Stated switching intent overstates observed switching in B2B software, and integration depth predicts the gap better than satisfaction does. The unhappy-but-integrated customer rarely moves, while the satisfied customer with a shallow deployment often does — which is why a churn thesis built on satisfaction scores reads the wrong variable.
What is stated switching intent worth?
As a ranking, a lot. As a rate, very little. Buyers reliably identify which vendors they are unhappy with and unreliably predict whether they will act on it.
The same pattern appears in prescriber research and in consumer categories, which suggests it is a property of stated intent rather than of software buyers.
The ranking is more useful than it sounds. Knowing which of six vendors a segment is most dissatisfied with tells a challenger where to concentrate, and it tells an incumbent's investor where the exposure sits — both decisions the ordering supports and the rate does not. The error is only in converting an ordering into a forecast.
How big is the gap?
Large enough to change a displacement thesis, and predictable enough to be adjusted for once you know what drives it. What drives it is not dissatisfaction but the cost of acting on dissatisfaction.
That is why satisfaction surveys are a poor churn proxy for embedded vendors: they measure the impulse and not the constraint, which is one of the things a retention number does quietly.
Predictability is what makes the gap workable. Because integration depth drives it, the gap can be estimated for a given customer base before any research is fielded: a base of shallow, recent deployments behaves differently from one of deep, mature ones, and that is observable from implementation records rather than from a survey.
What actually blocks a switch?
Integration depth and internal process change. A customer who dislikes their vendor but has built workflows, reports and downstream systems around it faces a project rather than a purchase.
Price is rarely the blocker and frequently the stated reason, because it is the most acceptable thing to say to a salesperson.
Internal process change is the part buyers understate most. Replacing a system means retraining, rewriting reports, renegotiating downstream integrations and absorbing a period in which both systems run — work that falls on people who did not choose the vendor and will not benefit from the switch. Their resistance is rarely modeled and frequently decisive.
How should you ask the question?
Anchored on the next renewal and the specific work a switch would require. "What would have to be true for you to move at renewal" surfaces the constraint; "would you consider alternatives" surfaces the impulse.
Asking what they have already evaluated is the other useful probe, because an evaluation that went nowhere is more informative than an intention that has not started — and it is a question a customer call settles in a minute.
The renewal anchor works because it forces the buyer to think about a date. An abstract question about alternatives invites an abstract answer; a question about what would have to be true by a specific renewal makes them consider the work, the budget and the internal argument they would have to win. Those three constraints are the ones that predict behavior.
What does this do to a thesis?
It moves the timeline rather than the direction. A challenger genuinely preferred by buyers can still take years to displace an incumbent, and a model that converts preference into share on a two-year view will be wrong about the hold period as well as the share.
It also identifies where displacement is achievable: shallow integrations, recent implementations, and customers who have already run an evaluation.
The timeline finding also changes what a challenger should sell against. Where displacement takes years, the winnable segment is the one currently implementing or recently implemented, not the one most unhappy — close to the opposite of where most competitive campaigns are pointed. Reading the intent data as a rate sends effort exactly where the constraint is strongest.