If your products aren't converting the way they should, the problem might not be your budget, your bidding strategy or your creative. It might be something far more unglamorous for most (although not for us!): your product feed.
We know. Feeds aren't exactly the sexiest topic in performance marketing. But right now, they matter more than ever, and most brands are significantly underinvesting in them.
Google's Shopping and Performance Max campaigns have always relied on feed data to match your products to the right searches - that's not new. But the shift toward AI-powered discovery, and the emerging world of agentic commerce, has raised the stakes considerably.
Information recently shared by Google confirmed what we'd been seeing in practice: feed quality is increasingly the difference between a product that surfaces and one that doesn't.
LLMs and AI-powered shopping tools don't just pull keywords. They read your product data holistically, and if that data is sparse, inconsistent, or outdated, your products lose out. Garbage in, garbage out - as the saying goes.
You don't need a complete overhaul. But if your feed has gaps in any of these areas, you'll want to prioritise them.
1. Rich titles and descriptions
Short titles and thin descriptions are still the most common feed problem we see. Google's guidance is clear: aim for titles over 30 characters and descriptions over 500. Include GTINs (Global Trade Item Numbers) wherever possible. These signals help Google match your products to the right searches, and they help LLMs understand what you're selling well enough to surface you in the right context.
2. Multiple images, including lifestyle
A single studio shot isn't enough any more. To surface across all placements (think Shopping, Discovery, Demand Gen, and AI-powered product recommendations), you need additional images, and lifestyle images in particular. Show the product in use. Show it in context. That visual variety unlocks placements your single-image competitors aren't appearing in.
3. Return policy and handling time
Your return policy and minimum handling time aren't just logistics, they're data that LLMs actively use to answer customer questions. "When will this arrive?" and "Can I return it?" are some of the most common purchase-intent queries going. If your feed doesn't have this information, an AI assistant cannot give your customer a straight answer. That's a lost sale.
4. The new product highlights attribute
This is the most exciting one. Google has introduced a product highlights attribute that lets you add bullet points for each SKU, short, punchy descriptions of what the product is and why it's worth buying. Think of it as writing for the LLM: clear, factual, succinct. This is how you tell an AI assistant what your product actually does, in plain language it can relay to a customer mid-query.
We use Channable as our preferred feed management platform, and we think it's brilliant for this kind of work. It gives you the flexibility to enrich titles, manage rules at scale, and push changes across channels quickly. That said, we work with multiple feed platforms and can work with whatever you're already using.
The platform matters less than the discipline. Feeds need to be treated as a living asset, not a one-time setup job.
If you haven't audited your feed recently, start with the four points above and benchmark where you are. Chances are you'll find quick wins in descriptions and images before you even get to the newer attributes.
If your brand is already running PPC with us and you're not sure whether your feed is holding you back, ask. It's one of the first things we look at.
Want to talk through your feed setup and what it could be doing for your performance? Get in touch with the team.
