By Steve Merrill, Founder of WRKNG Digital | September 10, 2026
AI assistants don't read your whole site. They read a few signals and decide. Seven of them carry most of the weight. Get these right and you show up in answers. Miss them and you don't.
I've watched search decide winners for fifteen years. The judges changed. The idea didn't. Answer the question clearly and you win. Here are the seven signals that matter.
What on-page signals do AI assistants weigh?
They weigh signals that prove you answer the question and can be trusted. Direct answers, structured data, matching specs, clear terms. Here are the seven, in the order they tend to matter.
1. A direct answer in the first sentence
The top sentence of each section should resolve the question. Models lift the top line. Bury the answer and it never gets read.
Lead with yes, no, or the number. Detail comes second.
2. Question-style headings
Headings written as the questions people ask help a model map your page to a query. "Does this run small" beats "Sizing."
The closer your heading matches the buyer's words, the easier you are to cite.
3. Structured data with real values
Product and FAQ schema, filled with live values, tell a model what your page is and what it answers. Validate with the Schema.org validator.
Empty schema is the same as no schema to an assistant.
4. Specs that match your feed
The specs on the page have to agree with your product feed, aligned to the Google Merchant product data specification.
When the page and feed disagree, a model discounts the whole product rather than pick a side.
5. Structured reviews
Ratings and counts marked up as data lower the risk of recommending you. No structured reviews reads as unknown.
Real ratings tied to the product beat a vague testimonial wall every time.
6. Clear shipping and return terms
Explicit shipping cost, delivery window, and return window reduce the friction to recommend. An agent completing a purchase needs them stated, not buried.
Vague terms read as risk and push you down the list.
7. Consistency across the page
Every claim should agree with itself across copy, schema, and feed. Consistency is what earns a model's trust.
This is the read a WRKNG audit gives you, signal by signal, so you can see which ones you're winning and which you're losing.
Further reading
- Schema.org validator
- Google Merchant product data specification
- Google product structured data guidelines
Frequently Asked Questions
What makes AI recommend one store over another?
The store that proves it answers the question and can be trusted. Direct answers in the first sentence, question-style headings, validated structured data, specs that match the feed, structured reviews, and clear terms all push a store up the list.
Do question-style headings help with AI search?
Yes. Headings written as the questions people actually ask help a model map your page to a query. Does this run small beats a heading like Sizing, because it matches the buyer's words and makes your answer easier to cite.
Why does page-to-feed consistency matter?
Because when the specs on the page disagree with the feed, a model can't tell which number to believe, so it discounts the whole product. Consistency across copy, schema, and feed is what earns an assistant's trust and a recommendation.
How many signals do I need to get right?
As many as you can, but start with direct answers, validated schema, and feed consistency. Those three carry the most weight. A WRKNG audit shows which signals you're winning and which are costing you recommendations.
Want to know if AI assistants can actually find and recommend your store? Get a free AI-visibility read on your Shopify store at WRKNG Digital. We show you exactly what ChatGPT, Perplexity, and Google AI see when they look at your products.
