How to Get Your Products Recommended by Perplexity and Google AI Mode

Two surfaces, two logics

Lump every AI assistant together and you will optimize for none of them. Perplexity and AI Mode share a foundation with ChatGPT, both rely on structured product data, but each adds its own filter on top.

Perplexity is citation-driven. It built its reputation on showing its sources, so every answer leans on documents it can point to: merchant feeds, review sites, comparison articles, community threads. When it recommends a moisturizer, that pick is backed by things written about that moisturizer. No paper trail, no pick.

AI Mode is feed-driven. Google's conversational search runs on the same Shopping Graph that powers Google Shopping, fed largely through Merchant Centre. When a shopper asks AI Mode for "a light jacket for Bangalore winters under 3,000," Google composes an answer from products whose feed data matches that intent. The feed is the ballot. If your data is thin or stale, you are not on it.

Both are downstream of the same shift we covered in the agentic commerce guide: people describing what they need in plain language and trusting an AI to shortlist for them.

How Perplexity decides, and what to do about it

Think of Perplexity as a researcher with a deadline. It fans your shopper's question into sub-queries, pulls candidate sources, and synthesizes an answer it can defend with citations.

That gives you two levers:

• Be a source. Get your product data into the structured feeds Perplexity reads, complete with the attributes that answer real questions: materials, dimensions, compatibility, who it suits, price and stock that are actually current.
• Be in the sources. Perplexity cites reviews, comparison posts, and community discussions. If your category's "best X for Y" articles and threads never mention you, the synthesizer has nothing to cite you from. This is the earning-mentions half of generative engine optimization.

A useful exercise: ask Perplexity the five questions your customers ask, then open every citation. That list is your placement map. Each citation is a place your brand either appears or should.

How AI Mode decides, and what to do about it

Google has spent years building the Shopping Graph, billions of product listings tied to price, stock, reviews, and seller data. AI Mode sits on top of it and turns a conversation into a query against that graph.

Three things matter most:

• Merchant Centre is your front door. Both Google and Microsoft Merchant Centre feeds decide whether your products exist in the AI's world. Treat the feed as a product, not an export.
• The new conversational fields are not optional for long. Google keeps adding attributes built for conversational shopping: Q&A content, document links, related products, popularity rank, variant options. Fields marked optional today have a habit of becoming the reason a competitor outranks you tomorrow.
• Context beats keywords. AI Mode reasons about use cases. "Light jacket" plus "Bangalore winter" plus "under 3,000" only resolves to your product if your data says what the jacket weighs, what temperatures it suits, and what it costs right now. Attribute context is the layer most feeds are missing entirely, as we showed in feed optimization for AI search.

Where brands actually lose

Run visibility checks across these surfaces for any category and the same pattern shows up. It is rarely one big failure. It is a missing attribute here, a stale price there, zero presence in the cited articles, and a competitor whose product card answers one more question than yours. Each small gap compounds, because the AI only has room to recommend three to five products.

The other quiet loss is the buy link. An AI can recommend your product and still send the shopper to a marketplace listing, where the margin and the customer relationship stop being yours. Tracking where buy links point is one of the three core metrics in measuring your AI share of voice.

The checklist

1. Audit your visibility. Ask both surfaces your top ten customer questions. Record which brands surface, at what position, and where each buy link goes.
2. Fix the feeds. Complete Merchant Centre data for every active SKU, including the conversational attributes. Aim for 15 to 40 real fields per product depending on category.
3. Add attribute context. For each attribute, capture why it matters: persona, occasion, climate, pain point. This is what lets an AI reason about your product instead of just sorting it.
4. Earn citations. Map what Perplexity cites in your category. Pursue presence in those reviews, comparisons, and communities.
5. Keep it current. Specs and required fields change weekly across platforms. Stale feeds sink silently. Automate the sync or assign clear ownership.

None of this is exotic. It is the same discipline SEO demanded in its first decade, applied to feeds and citations instead of links and keywords, and the brands moving first are compounding an advantage while their competitors still treat AI search as a curiosity. The wider playbook lives in AEO for ecommerce brands.