5 steps to better use Google Smart Shopping
Frequently seeing criticism of Google Smart Shopping is that it targets users who would have already bought from you. We also call
Frequently seeing criticism of Google Smart Shopping is that it targets users who would have already bought from you. We also call this the selection effect. You get more out of your campaigns if you can avoid this. In this article I explain what the biggest problem is with Google Smart Shopping and how you can prevent it by following 5 steps.
In 2019 Google introduced Smart Shopping: the automated variant of Google Shopping. Unfortunately, this 'smart' automation means you can control less on campaigns, which means that many conversions are unjustly combined.
I will first briefly discuss the advantages and disadvantages of Google Smart Shopping Campaigns. Google Smart Shopping Campaigns is a solution from Google that combines standard Shopping and display remarketing campaigns. It uses automatic bidding and ad placements to promote your products across all Google networks.
Smart Shopping, as you may have noticed, has many advantages. It is easy to set up and manage, you have a wide reach and it uses all the signals that Google uses for smart bidding. In addition, it offers during the auction (also called auction-time bidding) with a powerful, self-learning algorithm.
Algorithms do not distinguish
One of the biggest criticisms of Smart Shopping campaigns is the tendency of the algorithms (aka: a set of instructions for a specific purpose) to focus on your existing customers or on customers who have previously interacted with your site or brand.
After all, the algorithms optimize for the chance of a conversion. They cannot distinguish between the selection effect (people click on your ad, but would buy from you anyway) and the ad effect (people click on your ad and therefore buy from you).
Why is Google not solving this "problem"?
Google has made Smart Shopping non-transparent by removing search term data and data about the share of remarketing from the reports. Google itself indicates that it wants to make it simpler, but it is also likely that Google would prefer to have you steered by selection effect. After all, the more sales value they can attribute to the ads, the more you spend.
As a result, you cannot see the percentage of clicks that have arrived via your own brand name or, for example, via a remarketing banner on YouTube or the Google Display Network. In other words: the users at the end of your conversion funnel who might otherwise have bought from you anyway.
This is how you use Google Smart Shopping better: a 5-step plan. If you can control more on your Smart Shopping Campaigns (that which Google has disabled), you prevent Google from sending too much on these users who would otherwise have already bought from you. Below I explain how to do this in 5 steps.
- Collect the correct data
Always start by collecting the right first-party data: that is data from your own website visitors and buyers in your system (such as Woocommerce, Magneto, Wix, or Shopify). Save this data.
Given the amount of data, it is useful to use a system such as Google Big Query for this. You then need this data for analyzes and to ultimately be able to upload the data to Google Ads.
The GCLID is also one of the data components that you will be collecting. By using this parameter in your URLs you can recognize from which ad someone came to you and what type of visitor it is (User ID). Other data may also be important, such as when the last sale took place.
By keeping track of the GCLID and linking it to a specific order, you can later assign the correct conversion value to a particular ad click.
The GCLID is easiest to set up via the Google Tag Manager (GTM) and a hidden field in the order form.
- Divide your database into target groups
Divide the website visitors and buyers in your database into groups. You do this to correct the selection effect of the Google algorithm. Different target groups have a different effect on the selection effect (I will explain this further in step 3).
You can organize and characterize groups in the following ways:
- New visitors
- Returning visitors
- New customers
- Loyal customers
The above is therefore only an example. Which characteristics you assign to the groups will be different for everyone.
- New visitor = never visited your site before
- Returning visitor = visited your site before but hasn't ordered anything yet
- New customer = ordered for the first time, loyal customer = ordered more than once
You could also split the last group further if that is interesting. For example, use the RFM model (recency, frequency, monetary value). However, that is the next step.
- Determine the conversion value per group and per touchpoint
The next step is a bit more complex and requires the expertise of a specialist (with a data science background). Unless you want to do this on the basis of assumptions or possibly benchmark data.
Determine a suitable multi-channel attribution model. A multi-channel attribution model is a visually displayed model that assigns value to different touchpoints within a customer journey. A touchpoint is a piece of information about the interaction of the ad. It tells something about the who, what and where of the interaction.
For example, someone may have seen your YouTube ad first, then end up on your website via a click on a Google text ad and then buy a product from you the next day, after seeing a Facebook remarketing ad. The multi-channel attribution model determines how much value each of the different touchpoints above is assigned.
What Is Incremental Uplift and Why Does It Matter?
Supplement the multi-channel attribution model with results from an incremental uplift test. This way you can determine the correct incremental value per click, depending on the conversion path and target group.
With incremental uplift, you measure whether an event, such as a conversion, would not have happened without a specific interaction, such as an ad serving. This will answer the question of whether your advertisements actually had an impact on sales or whether they might be claiming credit for an action that would otherwise have taken place.
Because you perform this step, a conversion can become worth more or less for you. For example, a conversion from a loyal customer can become worth less because there is a good chance that he or she had already reached you via a direct channel.
- Create a new conversion action in Google Ads
Create a new conversion action in Google Ads so that you can import all conversions that you have saved (including those from the past) using the GCLID.
- Include the new conversion action in your conversions
Is the data in order? Then include the new conversion action in your conversions. To do this, go to the tools tab and then via the measurements to conversions heading. Click on the conversion action and click on adjust setting. Change include in conversions from œno to œyes and save the conversion action. Your Smart Shopping campaign then runs on the new conversion and focuses on better factors.
Using the key to smart shopping better: putting the right value on conversions
Because you attach the correct value to conversions through the above 5 steps, you avoid the selection effect.
Note: you actually only advertise at its best if you combine different disciplines in your "Ads recipe". A lot of knowledge about your company, a Google (Ads) specialist, a data scientist and good software or tooling.