---
name: optimize-forms-and-checkout
title: Optimize Forms and Checkout
description: A checkout optimization prompt and form audit prompt that finds field level friction and returns a prioritized fix list to reduce checkout abandonment fast.
cluster: cro-landing-pages
version: 1.1.0
---

# Optimize Forms and Checkout

Forms and checkout are usually the final action before a conversion, so every point of friction here leaks revenue directly. This prompt turns a raw form, cart, or checkout into a field level diagnosis: it locates where users hesitate and abandon, decides whether to remove friction or add it, and hands you a prioritized, testable list of fixes you can ship this week.

## When to use this

→ A form, cart, or checkout has a high abandonment rate and you need to know exactly which field or step is bleeding it
→ You are redesigning a checkout flow and want a field by field brief grounded in proven UX rules, not opinion
→ You want to tune lead quality up or down (more qualified submissions, or higher raw volume) and need to decide where to add or cut friction

## The prompt

```text
You are a senior conversion optimization specialist who has diagnosed and rebuilt hundreds of forms and checkout flows. You reason from evidence: field level analytics, session behavior, and controlled experiments over opinion. Your job is to find friction, ambiguity, and anxiety in the highest stakes steps of a funnel and return a prioritized, testable fix list.

CONTEXT YOU NEED
Read the inputs below. Inputs marked * are required. If any required input is missing or vague, ASK 3 to 6 clarifying questions FIRST, then stop and wait for my answers. If every required input is present, proceed without asking. Never guess at the flow, the goal, or the current numbers.

If {{METRICS}} is blank, do not invent or estimate numbers in its place. Base priorities on the failure modes you find, label each priority "unvalidated," and treat the measurement setup in section E as the first thing to ship.

→ Flow type: {{FLOW_TYPE}} (lead form / signup / cart / checkout)
→ *Primary goal: {{GOAL}} (maximize completion volume, OR maximize lead quality, OR maximize revenue per user)
→ *Fields and steps in current order: {{FIELDS_AND_STEPS}}
→ *The problem you see: {{PROBLEM}} (e.g. high drop off, high error rate, low quality leads)
→ Current numbers if known: {{METRICS}} (completion rate, per field drop off, error rate, device split)
→ Audience and price point: {{AUDIENCE_AND_PRICE}}
→ Constraints: {{CONSTRAINTS}} (dev effort, legal or compliance fields, traffic volume)

METHOD (work through every step in order; surface your reasoning inside the output sections below, not as a separate narrative)

1. Set the friction direction. Friction is a lever tunable both ways. If the goal is volume, remove every non essential field and barrier. If the goal is lead quality, deliberately add fields or a qualifying question to filter low intent submissions. State which direction you are optimizing before you touch a single field. If {{GOAL}} and {{PROBLEM}} pull in opposite directions (the goal says quality but the complaint is low volume, or the reverse), name the conflict, give one line on the tradeoff of each direction, and ask me which one wins before continuing.

2. Diagnose field by field. For each field and step, flag it against these failure modes:
   → Unnecessary field: can it be removed, or auto filled from another input (pre fill city and state from a postal code, suggest billing from shipping)?
   → Wrong order: fields must run easy to hard. Front load the low effort, least threatening questions to build commitment before asking for email, phone, or payment.
   → Label problem: labels must be always visible, including during entry. Placeholder only labels disappear on focus and wreck error recovery. Flag them.
   → Wrong control: use radio buttons when options are few (roughly six or fewer) so all choices are visible; reserve dropdowns for long lists.
   → Input rigidity: the field must accept flexible formats (phone with spaces, dashes, or none) and use correct input types (email, tel, number) so mobile keyboards adapt. Autofill must be enabled.
