Form diagnostic
Find form friction without guessing at conversion lift
A short form is not automatically a good form. The useful question is whether every request earns its place at this step and whether a visitor can recover when something goes wrong.
Observe first
Facts to collect before changing anything
These are public-page facts, not performance conclusions.
- 1The visible deliverable promised before the form
- 2Every field, required state, input type, and helper label
- 3Whether sensitive or high-effort information is requested before trust is established
- 4Inline validation, error wording, and whether entered values survive an error
- 5The confirmation, response time, and follow-up described before submission
Decision artifact
The five-check form-friction table
Use the warnings to isolate a reviewable problem. Field count alone is not a diagnosis; a necessary field can reduce wasted follow-up even when it adds effort.
1. Value exchange
Inspect
Read only the copy immediately above the first field and name what the visitor receives.
Healthy signal
The deliverable, format, and relevant eligibility are clear before data is requested.
Warning
The form asks for contact details to “get started” without naming the result.
Next move
State the immediate deliverable and remove any outcome the process cannot guarantee.
2. Field necessity
Inspect
For each required field, name the decision or delivery step that uses it now.
Healthy signal
Every required value changes routing, eligibility, personalization, or delivery.
Warning
Fields are required because they may be useful to sales later, not to complete this step.
Next move
Defer nonessential enrichment or explain why the information is needed now.
3. Sensitive-field timing
Inspect
Mark requests involving phone, budget, revenue, identity, credentials, or private URLs.
Healthy signal
Sensitive requests appear after purpose, handling, and necessity are visible.
Warning
High-trust information is required before the visitor sees process or privacy context.
Next move
Move the request later, make it optional, or add the specific handling explanation.
4. Error recovery
Inspect
Trigger an empty, malformed, and server-error state without submitting real personal data.
Healthy signal
The exact field is identified, valid entries persist, and the visitor can retry.
Warning
A generic error clears the form or leaves the failing value ambiguous.
Next move
Keep valid input, focus the failed field, and say how to correct or retry it.
5. Completion expectation
Inspect
Compare submit copy, consent text, confirmation state, and actual follow-up channel.
Healthy signal
The visitor knows whether submission starts an email, scan, call, account, or review.
Warning
The form implies immediate access but routes to an unspecified manual follow-up.
Next move
Name the real next step and timing only when the operating process supports it.
Original worked example
Remove an unearned request, not every field
A public-page review needs an email for delivery and a public HTTPS URL for the scan. Company size is collected only for future sales sorting.
Before
Required: name, email, phone, company, company size, role, website, budget, message
Reviewable hypothesis
Required: email, public HTTPS URL, processing acknowledgment
The revised set maps each value to scan delivery or consent. It does not claim fewer fields will increase conversion, and a different workflow may legitimately need more.
A bounded next move
- 1.Write the form's promised deliverable in one literal sentence.
- 2.Map each required field to a current decision or delivery dependency.
- 3.Test validation and recovery without using real sensitive information.
- 4.Align submit copy and confirmation with the actual follow-up path.
Limits
- This diagnostic cannot see abandonment rate, lead quality, sales acceptance, or causal lift.
- Regulated, payment, support, and qualification workflows may need additional fields.
- Do not submit another person's data or probe authenticated/private forms during review.
Check the live page, not just the pattern
Growth Scan observes a bounded set of public-page facts and returns one priority finding with assumptions, confidence, and limitations.
Continue the diagnosis