“Mission-critical” and “high switching costs” are the two most common phrases in every vertical market software CIM, and they're also two of the least examined. A seller's advisor writes them because they're true often enough to be a safe default, not because anyone ran a test. The result is that switching costs get underwritten as a qualitative impression, a general sense that customers seem unlikely to leave, rather than as something actually measured. That gap matters more in VMS than almost anywhere else in software, because the entire valuation premium for a niche, unsexy, single-industry product over a broader horizontal alternative rests on the claim being true.
This is a methodology for testing the claim rather than accepting it, built around five signals that are each independently checkable in a normal diligence process, none of which require anything more exotic than reading contracts closely and asking better questions on calls that are already happening.
Why the default level of skepticism should be higher than it is
Every seller says their product is sticky, for the simple reason that saying otherwise would be strategically irrational in a sale process. That alone doesn't make the claim false, but it does mean the claim carries no information on its own; the base rate of sellers asserting stickiness is close to 100%, which means the assertion should move a buyer's estimate of true switching costs by approximately zero. What should move it is evidence that would be inconvenient for a seller motivated to overstate retention, evidence the seller wouldn't manufacture voluntarily. That's the organizing principle behind everything below: each signal is chosen because it's harder to spin than a logo-retention percentage on a summary slide.
Signal 1: Contract structure and termination mechanics
Read the actual customer contracts, not a summary table of contract terms prepared by the seller's team. The summary table reliably rounds toward whatever makes the base look stickier, multi-year terms get emphasized, termination-for-convenience clauses get omitted, and auto-renewal language that technically allows a 30-day opt-out gets described simply as "multi-year."
What actually matters in the contract language:
- Initial term versus renewal term. A one-year initial term with annual auto-renewal is materially different from a genuine three-year commitment, even if both get summarized as "multi-year" on a slide.
- Termination-for-convenience rights. If a customer can exit with 30 or 60 days' notice regardless of contract length, the nominal term length is close to meaningless as a switching-cost signal.
- Notice windows and auto-renewal defaults. A contract that auto-renews unless cancelled 90 days in advance creates real friction; one that requires active re-signing creates none.
- Price escalators built into the contract. Contracts silent on pricing mean every renewal is a fresh negotiation, which is itself informative about how much leverage the vendor actually holds.
Pull a sample across the customer base, not just the top ten accounts by revenue, since the top accounts are the ones a seller has the most incentive to have negotiated well and the least representative of what a typical customer actually signed.
Signal 2: Integration depth
A product a customer logs into is different from a product woven into a customer's daily operating workflow, and the difference is measurable by asking a specific question: what other systems does this software actually talk to, and how would removing it break something else. A route optimization tool that only exports a CSV report has shallow integration depth. The same tool with a live API connection to a dispatch system, a billing system, and a driver mobile app has deep integration depth, and ripping it out means re-engineering three other workflows, not just switching a login.
Ask for an integration map, not a features list: which systems does the product connect to via API, EDI, or direct database access, at what percentage of the customer base, and how long did each of those integrations take to build during onboarding. That last number, actual implementation time for a new customer, is one of the most honest proxies available for switching cost, because it's an operational fact the seller has to report accurately for their own onboarding team to function, not a marketing claim. A product with a two-week implementation timeline has a fundamentally different switching-cost profile than one with a four-month implementation timeline, regardless of what either seller's pitch deck says about stickiness.
Signal 3: Migration and reimplementation cost, estimated from the seller's own numbers
The company's own implementation timeline and cost for a new customer is a reasonable, conservative proxy for what it would cost a competitor to win one of its existing customers away, since a competitor faces a comparable technical migration plus the added friction of extracting data from an incumbent system that has no incentive to make the export easy. If onboarding a new customer takes eight weeks and meaningful services revenue, a competitor's realistic cost to displace an existing customer is at least that, and usually more.
This is worth stress-testing directly rather than assuming: ask what percentage of the historical customer base churned specifically to a named competitor, versus churning simply by going out of business, getting acquired, or discontinuing the underlying business need entirely. Competitive churn, someone actively choosing a rival product over this one, is the number that speaks to switching costs. The other categories speak to market dynamics that have nothing to do with product stickiness and get conflated with it constantly in a seller-prepared retention analysis.
Signal 4: Price elasticity, the most direct evidence available
This is the single best test of a switching-cost claim, because it's the one signal that requires the company to have actually taken a real-world action with financial consequences, rather than a hypothetical. Has the company raised list prices, or renewal prices, on its existing base in the last two to three years? If so, what happened to logo retention and revenue retention in the cohort that received the increase, specifically, not blended into the overall retention number?
A company that has never tested a price increase has no real evidence of pricing power, whatever the retention slide says. A company that has, and held onto 95% of the affected accounts, has produced the closest thing to a controlled experiment on switching costs that exists in a diligence process.
If the company has never raised prices, that's itself informative, sometimes it means genuine reluctance to test the relationship, and it's worth asking directly what internal assumption is driving that reluctance. A founder who says "we've never tried because we're worried about churn" has just told you, more honestly than any retention chart, where they think the real switching-cost ceiling is.
Signal 5: Cohort retention curves, not headline NRR
A single blended net revenue retention number, discussed in more detail elsewhere on this site, hides more than it reveals about switching costs specifically, because it mixes expansion revenue from happy accounts with churn from unhappy ones and can look healthy even when a specific, informative subgroup is leaving at an elevated rate. For switching-cost purposes, the more useful cut is a cohort retention curve: of the customers who signed in a given year, what percentage are still active at 12, 24, and 36 months, tracked separately for each signing cohort rather than blended.
A business with genuinely high switching costs shows a retention curve that flattens, most of the churn happens in the first 12 to 18 months, while a founder discovers whether the product actually fits their workflow, and the curve goes nearly flat after that, since customers who survive the initial adoption period rarely leave. A business without real switching costs shows a curve that keeps declining steadily well past the two-year mark, which indicates ongoing, ordinary competitive churn rather than a one-time adoption filter, whatever the average tenure or blended retention percentage suggests.
A switching-cost scorecard
Putting the five signals together into something usable in an actual diligence process, rather than five separate conversations that never get synthesized:
| Signal | Genuinely high switching costs | Asserted, not yet verified |
|---|---|---|
| Contract terms | Real multi-year terms, no termination-for-convenience, meaningful notice windows | Nominal multi-year label, easy opt-out, or terms only reviewed via summary table |
| Integration depth | Multiple live API/EDI connections across most of the base, documented integration map | "Deeply embedded" asserted without an integration map or connection count |
| Migration cost | Long, services-heavy onboarding; low competitive (vs. non-renewal) churn | Fast, self-serve onboarding; churn reasons not broken out by cause |
| Price elasticity | Tested price increase with retained cohort data to show it | No price increase ever tested, or results not broken out by affected cohort |
| Cohort retention | Curve flattens after 12–18 months across multiple signing cohorts | Only a single blended NRR figure available, no cohort-level data |
None of these signals require anything a seller should be reluctant to provide in a serious process; a seller who resists producing cohort-level retention data or an integration map is itself a data point worth weighing. The switching-cost premium is one of the largest single drivers of valuation in vertical market software, which is exactly why it deserves the same evidentiary standard as anything else that size, rather than a sentence on a summary slide taken at face value.