AI assistants answer pricing questions whether or not you publish prices, so the real choice is between your pricing page and someone else's guess. A clear, crawlable pricing page gives assistants an accurate source to quote and lets you appear in budget-limited prompts. This post covers what a pricing page needs, how to handle quote-based pricing, the technical traps, and how to keep the rest of the web in sync.
What happens when AI can't find your price
When an assistant can't find an authoritative price, it fills the gap with whatever it can find: review profiles, comparison sites, old blog posts, reseller pages or forum threads. Those sources are often years old, quote a discontinued plan or mix up currencies. The model may also give a broad estimate drawn from similar products. Either way, the buyer leaves the conversation with a number you didn't choose.
Hidden pricing also costs you shortlist visibility. Prompts like "CRM under $30 per user" or "project management tool with a free plan for 10 people" can only include you if the assistant can confirm you meet the constraint. See how to fix AI hallucinations about your brand for the wider accuracy problem.
The pricing prompts buyers ask
Pricing prompts range from direct cost questions to constraint filters, and each needs a different fact on your page.
| Prompt type | Example | Fact the page must state |
|---|---|---|
| Direct cost | "How much does [Brand] cost?" | Plan prices and billing unit |
| Scale | "[Brand] price for 40 users" | Per-seat or tiered rules, volume breaks |
| Constraint | "Email tool under $50 a month" | Entry price in text |
| Free option | "Does [Brand] have a free plan?" | Free plan or trial terms |
| Comparison | "Is [Brand] cheaper than [Competitor]?" | Plan inclusions, not just prices |
| Hidden costs | "Does [Brand] charge setup fees?" | Fees, overages, add-ons |
| Value | "Is [Brand] worth it for a small team?" | Who each plan is for |
Add these to your prompt set so you can track accuracy, not just mentions. See building a prompt set.
What the page needs in text
A pricing page should state every fact a buyer or assistant would need to answer a cost question, in plain HTML text near the top. Work through this list:
- Plan names and the price of each, with currency
- Billing unit: per user, per location, per 1,000 contacts, flat
- Billing period and the annual discount, stated for both
- What each plan includes and its limits (seats, usage, storage, features)
- Who each plan is for, in one line
- Free plan or trial terms (length, card required or not)
- Add-ons, overages, setup or onboarding fees
- Discounts (nonprofit, education, startup) and eligibility
- Contract terms: monthly, annual, cancellation
- A visible "last updated" date
- An FAQ answering the hidden-cost questions directly
Open with a one-paragraph summary an assistant can lift whole: "[Brand] has three plans: Starter at $19 per user per month, Team at $39 and Business at $69, billed annually, with a 14-day free trial and no setup fees." For FAQ writing, see FAQ pages in the age of AI answers.
If you sell on quotes
Quote-based pricing can still be transparent: explain what the price depends on, give a starting point or typical range if you can, and describe the process. Enterprise, services and custom manufacturing companies often can't publish a fixed price, but "contact sales" alone hands the answer to third parties.
Publish:
- The pricing drivers: number of locations, volume, modules, contract length, support tier.
- A floor or range, if leadership allows it: "Most implementations start at $25,000 per year."
- What's always included and what costs extra.
- The process and timeline: "Send your requirements and get a quote within two business days."
- Worked examples of typical configurations, clearly labeled as examples.
Even without a number, a page that says "pricing is based on annual shipment volume and the number of warehouses" gives an assistant something accurate to repeat.
Technical traps that hide prices
Many pricing pages show correct prices to people but expose wrong or no prices in the HTML that crawlers read. Check for these:
- Monthly/annual toggles where only one set of prices is in the HTML, or where the default state shows a number without saying which period it is. Put both in text, or label clearly.
- Client-side rendering where prices load after JavaScript runs. See JavaScript rendering and AI crawlers.
- Prices in images or in design components that render text as graphics.
- Geo-based pricing that serves different currencies by IP with no stated default. Name the currency and link to regional pricing pages.
- Calculators only. Keep a calculator, but also state the formula and a few common results in text.
Test by fetching the page with a plain HTTP request and searching the source for your prices.
Markup that supports the page
Offer markup can make prices machine-readable, but it must match the visible page exactly. On a software or product page, a minimal example looks like this:
{
"@context": "https://schema.org",
"@type": "SoftwareApplication",
"name": "[Brand]",
"applicationCategory": "BusinessApplication",
"offers": [
{
"@type": "Offer",
"name": "Starter",
"price": "19.00",
"priceCurrency": "USD",
"description": "Per user per month, billed annually"
},
{
"@type": "Offer",
"name": "Team",
"price": "39.00",
"priceCurrency": "USD",
"description": "Per user per month, billed annually"
}
]
}
Update markup in the same release as the visible prices. Mismatches are worse than no markup. See schema markup for AEO.
Keep the rest of the web in sync
Your pricing page is only one of the sources assistants read, so update the others the day prices change. Keep a pricing source list and work through it:
- Review platform profiles (pricing sections are often edited by the vendor)
- App marketplaces and partner directories
- Reseller and affiliate pages
- Your own older blog posts, launch announcements and help articles that mention prices
- Comparison pages on your site
- Sales one-pagers that are publicly linked
Redirect or update old pricing URLs rather than leaving them live. See content freshness for AI search for update routines.
Example: auditing pricing answers
A 30-person software company (figures are illustrative) runs eight pricing prompts across four assistants, three times each.
- Finding 1: Two assistants quote a $15 starter plan retired 18 months ago. The source is a 2024 comparison article and the company's own old launch post. They update the post, ask the comparison site's editor for a correction, and add a clear plan summary to the pricing page.
- Finding 2: No assistant knows there's a nonprofit discount. It's only mentioned in a help article. They add it to the pricing page and FAQ.
- Finding 3: Annual and monthly prices are mixed up. The HTML only contained annual prices with no label. They add both, labeled.
Six weeks later, they rerun the same prompts and record how many answers are accurate.
The bottom line
Assistants will answer "how much does it cost?" with or without you. Publish prices or a clear explanation of how pricing works, put every fact in crawlable text with a date, avoid toggles and scripts that hide numbers, keep review sites and old content in sync, and check pricing prompts monthly for accuracy. AEO-Rank tracks pricing and other brand prompts across ChatGPT, Perplexity, Gemini, Claude and Grok so you can spot outdated answers early.
