by MichaelIt certainly seems as though AI Overviews are the crux upon which your ecommerce revenue sits. Or at least that’s what LinkedIn, conference panels and perhaps your current agency are leading you to believe. But are AI Overviews really the biggest opportunity to gain SERP real estate? I’d argue, in e-commerce, they’re outshone by a SERP feature from 2020. That feature is the product grid. In April 2020, Google made the Shopping tab primarily free listings. Merchants opt in, and the same product data becomes eligible across the Shopping tab, Google Search, Google Images and YouTube. Six years on, most large UK retailers still have not taken full advantage of it.
Nine searches in ten
Product grids fire on nine out of ten commercial searches in UK retail. To test this, we looked at five categories:
- Paint
- Pet food
- Footwear
- Home cinema
- Headsets
Every category we measured this July had a popular products appearance rate between 89% and 94% across the respective keyword universes. And yet, a lot of large retailers aren’t in them.
The thing is, it’s often not through lack of trying. A few months of optimisations, without much movement, is all it takes for somebody to say “Organic shopping just isn’t a real channel for us.”
Part of the reason it seems that way is that the product grid gets measured as one placement on one search results page. It is not. Because the free listings programme makes the same product data eligible on the Shopping tab, in Search, in Images and on YouTube, measuring only the box on the results page counts a fraction of the available placements, and it counts them on the surface where the click economics are worst.
Sure, the grid is a weak click channel
Get that out of the way first, because you’ll find it anyway and I’d rather you found it here. We’ve measured the average click-through rate for each SERP surface at its typical position. A top-three organic result gets around 55%. The product grid gets 4% to 7%.
Put another way, one appearance in the top three organically is worth roughly the same as eight to fourteen appearances in the product grid.
So as a place to collect clicks, the grid is poor. But here’s the thing. The popular product grid fires on nine searches in ten. It takes nearly two slots when it appears. And it sits above the first organic result on 34% of searches in paint, 33% in pet food, 22% in home cinema, 20% in men’s trainers.
On those searches, you’re competing for the top of what’s left of the page. You can hold position one, lose a third of your visibility across a third of your market, and see nothing at all in your data. And it’s accelerating. In the men’s trainers SERPs, over the last twelve months, grid prevalence went from 79.1% to 94.2%, and average position went from 8.53 to 5.95. So they’re appearing more often, higher up the page.
Of those two numbers, average position is the one to watch. Prevalence tells you how often the grid appears. Average position tells you how far down the page a user has to scroll before they reach your link. A grid that appears on 94% of searches at position 8.5 doesn’t change the picture that drastically. But the same grid at position 5.95 has a much bigger impact on organic performance.
Hygiene is not a strategy
Google’s product data specification tells us that hygiene means every product having id, title, description, link, image_link, availability and price, and for most products brand plus either gtin or mpn. GTINs have to validate against the GS1 guide with the check digit present and correct, in the right regional format, and every variant needs its own. The disapprovals that quietly remove products from a feed are nearly always the same handful, such as a wrong google_product_category, an invalid GTIN, missing variant attributes, or feed data that contradicts the product page. Get any of that wrong and you’re invisible. But getting it right does not make you competitive.
Google is more explicit about this than it gets credit for. Its own title guidance asks for the attributes that identify a product: size, material, intended age range, special features, technical specifications, and visual properties such as shape, pattern, texture and design. That is a description of how people search.
Then complete attributes, especially colour, size, material and age group. Variants that offer real choice instead of forty near-duplicates cannibalising each other. More than one image per product.
Reviews are worth setting out on their own, because the requirements are easy to miss. To show product ratings at all you need at least 50 reviews across your Product Ratings feed, and each individual product needs at least three reviews of its own before a rating appears against it. The feed has to be refreshed at least monthly. Onboarding can take a fortnight after you qualify. And if you use a third-party review platform, they submit on your behalf, which makes identifier alignment between their data and yours a real dependency.
Take a retailer with 12,000 reviews spread across 8,000 products. It clears the account-level requirement of 50 comfortably and still shows ratings on almost nothing, because the three-review minimum applies per product. Total review count tells you very little. Reviews per product decide whether a star rating appears next to you in the grid.
Stop fighting for the head terms
On short generic queries, the grid rewards price more than anything. In the paint keyword universe, Amazon holds a top grid slot on 87% of keywords, a price comparison engine on 65% and a trade merchant on 61%.
If you sell at full price, you will never win those, simply due to the intent of the buyer for those types of queries.
In the men’s trainers keyword universe, the grid outranks the first organic result on 20% of keywords, and those keywords account for 8.3% of search demand. The top ten keywords carry 60.2% of demand in that universe, so on the terms carrying the most demand the grid usually sits below the first organic result.
On generic queries you are least likely to win a grid slot and least likely to be affected by one. Specific queries are where the opportunity lies.
What we can’t tell you
Which signal Google weights most in grid selection varies by category and by market. We can see the correlations, and we can see what the competitors winning grid positions have in common. We can’t tell you in advance whether image count beats review volume in your category. A definitive ranking of signals that holds across all of retail is describing one category rather than the whole market.
If this sounds like a conversation you should have, we run a free 15-minute consultation for in-house marketers with a real problem to solve.
Stay human,
Michael
