Direct Derek - Workforce Enablement

Field-Service Software Living on the Technician’s Phone

2026-04-26T00:00:00+00:00

The software worth studying in this category does not live in an executive suite. It lives in the door pocket of a service van, gets dropped on gravel driveways, gets smeared with grease, and is opened twenty times a day by someone wearing work boots. Trade-press surveys put roughly seventy percent of field technicians running the entire workday off a mobile device as their primary tool. That is the operating environment, and it decides everything downstream: in that environment vertical software either becomes infrastructure or gets deleted.</p>

The Cracked Screen on the Dash</h2>
The day is paced by work orders. When a technician arrives at a job, the app is the first thing they touch. If it takes five taps to log a diagnostic reading, or if it loses its connection in a basement, the technician stops using it. They revert to a grease pencil and a scrap of cardboard, and the dispatcher spends the afternoon trying to find them.</p>
So the product that wins is the one that removes friction from the workday rather than the one that renders the prettiest chart. The directional numbers, which come from vendors and trade press rather than audited studies, run in one consistent direction: paper-based technicians burn something like six hours a week each on administrative work, and roughly seventy-three percent of technicians name paperwork as their leading daily frustration. Voice-to-form entry is reported to complete that paperwork about thirty-five percent faster. Run the arithmetic on a single technician and you get somewhere around two hours a week back, which the owner experiences as capacity rather than as a feature. Treat all of these figures as directional. They point the same way, which is the most you can ask of a vendor statistic.</p>

The Retention Anomaly</h2>
Small-business software is a difficult market to hold. A SaaS-focused lender calls seventy-five to eighty percent gross revenue retention quite strong</em> when you are selling to small businesses — fickle customers, owner-driven decisions, high underlying failure rates — against ninety percent and up when selling to banks and insurers. Median net revenue retention for the sub-$25,000-ACV segment sits around ninety-seven percent, which means the median company selling to small businesses shrinks inside its own base before it sells anything new.</p>
Vertical field-service platforms report numbers that do not belong to that segment. ServiceTitan's S-1 discloses net dollar retention above one hundred and ten percent for each of the last ten fiscal quarters and gross retention above ninety-five percent over the same window. That disclosure is contested: at least one analyst argues that once the quarterly figure is annualized it normalizes closer to eighty-one and a half percent — and lower still, since the disclosed metric counts only customers who churn to zero and ignores partial downsells — a gap of roughly thirteen points that is entirely definitional, not a dispute about the business. They also call eighty-one and a half percent "entirely believable and altogether fine" for the trades, which it is.</p>
Take the lower number and the claim narrows. Vertical SaaS gross retention averages around ninety-one percent, fintech-led vertical SaaS around ninety-six — so eighty-one and a half percent is not an anomaly against that cohort, it is ordinary. The anomaly is only against the small-business baseline of seventy-five to eighty percent, which is the right comparison for a base made of local plumbers and electrical contractors. The distance between the small-business baseline and the observed number is the depth of the switching cost, and it is worth knowing which methodology produced any retention figure before you admire it.</p>

The Retraining Tax</h2>
The barrier is a coordination cost, not a technical one. If a contractor wants to move from one field-service tool to another, the license fee is the smallest line in the calculation. The user base is not three analysts at head office; it is every dispatcher and every technician, retrained simultaneously. That cost scales with headcount and turnover, not with seat price.</p>
Work a hypothetical. A business with ten service vans has ten technicians and two dispatchers. Switching platforms means halting the schedule, bringing the crew off the road, and teaching twelve people a new system. Assume, for the sake of the arithmetic, sixty dollars an hour of billable revenue per technician and a single lost week: call it twenty-odd thousand dollars of foregone capacity, paid in cash, in one quarter, for a benefit that arrives later if it arrives. This analysis has no sourced figure for what that week actually costs across the industry, and the number moves with the trade and the region. The direction is what matters. The owner will tolerate a mediocre product, rising prices, and indifferent support for a long time before writing that cheque.</p>
The Payments Pipeline</h2>
The stickiest field software processes the money as well as the calls. Once a trades business runs estimating, job costing, invoicing, payments, and payroll through one platform, the tool is holding how the business gets paid and how it pays people. Jobber syncs approved timesheets into QuickBooks Online and runs payroll through Gusto. ServiceTitan's core spans call tracking, scheduling, dispatch, estimating, job costing, inventory, and payroll integration.</p>
The scale of that plumbing is the part institutions did notice. Roughly sixty-two billion dollars of gross transaction volume moved through ServiceTitan in the trailing twelve months, with the company capturing about one percent of that volume as revenue overall and something like a quarter of a percent on the payments-processing slice specifically. Both figures are analyst-derived from the S-1 rather than lifted from a headline number in it. Note the second engine hiding in there: because the take rate rides on the customer's transaction volume, vendor revenue grows without anyone raising a seat price. A pure per-seat scheduling tool has no equivalent.</p>
Removing a system that sits between a business and its bank account is a different operation from swapping a CRM. The fear of a missed payroll run is a stronger barrier to exit than any contract term, because a contract can be breached and a payroll cannot be un-missed.</p>
Sizing the Tail</h2>
None of the above is a secret. The bellwether went public in December 2024, grew revenue to roughly $961 million in FY2026, and is covered, owned, and priced by people whose job it is to price exactly this. The stickiness is real and it is already in the number. There is no capacity-constrained inefficiency in a multi-hundred-million-dollar-revenue software company that institutions can buy in size.</p>
The tail is a different question, and a narrower one than it looks. Vertical benchmark data suggests the barrier requires density</em> within a vertical rather than mere presence — a tool at five percent penetration of its niche does not carry the embedded switching costs of one at thirty or forty. That density requirement is the fence. It keeps large players out of the smallest, most fragmented niches, because winning them requires knowing the specific language of rotating-shift clinic staffing or single-trade dispatch and cannot be bought with advertising. Field adoption is not a channel you can scale by spending money on it.</p>
Which is also the honest capacity read. The same smallness that keeps the institutions out caps what you can deploy and thins the buyer pool for the asset itself. There is no sourced answer to how small a high-retention niche tool can be and still support a purchase — that is an open empirical question, not a solved one, and anyone who tells you the number is estimating. At minimum, a framework which locates the edge precisely where the data runs out is a convenient one.</p>
What the Habit Actually Proves</h2>
The benchmark literature marks a business-to-business tool as part of a regular workflow when daily actives run above twenty percent of monthly actives, and calls thirty percent and up strong stickiness. An app opened at every job, every day, sits at the top of that range by construction. The technician taps the same screen at every stop because the paycheck depends on it, and that repetition is the asset — not the feature list, not the dashboard, not the roadmap.</p>
The weakness in the argument is that habit is measured after the fact. Retention data tells you a tool was hard to remove last year; it cannot tell you the price at which owning that fact stops being worth it. Muscle memory is very difficult to destroy and completely indifferent to what you paid for it.</p>

Boring Workforce Software Before the Multiple Expands

2026-04-22T00:00:00+00:00

The Stickiness Trap</h2>
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.</p>
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.</p>
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.</p>
The Arithmetic of the Return</h2>
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.</p>
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.</p>
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.</p>
The Boring Software Paradox</h2>
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.</p>
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.</p>
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.</p>
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.</p>
Sizing the Exit</h2>
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.</p>
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.</p>
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.</p>
Underwriting the Compression</h2>
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.</p>
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.</p>
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.</p>