Some companies keep customers by delighting them. Others keep customers because leaving would be a nightmare. The second group has a moat that network effects get more press for, but that quietly protects some of the most reliable compounders in the market: switching costs.

The definition

Switching costs are everything a customer must give up or endure to move from one provider to another — beyond the price of the new product itself. When those costs are high enough, customers stay even when a competitor offers something cheaper or better. The incumbent doesn't have to win every year's product comparison; it only has to avoid being so bad that leaving becomes worth the pain.

That asymmetry is the moat. It converts customers into annuities.

The three types

Financial switching costs. Direct money: migration projects, early-termination fees, retraining budgets, parallel-running two systems during a transition, rebuilding integrations. Enterprise software migrations routinely cost multiples of the annual license fee — which is precisely why they rarely happen.

Procedural switching costs. Time, effort, and risk. Years of accumulated data in one system, workflows built around one tool, an operations team certified on one platform. The killer variant is risk to mission-critical operations: if the system schedules the factory or clears the trades, the downside of a botched migration dwarfs any subscription savings. Nobody gets fired for renewing.

Relational switching costs. Habit, familiarity, and organizational muscle memory. Ten thousand employees who know the current software keystroke-by-keystroke are a switching cost, even if no invoice ever says so.

What switching costs buy the business

  • Retention that borders on gravity. Mature enterprise software businesses routinely retain 95%+ of revenue year after year.
  • Pricing power in small doses. A vendor that is painful to replace can raise prices a few percent every year, essentially forever, without triggering churn. Compounded, those "small" increases are the difference between a good business and a great one.
  • A base for expansion. Locked-in customers are the cheapest audience for new modules and cross-sells — each one deepening the lock further.

Companies like Microsoft (MSFT) stack all three types: financial (enterprise agreements), procedural (data, integrations, IT skills), and relational (a workforce raised on its tools) — then layer ecosystem network effects on top.

The metrics that reveal real lock-in

Claimed stickiness is cheap; measured stickiness isn't. Look for:

  • Net revenue retention (NRR) above 100% — existing customers not only staying but spending more.
  • Gross churn in the low single digits for enterprise-focused businesses.
  • Contract length and prepayment — customers who sign multi-year deals are telling you they don't expect to leave.
  • Price increases that stick — the cleanest evidence. If the company raises prices annually and churn doesn't move, the lock is real.
  • What happens in bankruptcy courts and case studies — when customers describe migrations as multi-year projects, believe them.

Where switching costs break

The great weakness of switching-cost moats is the platform shift. Lock-in protects incumbents only while customers stay on the current architecture; when a technology transition forces everyone to rebuild anyway — on-premise to cloud, desktop to mobile, and now conventional software to AI-native workflows — the switching cost resets to zero for the whole market at once. Incumbents that navigate the shift can re-lock customers on the other side; those that don't discover their moat was tied to the old platform, not to them.

How MoatScan scores it

MoatScan's AI rates the switching costs pillar from 0 to 10 for every company it analyzes, weighing integration depth, retention economics, contract structure, and the realistic pain of leaving. See every company in the database scoring 7 or higher on the stocks with high switching costs list, or open a full analysis from the database to see how the pillar interacts with the other four.

The rest of the series