75% don’t buy what they’ve searched for on Google Shopping!

A Google user searches for a particular product, sees the corresponding product listing ad, clicks on it, and makes a purchase. Common sense would have us believe that this is normally the case. But, is this actually true? The following analysis shows that only in rare cases do users buy exactly what they were searching for. This insight calls many advertisers’ strategies into question.


Only about 25% buy what they’ve searched for

To answer this question, I’ve taken data from a large, international fashion retailer and used the product “Nike Air Max Thea” as an example.


Over the past few months, Shopping Ads for this product generated a total of 1,300 sales in Great Britain. However, only in 336 cases (which represent 25.8% of all sales) did the users actually purchase the Nike Air Max Thea shoe model in various colors. So, here, only one in four customers actually purchased what they originally sought out.

Only 6.7% purchased other “shoe” products

So far, so good, but what other products make up the remaining 74.2% of sales? Certainly other shoes, right? Far from it. To our continued surprise, only 6.7% of sales (86) were shoes in the broadest sense.

Performance of Shoes
Performance of shoes

Here, I would have expected far greater shoe sales, as the search term implies that this is the user’s intention. The landing pages could possibly provide an explanation. Target pages for Shopping Ads naturally have a strong focus on a single product. If related products from a category or brand page aren’t easily reached through a gallery or site navigation, or if the related products aren’t relevant enough, the customer may return to the shop’s homepage and search within the site. Related product visibility may influence this change in the customer’s focus.

T-Shirts, Long-Sleeved Shirts, and Accessories are strongly represented

Among the most frequently purchased product groups are t-shirts and long-sleeved shirts, which make up a total of 17.7% of sales. These are followed closely by accessories (sunglasses, belts, hats, rings) with 16.9%. The remaining purchased products are divided up relatively evenly among outerwear tops (jackets, coats), bottoms (pants, jeans, leggings, dresses, and skirts), and other products which aren’t categorized as easily (bikinis, socks, cosmetics).

Relation Categories
One search query leads to different purchases.

The bottom line

With my example, I’ve shown that on average, only one in four consumers buys exactly the product they search for. The other customers choose different products instead. So, what does this mean for advertisers’ strategies?

  1. Even low-inventory products should be advertised, since they can lead to additional sales.
  2. Reports in Google Shopping distort the reality of actual sales. A real top-brand analysis is only possible by looking at actual purchased products, not product ad clicks.
  3. It’s short-sighted to orient bids toward the margins of individual products, as, for example, our colleagues at Volusion suggest. Doing so assumes an exact match between ads and purchased products.
  4. Google Shopping is of limited use when it comes to steering sales of specific products.

What are your experiences regarding the relationship between search terms and purchased products? Do you agree with my analysis, or do you have other opinions? I look forward to your comments!

[shareaholic app="share_buttons" id="19406647" link=""]


Huey-Yng Cherng-Griga

Huey-Yng is a PPC Analyst for crealytics. Before, she worked as Account Manager and took care of some of the biggest online retailers in Europe.

  • James Pomeroy

    This is really interesting. I think it emphasises the importance of showing alternate product suggestion on the product page to encourage the user to continue browsing.

    • Mike

      True. A pity that google doesn’t provide search referrers any more. Otherwise you could just suggest products related to the search which triggered a click.
      Still, a more prominent link to the brand or category page would make sense.

      • Huey-Yng

        Thanks James! @Mike I totally agree with you. I’ll keep on digging in numbers to get more insights.

  • Interesting article. I think you make some awesome points, especially in getting us to think more deeply about the buyer psychology in Google Shopping. That can only help us as we all grow in creating better Shopping Campaigns!

    I would however push back on your conclusion (except for the 1st bullet point. Agree there!). Perhaps I misunderstood your article, but it appears that the main conclusion you arrived at was “we can’t really know what product users want since they don’t buy exactly what they clicked on so when it comes to product/margin specific bidding/organization in Google Shopping… well, we just can’t know.”

    If my summary of your conclusion is accurate, than my push-back would be “Then what is the better alternative for bidding on and grouping products?” The grouping of products by margin as Volusion does, or the purposeful bidding of specific products (as I do) is usually based upon the conversion data that comes in. In other words, while you are correct that we can’t guarantee a specific product will be purchased, we still need some way to segment products to identify the better performing groupings. To say it’s “short-sighted” to bid on margins as Volusion mentioned makes sense in an ideal sense, but if that means organizing the product data results in more successful segmentation and more conversions I’m not seeing how it is short-sighted.

    In other words, I could care less if 200 people click on my Red Shirt PLA but then 20 of them buy a Blue Shirt instead. They still bought something, thus verifying the profitability of that product and that is primarily what I care about when determining what to invest Shopping energy/budget into. I will perhaps, use that information to try to tweak the feed or bidding on the Blue Shirt with the hopes that I can entice more in Google Shopping at that point (though as you said, doing that might not help all that much anyway if they don’t know what they want). But at the end of the day, if that many people consistently purchase through that Red Shirt PLA, I will continue bidding specifically on that because it has demonstrated profitability, regardless of the specific outcome of their transactions.

    I think the article is super, and it has me thinking a lot more on this. I think you are absolutely correct that we need to continue thinking of better ways to group our Shopping campaigns so we can increase exposure of profitable products & limit traffic for historically proven “non-profitable” products. I really do appreciate you writing this post, but I’m not sure if the data supports (all of) your conclusion when it comes to practically making decisions in the account.

    • Hi Kirk, good point. If people don’t buy what they search for and click on, then what is the alternative?

      I think you described it quite well: “if 200 people click on my Red Shirt PLA but then 20 of them buy a Blue Shirt instead. They still bought something, thus verifying the profitability of that product and that is primarily what I care about when determining what to invest Shopping energy/budget into.”

      Given that people don’t buy what they are looking for, the margin of the products is a weak proxy for bids. To overcome the problem of scarce historical data when looking at product item ID level, you can e.g. rely on attributes of products to build clusters of similar products. If you don’t find enough data for “Women’s Nike Air Max Thea White” you can factor in data from e.g. all Women’s Air Max Tea, then all Air Max, and finally all Nike products. You can use a weighting which takes into account the volume on the different levels.

      On the other hand, the margin of the products which are actually sold do matter (as you described). By tracking the exact margin of every product you’re selling and calculating your bids accordingly, you would automatically raise bids for products which have a small margin themselves, but in the end generate sales in areas of high margin.

      The same for managing inventory:
      If you want to push product A which has high stock levels, you have to increase bids for the products which lead to sales of product A. As there will be dozens of products which generate sales of product A, all these products will get a slight bid increase then. Still, managing inventory in a almost 1:1:1 relationship “search, click, buy” is not possible or at least not as easy as some might believe.