium multiple for those metrics is an expensive mistake.
The Boring Software Paradox
The market for frontline training and workforce software is structurally attractive, though not for the reason the sector's boosters usually give. The common story is that a skilled-labor shortage forces employers to cut internal training headcount and buy external tools instead. Training Magazine's 2025 Industry Report does not show that. Total U.S. training expenditure rose 4.9% in 2025 to roughly $102.8 billion, after about $98 billion the year before, and training payroll rose with it — up roughly 7% to $64.7 billion. Nobody was cutting L&D staff to fund software.
What the same report does show is a mix shift inside a growing budget. Spending on outside products and services rose 29% to $16 billion, against that 7% payroll line. Both grew; the external-tools line grew about four times faster. That is the narrower claim the data supports, and it is enough: the vendor's addressable budget is expanding faster than the department around it.
This revenue tailwind is real, but it is also highly visible. The mistake investors make is assuming that a structural tailwind for the industry translates automatically into a tailwind for investment returns. The industry tailwind comes for free; the entry multiple is arithmetic you have to do yourself. Confusing the two leads to paying growth-software prices for slow-growing annuities.
A boring workforce software company is a resilient asset, and a limited one. The addressable market for training software in a single niche, something like vocational training for regional HVAC technicians, is small and highly fragmented. A vendor can saturate the niche quickly and find that expansion beyond it is difficult. The defensible multiple for this type of business is somewhere around five times annual recurring revenue, which is roughly where SaaS Capital's model puts private companies generally. Paying eight or nine times revenue means you are assuming a growth trajectory that the market structure cannot support.
Sizing the Exit
When evaluating an opportunity in this space, size the position to leave through a door built for one person. The capacity constraints of these niches cut both ways: they protect the small vendor from larger competitors, and they trap the investor who overpays and needs a large exit.
Consider a software company with five million dollars in annual recurring revenue. At a disciplined multiple of five times ARR, the enterprise value is twenty-five million dollars. That is a meaningful position for an individual allocator. It is also highly illiquid, because the business is private and the buyer pool is thin. If you pay an inflated multiple assuming you can sell it to a larger fund later, you are relying on the existence of a greater fool.
The larger funds cannot buy a twenty-five-million-dollar business, because the diligence cost does not shrink with the check size and the position does not move the needle against their capital base. The realistic exits for a small, boring software position are holding it for the cash flow or selling it to another solo operator. That reality forces you to underwrite the investment on cash yield rather than multiple expansion.
Underwriting the Compression
There is no terminal or research team behind this kind of analysis, and there does not need to be. The work is a spreadsheet that calculates the implied return over a ten-year holding period under different exit multiples. The return has to make sense even if the multiple compresses from five times ARR to four during the hold, because that is the assumption most likely to be wrong in the direction that costs money. The public re-rating in early 2026 is the argument for underwriting a harder compression than that: a category median can halve in a quarter on a story about AI that nobody has finished telling yet.
The whole frame is not about chasing the fastest-growing asset. It is about waiting for the one priced below what its cash flows are worth. A small software company with high customer retention and low growth is a fine asset at four times ARR and a capital trap at eight, and nothing about the software itself changes between those two sentences.
Which is the obvious weakness in all of this. Every number above is a market-level average — an index median, a predicted private multiple, a survey of training budgets — and no individual company is an average. The discipline of refusing to pay eight times is easy to state and mostly untested, because the piece of the argument that actually matters is what the multiple does over ten years, and that part is guesswork here as much as anywhere. The business never needed to be exciting to make you money, but the price you paid for it is the only part of the trade you ever controlled.
