AI Recommendation Readiness: Does Your Page Give an Assistant Enough to Recommend You?

Suppose someone asks an assistant: "I run a three-person design studio. What's a good project management tool under 20 pounds a month per user?" To recommend a product, the assistant needs some facts. Who the tool is for. What it costs. What it does. What it doesn't do. If your page doesn't state these plainly, the assistant has to guess or pick another product that does.

Recommendation readiness is how well a page supplies those facts.

The facts a recommendation needs

Fact	Why it matters
What it is	A one-sentence description in plain words
Who it's for	Audience, company size, industry, skill level
What problem it solves	The situation that makes someone need it
Key features or inclusions	What you actually get
Price or price range	Needed for any budget-related question
Availability	Regions, platforms, languages, delivery time
Limitations	What it doesn't do or who it doesn't suit
Evidence	Reviews, case studies, awards, customer numbers, with dates
How to start	Trial, demo, booking or purchase route
Support and terms	Cancellation, guarantees, response times

Step 1: choose the pages

Start with the pages for your main products or services. If a page is mostly slogans and a contact button, it's probably the one most in need of work.

Step 2: run the checklist

Score each page against the table above: stated clearly, mentioned vaguely, or missing. Count the "stated clearly" items. A page with fewer than six out of ten is a weak base for a recommendation.

Step 3: test with real scenarios

Write three or four realistic requests, such as "best accounting tool for a one-person consultancy under 15 pounds a month", and give an AI assistant only your page text. Ask whether it would recommend the product and why. Look at whether it:

• Gets the audience right
• Knows the price
• Names a reason that matches your actual strengths
• Mentions limits correctly

Hesitation or invention shows you which facts are absent. The answer extraction test uses the same method.

Step 4: add the missing facts

Put them in visible text, near the top where possible. A short block under the main heading often works well:

• What it is: invoicing software for freelancers
• Best for: self-employed people in the UK with up to 50 clients
• Price: from 9 pounds a month, free trial for 14 days
• Not for: businesses needing multi-currency payroll

Then expand the detail below.

Step 5: add structured data where it fits

Use Product, Service or SoftwareApplication schema with matching values for name, description, price, availability and review data. The values must match the page. If the schema says one price and the text another, you've created a conflict.

Step 6: show evidence with dates

Customer numbers, ratings and testimonials carry more weight when they say who and when. Link to the original review platforms where the ratings come from. The guidance in fact verification applies here.

Step 7: keep it current

Outdated prices and discontinued features turn a recommendation into a bad one. Review these pages whenever something changes, and at least twice a year.

Common mistakes

• Replacing facts with adjectives such as "powerful" and "seamless".
• Hiding all pricing behind a form.
• Describing features without saying who benefits from them.
• Leaving out limits, then losing the trust of people who find out later.

A readable, honest, fact-rich page helps assistants and humans alike.

Further reading: schema.org's SoftwareApplication type and Google's product structured data guide are good places to check the details straight from the source.