Back to News Hub
GEO & Tactics November 17, 2026

ChatGPT Follow-Up Forms: A Step-by-Step SEO & AEO Guide

ChatGPT Follow-Up Forms: A Step-by-Step SEO & AEO Guide

ChatGPT is now asking buyers what they want before it recommends anything

Ask ChatGPT a commercial question like "best ATS software" and it may not just give you a list. More and more often, it ends its answer with a short questionnaire. You get three to five questions with radio-button answers about budget, company size, use case or priorities.

SEOs on LinkedIn, including John-Henry Scherck, have started calling it the "User Requirement Box." It's no longer limited to physical products. It shows up for software, services and other B2B prompts too.

This matters because the form turns a vague, top-of-funnel question into a detailed, bottom-of-funnel one. ChatGPT then answers that new prompt. If your content doesn't speak to the requirements the form asks about, you'll have a hard time showing up in the answer the user actually reads.

This guide explains what the follow-up form is, how it works, and how to optimize your site for it step by step.

What is ChatGPT's follow-up form (the "User Requirement Box")?

It's a short questionnaire ChatGPT can add to the end of an answer so the user can narrow down what they need. Cloro, an AI-visibility API provider, describes it as a refinement prompt that ChatGPT can end an answer with. Its example asks about budget, experience and priority, each with radio options, followed by a button to find a product (Cloro).

The idea isn't brand new. When OpenAI launched Shopping Research, ChatGPT already asked follow-up questions about preferred brands, size ranges, or whether the user cares more about performance, comfort, style or price (OpenAI Help Center). What's new is how widely it now appears:

  • In regular chat. Users no longer have to switch on a separate shopping mode.
  • For non-physical products. SaaS, services and B2B vendors are seeing it. They were largely left out of the shopping experience.
  • More often. Practitioners report seeing it on a growing share of commercial prompts.

That last point about non-physical products is the big one. Profound's analysis of ChatGPT Shopping found that if a prompt's main noun describes a service, a software platform or a concept, it probably won't trigger a Shopping card (Profound). Follow-up forms bring the same move from vague question to specific requirements to those categories.

How the follow-up form works

The form is a template. Each answer fills a slot in a new, more specific prompt, and ChatGPT runs that prompt as soon as the user submits.

  1. The user asks a broad commercial question. For example: "What's the best ATS software?"
  2. ChatGPT answers and adds a form. It's usually three to five questions. Each one has a label, an input type and a set of options. Cloro's documentation lists radio, checkbox, select, text, textarea and date inputs (Cloro).
  3. The user picks options. Each option has a visible label and a value that goes into the prompt. The two can differ.
  4. ChatGPT re-prompts itself. The form carries a hidden follow-up prompt template. Submitting the form fills in each placeholder with the user's answers and sends the result.
  5. A new, narrower answer is generated. That new prompt triggers its own searches (query fan-out), so the sources and brands ChatGPT cites can change completely.

A real example

Here's the kind of re-prompt the form produces for an ATS query:

"Recommend the best ATS software based on these requirements: Hiring employees for my company; company size: 100–500 employees; priority: AI resume screening and candidate matching; budget: …"

Compare that to the original "best ATS software." The new prompt names a segment (mid-market), a must-have feature (AI screening and matching) and a price limit. Each one is a filter your content either matches clearly or doesn't.

Why follow-up forms matter for SEO and AEO

  • The form shows you, in ChatGPT's own words, which criteria it thinks buyers in your category use to decide. That's rare, first-hand insight into how an answer engine breaks down intent.
  • More specific prompts mean more specific fan-out. Every submitted form creates a long-tail, requirement-heavy prompt. That prompt sends its own searches, so the pages ChatGPT retrieves and cites can be very different from the ones it used for the broad question.
  • The funnel gets compressed. ChatGPT pushes users from "what's out there" to "what fits my exact situation" in one click. Brands that clearly match a segment and use case win that moment.
  • Vague copy loses. A page that says "powerful, scalable, for teams of all sizes" doesn't match "100–500 employees, AI screening, under $X." A page with explicit segments, features and pricing does.
  • The stakes are high. HubSpot, citing Similarweb, reports that AI referral traffic converts at 11.4% in ecommerce, compared with 5.3% for organic search (HubSpot). Prompts this far down the funnel are where that conversion edge is likely to be largest.
  • It's measurable. Tracking tools already capture the form as structured data, so you can watch which questions show up for your category over time.

Step-by-step: how to optimize for ChatGPT's follow-up forms

Step 1: Build a prompt set for your category

Write 20–30 commercial prompts your buyers would really type. Mix broad prompts ("best CRM"), "for" prompts ("best CRM for real estate agents") and comparison prompts ("HubSpot vs Pipedrive"). Use the language you hear in sales calls, support tickets and reviews, not your keyword tool's phrasing.

Step 2: Run each prompt several times and capture the forms

The questions are generated fresh each time, so one screenshot proves nothing. Run every prompt at least three to five times over a couple of weeks, logged out and logged in, and from your main markets. Record every question, every option and the final re-prompt. A tracking tool that captures the form as data will save you hours.