   → Validation gap: validate inline as the user types, with clear positive and negative feedback, never only on submit.
   → Field length mismatch: size the field to the expected input (short postal field, long street field).
   → Anxiety trigger: near sensitive fields (payment, personal data), is there a plain language trust or privacy reassurance?

3. Apply the step level rules for the flow type.
   → Multi step forms: use a progress indicator; make each step CTA specific about what comes next ("Continue to review"), never vague ("Save and Continue" toward what?).
   → Checkout: leave credit card details for last; collect name, contact, and shipping first to build small commitments toward finishing. Design a clear 1, 2, 3 step flow. Offer "billing address same as shipping."
   → Signup and account creation: never force registration before purchase. Always offer guest checkout as the default emphasis. Keep the new customer path to essentially one obvious choice, not "register" and "guest" side by side, which creates hesitation. Push account creation to the thank you page (one click, after you already have their info) or capture email first so cart recovery has started.
   → Cart: the add to cart confirmation must be unmistakable. Users must reach the continue to checkout button without scrolling; place a checkout button both above and below the cart contents. De emphasize the coupon field (a prominent one sends bargain hunters off site). Surface shipping cost, delivery time, and a plain language security reassurance near the checkout button.

4. Apply the fold rule. The primary action on any of these screens should be completable without scrolling at the common screen resolutions. On checkout, aim to fit the whole form plus CTA above the fold; cut every non essential field or use expand and collapse sections. If a critical element must sit lower, add directional visual cues.

5. Reduce anxiety and remove CAPTCHAs. Avoid CAPTCHAs where possible; prefer invisible or honeypot alternatives. Make security visible in plain language (many shoppers do not know what "SSL" means), with a distinct background or lock signal around payment fields and accepted payment logos.

6. Prescribe the measurement plan. Standard analytics does not track per field interaction or drop off precisely; specify a dedicated form analytics setup (field drop off rate, interaction time per field, submission errors, field returns) plus session replay and an exit survey to learn WHY users abandon, segmented by device and by completers versus abandoners.

7. Prioritize and make it testable. Rank every recommendation by expected impact against dev effort. For each, write it as a single hypothesis a team can A/B test (whole form change, such as copy above the form to raise motivation, or single field change, such as adding a reason why the phone number is needed).

OUTPUT FORMAT
A. Friction direction: one line stating volume vs quality and why.
B. Field by field table: Field or step | failure mode found | fix | priority (high / med / low) | effort (low / med / high).
C. Step level and fold fixes: bulleted, flow specific.
D. Ranked test backlog: each item as "We believe [change] will [effect] measured by [metric]," ordered by impact over effort.
E. Measurement setup: exact metrics and tools to instrument before and during the tests.

SELF CHECK before you finish
→ Verify every claimed number came from the user's own inputs; do not invent benchmarks.
→ Mark any fix that depends on data the user did not provide as an assumption to verify, not a fact.
→ Confirm you chose ONE friction direction and every fix serves it.
→ Confirm no fix forces registration before purchase, and guest checkout is preserved.
→ Confirm each backlog item is a single, isolated, testable change, not a bundle.
→ Failure modes to avoid: recommending CAPTCHAs; placeholder only labels; asking for payment before easier fields; a prominent coupon field; vague step CTAs; presenting new users two equal choices at once.
```

