form-friction
Skill0 starsChanged 2 months agoMITView on GitHub
Find where a lead form or checkout loses people, and size the loss against the cost of the traffic feeding it. Distinguishes a form problem from a traffic problem before anyone rewrites copy. Use when conversion is below expectation, or before increasing spend on a page that already receives traffic.
In this repository9 other files it gives its agents.
- activationSkill
- capacity-economicsSkill
- channel-fitSkill
- data-trustSkill
- experiment-designSkill
- funnel-diagnosisSkill
- pricing-decisionsSkill
- retention-diagnosisSkill
- unit-economicsSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 17
Ask for three numbers:
- Distinctive: line 56
Fields that are not needed to make the sale: Every field is a chance to stop.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Form friction
The most expensive conversion problem is usually not the headline, the offer, or the audience. It is that a meaningful share of people who decided to convert and started filling something in did not finish.
Those people are the cheapest customers you will ever get, because you have already paid for them.
Size it before diagnosing it
Ask for three numbers:
- starts — people who interacted with the first field
- completions — people who submitted
- cost of the traffic that produced the starts
Then:
completion rate = completions / starts
abandons = starts − completions
implied waste = abandons × cost per startimplied waste is the number that gets a form fixed. A completion rate in the
20-30% range on a short form usually means something is broken, not that
visitors were unserious — but treat that band as a prompt to investigate, not a
diagnosis, because acceptable rates vary enormously by form length, price point
and how qualified the traffic is.
If starts are not instrumented, that is the first finding. Submit-only tracking cannot distinguish a form nobody reached from a form everybody abandoned, and those have opposite fixes.
Separate the form problem from the traffic problem
| Pattern | Reading |
| Many starts, few completions | Form problem. Fix the form |
| Few starts relative to sessions | Page or offer problem. The form is innocent |
| Completions fine, sales poor | Neither. Look downstream at qualification or follow-up |
Running a copy test on a page whose form is the constraint wastes the test. Establish which of the three you are in before recommending anything.
Where forms actually lose people
Ordered by how often it turns out to be the cause:
- Fields that are not needed to make the sale. Every field is a chance to stop. Ask what breaks if it is removed, and remove it if nothing does. The sales team wanting it is not the same as the sale needing it
- Mobile input mismatch. A numeric field that summons a full keyboard, a date picker that fights the OS, a submit button under the keyboard. Where most traffic is mobile this dominates everything else, so check the mobile completion rate separately — a healthy blended rate can hide a broken one
- Validation that punishes. Errors shown only on submit, a cleared form after a failure, a phone format rejected without explanation
- Silent failure. Submissions that never arrive, or a success screen that fires while the write fails. Test the full path end to end with a real submission and confirm it reached the system of record
- Trust asked for too early. Payment or identity details before the value is clear
- Unexplained cost. A price, fee or commitment first revealed at the last step
Mobile is a separate funnel
Report completion by device always. They are different products with different failure modes, and the fix for one frequently does nothing for the other.
Verify the instrument before believing the number
A collapsed completion rate is as likely to be a tracking change as a behaviour
change. Before diagnosing, confirm the start and completion events still fire,
still fire once, and that two similarly-named events have not been merged into
one. See data-trust.
When to refuse
- Starts are not tracked, so abandonment cannot be separated from non-arrival
- Fewer than ~100 starts in the window — completion rates on small samples swing wildly
- The tracking changed inside the comparison window
- Device split unavailable and most traffic is mobile
Output
Starts, completions, completion rate overall and by device, abandons, and implied waste in currency. Then the single most likely cause from the ordered list with the evidence for it. Then the one change to make first, and what you expect it to recover — stated as a range, not a point estimate.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

