By Mehdi HassanHow should an AI company structure its use case pages?
An AI company should structure each use case page around one job for one buyer, with sections phrased as the questions that buyer asks before they try the product. Each section opens with a direct answer, then shows detail and proof. That shape serves people scanning the page and AI answer engines quoting it.
Use case pages do more selling than most AI founders expect. TripleDart's study of 28 B2B SaaS companies found homepages earned 31.9% of LLM citations, which is why we'd put pages that answer buying questions ahead of blog volume. This article gives you a template you can copy. For where these pages sit in the wider site, see our hub on AI startup website design.
What is the difference between use case, solutions and product pages?
A product page explains what the product is, a solutions page speaks to an industry or company type, and a use case page covers one specific job. AI companies need all three, but use case pages are the ones that match how buyers search and ask: "AI for summarizing sales calls", not "AI platform".
Page type | Organized by | Answers | Example |
|---|---|---|---|
Product | Capability | What is it and how does it work? | Platform overview |
Solution | Industry or company type | Is it built for teams like mine? | AI for insurance carriers |
Use case | Job to be done | Can it do this specific task for me? | Drafting first notice of loss summaries |
Role | Buyer or user role | What does it do for my job? | For claims managers |
Our guide to solutions pages vs product pages vs use case pages covers navigation and how to stop these from duplicating each other.
What is the template for an AI use case page?
The template below has seven sections, each phrased as a buyer question. Use it as a CMS structure, with one field per section, so every page follows the same pattern and new pages are quick to add.
Hero: what does it do for [buyer]?
One line naming the job and result, a subhead with input and output, and a product view showing this exact use case.
How does it work?
Three steps from the buyer's side: connect, run, review. Name the systems it reads from and writes to.
What goes in and what comes out?
A side-by-side example of real input and AI output in the buyer's format. This is the most important block on the page.
Where does a person stay in control?
The review and approval step, what users can edit, and what the AI won't do.
How accurate is it, and how do you know?
Your testing method and any results you can stand behind, or a plain description of where it needs review.
What does it cost for this use case?
The plan that fits and roughly how much usage this job consumes, linked to pricing.
How do I start?
A CTA tied to the use case, like "Try it on five of your own claims", plus a short FAQ.
What should the input and output section show?
The input and output section should show a realistic example of what the user gives the AI and exactly what they get back, side by side, in the format they'd really use. It turns an abstract claim into something the buyer can judge with their own expertise.
- Use realistic, anonymized data. A real-looking contract clause, call transcript or support ticket, never lorem ipsum.
- Show the full output, including citations, flags and confidence markers, not a cropped best case.
- Include one imperfect case where the AI flags uncertainty. It makes the good cases more believable.
- Make it interactive when you can. A toggle between examples or a small live demo, if your product allows it safely.
Hypothetical example: a filled-in use case page
For example, suppose a hypothetical AI product for insurance carriers builds a page for "First notice of loss summaries".
Hero: "Every first notice of loss summarized for the adjuster in minutes." Subhead: "We read the call transcript, photos and policy, then draft a summary with coverage questions flagged."
Input and output: a short anonymized transcript on the left; on the right, the summary with date, location, parties, damage description, two flagged coverage questions and links back to the transcript lines.
Control: "Adjusters edit and approve every summary. We never make coverage decisions." Accuracy: "We test on anonymized claims reviewed by senior adjusters; here's the method." Start: "Run it on ten of your own claims."
How should AI use case pages present metrics?
AI use case pages should present metrics only with the method beside them: what was measured, on what data, against what baseline and when. A number without a method reads as marketing, and with AI products it can also read as AI-washing.
Claim type | OK to publish when | Safer alternative |
|---|---|---|
Time saved | Measured with real customers, with permission | Describe the steps the AI removes |
Accuracy rate | Defined task, test set and scoring method shown | Describe the review step and what gets flagged |
Volume handled | A real production figure you can verify | Say what volume the product is built for |
Cost reduction | A customer agreed to share it, with context | Show the pricing for this use case |
Should AI use case pages be built programmatically?
Build AI use case pages from one template, but write each page's core content by hand. A template with structured fields keeps pages consistent and cheap to add. Fully generated pages with only the job name swapped are thin content that ranks poorly and convinces no one.
Pros
- One design and build, then each new page is a CMS entry
- Consistent structure that buyers and AI engines learn to scan
- Easy to update sitewide when pricing or product views change
Cons
- Tempting to fill fields with near-identical copy
- Needs a real input and output example per page, which takes product time
- Too many weak pages can dilute the site's overall quality
Google's guidance is that AI search features run on ordinary SEO, so there's no special schema or file format that makes these pages more quotable. Clear answers, real examples and good internal linking do the work. Customer stories can sit beside use cases; see our SaaS customer story page template.
How many AI use case pages do you need?
Start with three to five use case pages: the jobs your best customers actually bought you for. Add more only when sales calls, search data or product usage show buyers looking for a job you can genuinely do well.
Link each use case page from the homepage or product page, from related use cases, and from the pricing page. Our list of must-have pages for an AI startup website shows where use cases fit in the build order. For layout ideas on technical products, see these developer tool landing page examples, and for service-led AI offerings, B2B services landing pages.
