AI Shopping Is Here: Clean Up Your Product Data Before the Robots Skip You

Your product photos are gorgeous. Golden hour. Perfect styling. Very moody.

The AI shopping assistant doesn't care. It can't tell from a vibe whether the couch fits through your door.

Neil Patel broke down AI shopping on Oct 15, 2026, and the takeaway is simple: the AI builds the shortlist. Messy product facts mean you're not on it.

Quick words (one line each):

• AI shopping agent: an AI assistant that researches, compares or buys products for a person.

• Product feed: a file listing every product with its price, stock and details, made for machines to read.

• Variant: a version of one product, like a size or color.

• Referral: a visit that came from a link on another site or app.

• Shortlist: the few products the AI shows the shopper.

Why this matters right now

Numbers Neil Patel showed (Oct 15, 2026):

1. Accenture (25,000+ shoppers, 16 countries): 74% would let AI help research and compare. 32% would let it choose within limits if they approve payment. Only 9% would let it buy on its own.

2. His team's data on 412 companies: AI drives an estimated 6.2% of traffic but about 11.4% of sales. An AI visitor is nearly twice as likely to buy.

3. Accenture again: 56% will tell the agent which brands to look at.

So people still pull the trigger. The AI picks what they see first.

Step 1: Run the no-pictures test

This is Neil's test, and it's brutal.

1. Open your best product (or service) page.

2. Ignore every photo and every design detail.

3. Read only the text.

4. Can you tell the size, price, ship time and return policy?

If you can't, neither can the agent. Neil's example: AI narrows down a couch by dimensions, delivery and reviews, then a person picks. Written dimensions were a top fit check in a Baymard survey he cited.

Step 2: Write the facts as plain text

Neil's Job 1 is clean product data. Every product page gets:

1. Clear specs (size, material, dimensions).

2. Every variant listed.

3. The current price.

4. Availability, and it must be true.

5. Shipping cost and time.

6. The return policy.

Put them in text on the page. Not inside an image. Not hidden in a tab only a human opens.

Step 3: Never say "in stock" when it isn't

Neil's warning: if your feed says in stock when you're not, the agent learns not to trust you.

He also pointed to Walmart: its in-chat checkout converted far worse than its own site, partly because of bad product data like wrong stock and wrong shipping costs. And OpenAI scaled back in-chat checkout in March to focus on discovery.

So when something sells out, mark it sold or unavailable right away. Stale stock teaches the agent to skip you.

Step 4: Publish a product feed

Neil says publish a product feed with the same facts as your pages. Same price, same stock, same shipping. If the page and the feed disagree, you look like a liar to a machine.

Step 5: Get reviewed where shoppers check

OpenAI says ChatGPT shopping results use brand and third-party product data, including public review summaries, and you can't buy your way in (as Neil Patel reported).

Neil's Job 2:

1. Pick one review site your customers check before buying.

2. Ask your last 20 customers to review you there.

3. Real customers only.

Neil's survey of 500 marketers ranked brand mentions (94%) and reviews (91%) far above backlinks (3%). He says himself that's what marketers told them, not a measurement of the algorithm. The direction still points at reviews.

Step 6: Check if AI shoppers already visit

Neil's list of AI referral sources to look for in your analytics:

1. chatgpt

2. perplexity

3. gemini.google

4. copilot.microsoft

Barely there? He says that traffic is probably being counted as "direct."

Step 7: Not a store? Same rules

Selling services? Your "product" is the service page. Run the no-pictures test on it. Can a machine read the price range, the area you serve, how fast you start and what's included? If not, fix it.

Do this today

1. Read your best product page with the photos ignored.

2. Add size, price, ship time and returns as plain text.

3. Mark anything sold out as unavailable today.

4. Pick one review site and ask your last 20 customers.

5. Look for chatgpt and perplexity in your referrals.

Related reading

• How to measure AI search traffic

• Get reviews where AI looks

• Make your pricing page the one AI quotes

Sources

• Neil Patel - AI Shopping Is Here, and Amazon Isn't Too Happy About It (2026-10-15)

Don't worry if this seems like it's a pain in the ass for someone who doesn't build websites. That's because it is. The good news: just click the link below, pull the trigger, and we'll take care of the rest.