I recently had the chance to sit in on a digital Academy training with Google on the Shopping Product. Most of what they covered was generic and review for the sharp PPCer paying attention, but there were some interesting things worth calling out.
First, Google detailed the four quadrants of the Shopping Ads algorithm for determining what you will pay and how your ad will show in SERPs.
Those four quadrants are:
- Data quality (your feed attributes)
- Product relevance
- Predicted CTR
- Bid
As with Search, Predicted/Expected Click-Through-Rate is an interesting one to include here as it uses Google's machine learning to determine the likelihood your shopping ad will be clicked on by the audience in this specific auction and that assumptive decision helps determine what you actually pay in that auction. This means, things like price can also have an indirect impact on the auction since Google may determine someone is more likely to click on a lower priced product, and thus allow a lower bid to win for that advertiser with the lower price.
Nothing too surprising there, but it's still interesting to determine what Google actually details is included in the algorithm since my assumption is that likely means those four are weighted fairly heavily in the decision (albeit, with potential other factors).
Next, we see Google talk about a few feed attributes. There were some interesting datapoints to ponder in this section of the training for me.
- Title
- GTIN
- Product Type
- Description
- Images
- Variant Attributes
Here is why this is interesting, because again, whenever Google calls out something to pay attention to in the midst of a host of other attributes (there are many!), I pay attention. In this Academy video, the Googlers noted to ensure attention was spent on optimizing these six attributes... which makes a lot of sense since they are likely to impact the previous four algorithm elements. For instance, if you have a better image Google believes will be more likely to get clicked on, then it stands to reason that they'll likely assign you a lower bid based on the higher Predicted CTR. It's an indirect assumption, but it's worth optimizing things like your images because of that.
That being said, Google did go into some detail on these attributes. Nothing was terribly new, though it's worth calling out (with them) how essential the Variant Attributes are to include whenever you can. We've seen these be tested by Google in displaying as actual Filter options in Shopping Ads test types. We also know they help Google better associate the correct product variant with the search (thus, product relevance in the algorithm above). So if someone is searching for a red dress, make sure to have Red in the title, and as a Variant attribute.
Google also gave some tips on Title and Description that may be helpful:
Product Types was an interesting one to listen to, as this is the most descriptive I've ever heard Google get on Product Types, and it resembled things I've pieced together from various Googlers for years (and presented in conferences!). Especially of interest was the guidance to prefer Product Types as the organizational means within your campaigns so you can reserve custom labels for more... customized... groupings. More importantly, Google has revealed for awhile now, and reiterated, that their Product Types are weighted somewhat heavily in Google's system since it reveals how a store views the relationships between its products, and that can impact serving and product relevancy.
There were additional interesting notes about Shopping Video for Action campaigns and Local Inventory Ads (LIA) and new automated feed options (cool!), but I found this slide particularly interesting and wanted to share it with you as well as it gives some insight into impact of optimizing various key attributes:
Finally, Google closed the Academy video by noting a couple of key changes to be aware of (which we already knew, but I guess maybe if you don't monitor the change-log closely like me cause you actually have a life outside of Google Shopping, haha):
- Availability changes in September 2023 (have this ready before Black Friday!)
- LIA offers to require availability instead of quantity