The Difference Between Detecting Friction and Resolving It

Most teams have gotten pretty good at detecting friction. The tools exist, the dashboards are populated, the heatmaps are running. If a customer struggles somewhere on your site, you will probably find out about it. The question is when, and what happens next.

Detecting friction and resolving it are not the same thing. They are not even close to the same thing.

What does friction detection actually give you?

Evidence that someone struggled and where. Analytics, session replay and heatmaps all do this well, and each still leaves a human to interpret what it captured.

Detection gives you a report. A session replay shows a subscriber hesitating on a cancel page. An analytics platform flags a spike in cancel-page visits. A funnel report marks it as a drop-off. Someone schedules a meeting to talk about churn.

This is all useful. But while the report is being generated, the subscriber has already made their decision. The moment has passed.

Detection is backward-looking by design. It tells you what happened across many users over some period of time. The patterns it surfaces are real, and acting on them makes future experiences better. That is worth doing. But it does nothing for the person who was stuck today.

What is the gap between detection and resolution?

Diagnosis. A signal shows that a customer is stuck but not why, and the same signal at the same step can have several causes that need different responses.

There is a gap between "we know friction exists here" and "we did something about it while the customer was still present." That gap is where customers are lost.

Think about the cancel-page moment. Analytics sees the visit. Session replay shows the hesitation. Both systems are doing exactly what they are built to do. But neither of them acts. They observe. The subscriber sees a standard cancel flow with no acknowledgment that anything is unusual about this visit.

If Pulse is on that page, the experience looks different. It detects the cancel-page visit, asks one question: "What would make staying easier?" or "What brought you here today?" The subscriber picks an option, and Pulse delivers a pre-approved response based on that answer. A billing concern gets a billing path. A feature complaint gets a feature acknowledgment or a roadmap note. An "I never used it" gets a re-onboarding prompt.

The detection is still there. The data still flows. But now something also happened in real time for that specific person.

What are the limits of detection-only tooling?

It can only improve the next customer's experience, and only after someone reviews, decides and ships. Neither ceiling moves by detecting more accurately.

Detection has a ceiling. Its ceiling is insight. A very good detection system gives you very good insight into where your experience breaks down and for whom. That is genuinely valuable.

Resolution has a different ceiling. Its ceiling is what happens next for the customer. Did they continue the journey? Did they move from stuck to moving? Did the cancel-page visitor decide to stay, or at least understand why they were leaving?

Neither ceiling is higher than the other by default. A team with excellent insights that actually acts on them will outperform a team doing lazy resolution with poor targeting. But when both exist, the combination is powerful: you detect at scale and resolve at the moment.

The pattern described in "The Anatomy of a Stuck Moment" captures this well. Stuck moments have a structure: a customer who has intent, hits an obstacle, and faces a choice between continuing and leaving. Detection tells you the obstacle is there. Resolution is what happens at the moment of that choice.

What makes resolution different from detection?

The response happens inside the session that produced the signal, and it is selected from what the customer said rather than from what the pattern implied.

Resolution requires a few things detection does not. It requires a signal that fires while the customer is still present, not after. It requires a question that is specific enough to be diagnostic and short enough to actually get answered. And it requires a pre-approved response for each answer, something the organization has already decided is the right thing to say in that case.

That last part is where most teams are underprepared. The question is easy. The diagnosis is easy. The response library is the work. But once it exists, resolution becomes repeatable. You apply the same pattern across every journey where customers get stuck: pricing confusion, onboarding drop-off, checkout hesitation, cancel-page visits. One model, many moments.

The measurement side is straightforward: did the customer who got a response continue the journey at a higher rate than the one who did not? If yes, the resolution is working. The "Stop Analyzing Friction. Start Resolving It." post goes further into what it takes to make that measurement meaningful.

Detection without resolution is a very good alarm system in a building where nobody is quite sure who is supposed to respond. Resolution adds the responder. Together, they handle the moment.

Frequently asked questions

What is the difference between detecting friction and resolving it?

Detection notices that a customer is struggling. Resolution does something about it in the same session. Most of the category detects well and stops there.

Which tools detect friction?

Analytics, session replay and heatmap tools all detect it, each in their own way. They differ on fidelity and on how much analyst time it takes to interpret what they captured.

What has to happen between detection and resolution?

Diagnosis. A behavioral signal shows that someone is stuck but not why, and the same signal at the same step can have several different causes that need different responses.

Can detection alone improve conversion?

Indirectly, and on a delay. Detection informs a change that ships later and helps future visitors. It does nothing for the customer whose session produced the signal.

How quickly does a resolution have to happen?

Inside the session, which in practice means seconds. Once the customer has left there is no interface left to help them in.

For the detect, diagnose and resolve breakdown as an evaluation framework, see customer intervention platform.