Friction is effort a person spends that does not move them toward what they came for. It is measurable: count the steps, the fields, the decisions and the waits between intent and outcome, and watch where people stop. Used loosely it explains everything and predicts nothing, which is how teams end up removing fields from a form that was never the problem.
This is how to measure it, where it accumulates, and the more important half: how to tell friction from a missing reason to act, because they look identical in a funnel and need opposite fixes.
The newsletter
Join our KISS newsletter
One short read a week on what actually moves revenue, in a free email. Read by 10,000+ operators and founders.
No spam. Unsubscribe in one click.
I.Friction is a measurable thing, not a metaphor
Define it as effort that does not advance the person, and it becomes countable. Leave it undefined and it becomes an excuse.
A.The definition that makes it useful
Friction is effort a person spends that does not move them closer to what they came for. The second clause is the whole definition. Effort that does advance them is not friction: choosing between two plans is work, and it is work they need to do.
This matters because the folk version, friction is anything that takes effort, licenses removing things people need. Teams strip a pricing comparison down until nobody can tell the tiers apart, then wonder why conversion fell.
Ask of any step: does completing it get them closer, or does it only get them past it? Address verification gets them closer to a delivery. A newsletter checkbox during checkout gets them past a thing you put there.
B.It accumulates, which is why the drop-off lies
The most common analytical error here: assuming the step where people leave is the step that lost them.
Friction is cumulative. Someone tolerates a slow page, then an unexpected required field, then a plan choice they do not have the information to make, and abandons at step four. Step four gets the blame and is often the least frictional of the set. It was simply where the budget ran out.
The tell is that fixing the last step moves nothing. If you have optimised a drop-off point twice with no effect, the cause is upstream.
Where it is spent versus where they leave
Funnels viewII.Finding where it accumulates
Four kinds, four signatures. Naming which one you have is most of the fix.
A.Mechanical: the work is literally the work
Fields, clicks, steps, typing. The easiest to count and the one everybody starts with, which is why it is usually already the least of your problems.
Signature: abandonment concentrated on the longest forms, sharply worse on mobile, and worse again on any field requiring information the person does not have to hand. See form optimisation.
B.Cognitive: they cannot work out what to do
More expensive and less visible. A choice with no basis for choosing, a term nobody outside the company uses, a price that cannot be estimated.
Signature: long dwell with high interaction and no progress. People expanding comparison tables, moving between two options repeatedly, opening a help link and coming back. They are trying, which distinguishes this from indifference.
C.Trust: they can, and they are not sure they should
The request feels disproportionate, or the page asks for something before earning it. A card required for a trial, a phone number to see a demo, a security question with no answer on the site.
Signature: the detour. Someone leaves the flow for the security page, terms, or a review site, and either returns or does not. The path names the doubt, which is the same diagnostic as the pricing page stall.
4.Wait: nothing is happening and they do not know why
Load time, but also processing states with no feedback, verification emails that take four minutes, and approvals that happen offline.
Signature: abandonment that correlates with duration rather than with any step. The version measured per person rather than per session is in page speed and revenue.
The four, and how to tell which one you have
Diagnosis view| Kind | Signature | Fix | Common mistake |
|---|---|---|---|
| Mechanical | Long forms, mobile-heavy drop | Remove or defer fields | Removing needed ones |
| Cognitive | High interaction, no progress | Make the choice decidable | Adding more explanation |
| Trust | Detour to security or terms | Answer it in the flow | Adding testimonials |
| Wait | Correlates with time, not step | Speed, or show progress | Optimising the wrong page |
III.When it is not friction at all
Friction means they wanted to and could not. Most of what gets called friction is people who could and did not want to.
A.The distinction that decides the whole diagnosis
Two people abandon the same form. One could not complete it. One completed it in their head, decided it was not worth it, and closed the tab.
They are indistinguishable at the step and need opposite work. Removing a field helps the first and does nothing for the second, who does not want the thing at any number of fields.
Friction work has an upper bound: it recovers the people who were trying. If most of your abandonment is the second kind, a perfectly frictionless flow moves your conversion rate by very little, and teams routinely discover this after a quarter.
B.Telling them apart by what came before
The evidence is upstream of the drop-off, not at it.
Someone who read three pages, returned twice and compared two plans was motivated. If they stopped, something stopped them, and friction work will recover a share of them.
Someone who arrived from a broad social ad, hit the form immediately and left was never in the market. Their abandonment is a targeting result, and no amount of form work touches it.
This is why the diagnosis needs the person's history rather than the session. In a session view both look like a form abandonment.
C.Sizing the recoverable half before you spend the quarter
Before optimising anything, split abandoners by prior engagement: number of prior visits, pages read, whether they compared options. The engaged group is the friction-recoverable population, and the unengaged group is an acquisition question wearing a conversion costume.
Kissmetrics keeps the full ordered history per person, so that split is a population rather than an inference, and you can open it and look at what the engaged abandoners actually did before stopping. That is usually where the specific friction names itself.
The honest limit: behaviour tells you they hesitated, not what they were thinking. Signatures narrow it to one of four, and the last step is still asking five of them.
Verdict
Friction is effort that does not advance the person. Defined that way it is countable, it comes in four kinds with four different fixes, and it accumulates, which means the step where people leave is frequently not the step that lost them.
Before removing anything, split abandoners by whether they were engaged beforehand. Friction work recovers the people who were trying and does nothing for the people who were not, and knowing which group is bigger is the difference between a quarter well spent and a flow with no fields left in it.
Continue Reading
Form Optimization: How to Reduce Friction and Increase Completions
Every form field you add reduces completions by 3-5%. But removing fields means less data. This guide shows you how to balance data collection with conversion using progressive profiling and smart defaults.
Read articleMicro-Funnels: How to Find and Fix Hidden Bottlenecks in Your Customer Journey
Your main funnel shows the big picture. Micro-funnels reveal the hidden friction points within each stage. This technique consistently uncovers the small fixes that produce outsized improvements.
Read articleWhere Qualified Leads Stall on Your Pricing Page
Someone on your pricing page has already decided they want it. The ones who leave are pre-qualified and stuck on something specific, and page metrics cannot tell you which of the four it is.
Read article