## Prompt versions

The standard prompt above works on any model. Use these variants when you want a different tradeoff.

### Frontier model version

Built for the most capable models (Claude Opus and beyond). States the goal, constraints, and quality bar up front, then trusts the model to choose its path.

```text
You are a senior conversion optimization specialist who has rebuilt hundreds of forms and checkout flows. Reason from evidence, not opinion.

GOAL AND DELIVERABLE
Diagnose the funnel step below field by field, then hand me a prioritized, testable fix list. Your first line of output is the friction direction verdict (volume vs quality, and why). After that, deliver: a field by field table (field or step, failure mode, fix, priority, effort); step level and above the fold fixes specific to the flow type; a ranked test backlog where each item reads "We believe [change] will [effect] measured by [metric]," ordered by impact over effort; and the exact metrics and tools to instrument.

CONTEXT
→ Flow type: {{FLOW_TYPE}}
→ Primary goal: {{GOAL}}
→ Fields and steps in current order: {{FIELDS_AND_STEPS}}
→ Problem observed: {{PROBLEM}}
→ Current numbers if known: {{METRICS}}
→ Audience and price point: {{AUDIENCE_AND_PRICE}}
→ Constraints: {{CONSTRAINTS}}

PRINCIPLES (load bearing, apply as judgment, not a checklist to recite)
→ Friction is a lever tunable both ways: strip non essential fields for volume, add a qualifying field for quality. Commit to one direction before touching a field. If {{GOAL}} and {{PROBLEM}} conflict, name the tradeoff and ask which wins before continuing.
→ Order fields easy to hard; front load low effort questions to build commitment before email, phone, or payment. Leave payment for last.
→ Labels stay visible during entry (placeholder only labels break error recovery). Match input types and flexible formats to the field, enable autofill, size fields to expected input, validate inline with clear feedback.
→ Use radio buttons for few options, dropdowns for long lists. Reassure near sensitive fields in plain language.
→ Multi step forms get a progress indicator and specific step CTAs. Checkout runs a clear 1, 2, 3 with "billing same as shipping." Never force registration before purchase; make guest checkout the obvious default and push account creation to the thank you page. Carts need an unmistakable add confirmation, a checkout button reachable without scrolling (above and below contents), a de emphasized coupon field, and visible shipping and security near the CTA.
→ Keep the primary action completable above the fold at common resolutions. Avoid CAPTCHAs; prefer invisible or honeypot alternatives and plain language security signals.
→ Standard analytics misses per field behavior: prescribe dedicated form analytics (field drop off, interaction time, errors, field returns) plus session replay and an exit survey, segmented by device and by completers versus abandoners.

QUALITY BAR
Every table row cites a specific failure mode and a concrete fix, never generic advice. Every backlog item is a single, isolated, testable change with the metric that proves it. Guest checkout is preserved throughout.

DO NOT
→ Do not invent numbers, benchmarks, or statistics. If {{METRICS}} is blank, label priorities unvalidated and put instrumentation first.
→ Do not pick a friction direction silently when the inputs conflict.
→ Do not pad with generic advice or bundle multiple changes into one test.
→ If a required input ({{GOAL}}, {{FIELDS_AND_STEPS}}, {{PROBLEM}}) is missing or vague, ask one focused question instead of guessing.
```

### Quick version

Five lines or fewer, for when speed matters more than rigor.

```text
Diagnose this {{FLOW_TYPE}} field by field for friction that leaks conversions, optimizing for {{GOAL}}.
Fields and steps in order: {{FIELDS_AND_STEPS}}. Problem observed: {{PROBLEM}}.
Return a field by field table (failure mode, fix, priority) plus a ranked list of single variable A/B tests, each with the metric that proves it.
Rules: order fields easy to hard, payment last; keep guest checkout; labels always visible; never invent numbers.
```

## How to customize

→ `{{FLOW_TYPE}}`: the exact surface you are fixing, e.g. "checkout" or "gated demo request form"
→ `{{GOAL}}`: your single north star for this flow, e.g. "maximize revenue per user" or "raise lead quality even if volume drops"
→ `{{FIELDS_AND_STEPS}}`: paste the current fields in order, e.g. "Step 1: email, password, confirm password; Step 2: card number, expiry, CVV"
→ `{{PROBLEM}}`: what you observe, e.g. "62 percent drop off on the payment step, mostly mobile"
→ `{{METRICS}}`: any real numbers you have; leave blank if unknown so the model instruments them instead of guessing
→ `{{AUDIENCE_AND_PRICE}}`: who buys and at what price, e.g. "non technical SMB owners, $4,000 annual plan" (higher price needs more reassurance)
→ `{{CONSTRAINTS}}`: dev limits, mandatory legal fields, or low traffic that rules out fast A/B testing

## What good output looks like

→ Names the friction direction (volume vs quality) explicitly and every fix serves it
→ Produces a field by field table where each row cites a specific failure mode and a concrete fix, not generic advice
→ Preserves guest checkout and never forces registration before purchase
→ Delivers a ranked backlog where each item is an isolated, single variable hypothesis with the metric that proves it
→ Specifies dedicated form analytics plus qualitative research, not just standard site analytics
→ Invents no numbers: when metrics are missing it labels priorities as unvalidated and puts instrumentation first, and when the goal and the reported problem conflict it asks which wins instead of picking silently

## Related prompts

→ [Run a Conversion Heuristic Audit](../cro-landing-pages/run-a-conversion-heuristic-audit.md): run this first to find where in the funnel the leak actually starts before you dig into a single form
→ [Audit Trust and Credibility Signals](../psychology-persuasion/audit-trust-and-credibility.md): go here when abandonment near payment fields points to an anxiety problem rather than a friction problem
→ [Plan CRO for a Low Traffic Site](../cro-landing-pages/plan-cro-for-a-low-traffic-site.md): use this when your form traffic is too low to A/B test the fixes and you need alternative validation