How Ecommerce Brands Actually Get Recommended by AI

From position one to ‘the AI picked us’

For fifteen years, ecommerce SEO had one job: get your product page onto page one. Beat the competitor on backlinks, nail the title tag, win the click. That game still exists, but it’s no longer the only game in town, and increasingly it’s not even the main one. When someone asks ChatGPT ‘what’s the best waterproof running jacket under £150’ or asks Google’s AI Overview to compare three air fryers, nobody’s scrolling through ten blue links to decide. The AI picks a shortlist and says them out loud. If your brand isn’t in that shortlist, it doesn’t matter that you’re ranked third on Google for ‘running jackets Manchester’ because the shopper never sees the list at all. This is the bit a lot of ecommerce owners in places like the Northern Quarter or Ancoats are only just waking up to: being recommended by an AI system is a different achievement to being ranked by a search engine. It draws on different sources, rewards different behaviour, and frankly punishes a lot of the tricks that used to work for classic SEO. Getting comfortable with that distinction is step one.

What AI answers are actually reading

So what is an AI system actually reading when it picks three products to recommend? Not your homepage copy, mostly. It’s pulling from a mix of sources: structured product data, price, stock, specs, in a format machines can parse cleanly, independent reviews on places like Trustpilot or Google Reviews, comparison and ‘best of’ articles written by other sites, Reddit threads where real people argue about which brand is actually worth it, and increasingly your own product pages if they’re clear and well structured rather than stuffed with marketing fluff. Notice what’s missing from that list: your ad spend, your Instagram follower count, and how clever your meta description is. AI models are trying to answer a question honestly, using whatever text they can find that looks trustworthy and specific. A product page that says ‘premium quality, unbeatable value’ tells the model nothing useful. A product page that says ‘320g, machine washable at 30°C, waterproof to 10,000mm, true to size’ gives it something to actually recommend. The brands winning AI visibility right now are the ones whose product data reads like a spec sheet, not a slogan.

Why your Google ranking doesn’t guarantee an AI mention

Here’s the uncomfortable bit for anyone still measuring success purely by Google rank. Since AI Overviews became a normal part of the search results page, organic click-through rates on queries where an AI answer appears have dropped sharply, some analysis has put the drop as high as 61% on affected queries. That means even a page one, position one ranking is worth a lot less traffic than it used to, because a chunk of shoppers now get their answer from the AI summary and never click through to any website at all. It’s not that SEO stopped working. It’s that the reward for winning at SEO shrank, while a new reward, being the brand the AI actually names, appeared alongside it. If you’re only optimising for the old scoreboard, you’re chasing a shrinking prize and ignoring the one that’s growing. The two aren’t mutually exclusive, thankfully. Good structured data and genuinely useful product pages help both. But the mindset shift matters: rank is no longer the finish line, being mentioned by name in the answer is.

What’s actually worth fixing this quarter

Practically, here’s where to spend your time if you’re running an ecommerce brand and want AI systems to start naming you. First, audit your product feed and pages for genuinely specific, factual detail, materials, dimensions, compatibility, care instructions, the boring stuff that AI models love because it’s checkable. Second, chase reviews actively across two or three platforms rather than hoping they trickle in, because AI systems seem to weight independent third-party reviews heavily. Third, look for the comparison and ‘best of’ content already ranking in your category and work out how to get your product included in it, whether that’s outreach, sending samples, or simply having the best publicly available spec sheet when a writer goes looking. Fourth, keep an eye on Reddit and forum threads in your niche. It sounds unglamorous but AI models scrape and weight this stuff surprisingly heavily, and one well-placed, honest mention can outperform a paid ad campaign in terms of AI visibility. None of this replaces classic SEO. It sits alongside it.