Step 3: Build a requirements map

Put everything into one spreadsheet with one row per question. Note how often it appears and which options keep coming back. You'll usually see the same five or six criteria:

Criterion Typical options ChatGPT offers Where your site should answer it
Company size / segment 1–10, 11–50, 100–500, enterprise ICP or segment pages, case studies
Use case / goal Hiring, onboarding, reporting Use-case and solution pages
Must-have feature AI screening, integrations, mobile app Feature pages, help center docs
Budget Free, under $X/month, custom pricing Pricing page with clear tiers
Industry Healthcare, retail, agencies Industry pages
Experience / skill level Beginner, intermediate, expert Onboarding docs, comparison pages

Step 4: Audit your content against the map

For each criterion and common option, ask: is there a page that clearly states we fit this, in plain language a machine can pull out? Mark each cell as covered, weak or missing. Watch out for three common gaps: pricing that isn't public, "for teams of all sizes" with no segment detail, and features only mentioned in a demo video.

Step 5: Close the gaps with real substance

  • Segment and ICP pages: say who you serve ("built for 100–500 employee companies"), what's different at that size, and add proof such as case studies, logos and numbers.
  • Feature pages: one clear page per must-have feature. Explain what it does, how it works and its limits.
  • Pricing: publish tiers or at least ranges and what each includes. If ChatGPT can't find a price, it can't match a budget filter.
  • Help center documentation: this is where specific capabilities (integrations, compliance, limits) are written down. Keep it public and crawlable.
  • Comparison pages: write honest "X vs Y" and "alternatives" pages that state which buyer each option suits best.

Step 6: Write so answers are easy to pull out

Use question-style headings that match the criteria ("How much does [product] cost for a 200-person company?"). Answer in the first one or two sentences under each heading. Use tables for specs and pricing, and keep facts consistent across your site, G2/Capterra profiles and directories.

Step 7: Add structured data and keep pages crawlable

Add schema where it fits: Organization, Product or SoftwareApplication with Offer for pricing, FAQPage, and HowTo. Make sure key pages aren't hidden behind JavaScript-only rendering, logins or blocked crawlers. AI retrieval still depends on crawlable, indexable pages.

Step 8: Re-test and iterate monthly

Re-run your prompt set every month. Track two things: whether your brand appears in the answers that come after the form, and whether the questions themselves change. A new criterion in the form is an early signal of a content gap.

Worked example: a mid-market ATS vendor

Imagine an applicant tracking system (ATS) vendor with a strong homepage but thin product content. Here's how the process plays out.

What the forms reveal. Across 25 prompts run five times each, ChatGPT keeps asking about four things: company size (with 100–500 employees coming up often), top priority (AI screening, onboarding, compliance), budget band, and whether the buyer is hiring for one location or several.

What the audit finds.

Criterion Current coverage Fix
100–500 employees Weak: "for growing teams" A mid-market page with two case studies and named limits
AI resume screening Missing: only in a demo video A dedicated feature page and a help-center article on how matching works
Budget Missing: "contact sales" Public pricing ranges per tier
Multi-location hiring Covered in help docs only Linked from the mid-market page, with an FAQ on the pricing page

The result to look for. When the re-prompt reads "100–500 employees; priority: AI resume screening; budget: under $X," the vendor now has a page that matches each part directly. That's the page most likely to be retrieved and cited.

Common mistakes to avoid

  • Building a page for every radio button. A page per option ("ATS for 101–250 employees," "ATS for 251–500 employees") becomes thin doorway content fast. Build one strong page per real segment or use case.
  • Trusting a single screenshot. The questions change between runs, accounts and model updates. Only patterns seen repeatedly are worth acting on.
  • Claiming fit you can't back up. If you add "perfect for enterprise" without proof, the claim will clash with reviews and third-party sources. Answer engines compare sources.
  • Hiding the answers. Pricing behind a form, specs in PDFs and features in videos are all hard for ChatGPT to match against a requirement.
  • Forgetting third-party sites. ChatGPT doesn't only cite your site. Keep your review profiles, directories and partner listings in line with the same segment and feature claims.
  • Ignoring traditional SEO. Crawlability, indexation, internal links and authority still decide whether your pages are eligible to be retrieved at all.

How to measure impact

Success means showing up in the answer that comes after the form, not just the first one. Track these monthly:

Metric What it tells you How to collect it
Form trigger rate How often your category's prompts produce a form Prompt-tracking runs
Criteria frequency Which requirements ChatGPT cares about most Your requirements map
Post-form mention rate % of re-prompted answers that name your brand Re-run with common option combinations
Citation share Which of your URLs get cited after the form AI visibility tools or manual checks
AI referral sessions and conversions Whether visibility turns into pipeline Analytics: filter referrers like chatgpt.com

Tip: test the option combinations your ideal customers would choose, not every combination. Ten well-chosen combinations will tell you more than a hundred random ones.

Sources

  • Cloro: ChatGPT follow-up form documentation
  • OpenAI Help Center: Commerce policies (shopping research)
  • Profound: Predicting when ChatGPT Shopping triggers
  • HubSpot: Answer engine optimization trends in 2026

Frequently Asked Questions

It's a short questionnaire ChatGPT can add to the end of an answer to commercial prompts. It asks about criteria like budget, company size and priorities, then re-prompts ChatGPT with the user's selections.

Not exactly. Shopping Research already asked clarifying multiple-choice questions inside its dedicated mode. Follow-up forms bring a similar step into regular chat, including for software and services.

Mostly broad commercial prompts where the best answer depends on the buyer's situation, like "best ATS software" or "best project management tool." Prompts that already include detailed requirements are less likely to need one.

Not directly. The questions are generated based on how ChatGPT understands buying decisions in your category. You can influence how well you match the answers by publishing clear segment, feature and pricing information.

Sometimes. Most sites mainly need to make existing pages more specific. Create new pages only for real segments, use cases or features that have no page today.

Keyword research tells you what people type. Follow-up forms tell you how an answer engine splits up the decision behind what they type. Use both.

Related Reading