When Amazon Marketing Cloud (AMC) was first launched in 2021, it was described as ‘a new, holistic measurement and analytics solution’. If was also called ‘a secure, privacy-safe and dedicated cloud-based environment in which advertisers can easily perform analytics across multiple, pseudonymised signals to generate aggregated reports’. Understandably, advertisers weren’t really sure what to make of it.
So what is AMC really?
AMC is a huge lake of data that relates to everyone who has seen one of your ads in the last 13 months. For each one of those people, Amazon holds metrics such as:
- which ads they have been exposed to and when,
- what categories and ASINs they’ve looked at,
- what they’ve added to basket,
- the things they’ve bought,
- how far they watched through video ads,
- what searches they’ve made and
- which ads they clicked on.
Depending on your point of view, this is either terrifying or transformational. So first, some reassurance.
The ‘pseudonymised signals’ that were mentioned in the first paragraph, refers to the fact that none of this rich data can be tied back to an individual shopper by users of the system. In AMC they aren’t people, but pseudonymised data. In addition, you aren’t able to analyse any data sets that contain data from less than 2,500 people. So even if you’re looking at really granular analysis (say, the post-purchase behaviour of people who have seen exactly 4 of your ads in the last 6 months and added your slowest seller to their Subscribe and Save subscription), you will only be able to do this if there are 2,500 such people, Consequently you won’t be able to see enough detail to identify anyone. The full Amazon UK privacy notice is here.
why is Amazon MArketing Cloud transformational?
You can conduct incredibly granular analysis to understand how your ads are working. But then you can turn this analysis into action. You can build ‘buckets’ of shoppers based on the analysis which you can target with your ads.
Example – Optimal Frequency
Using Amazon Marketing Cloud we can analyse the optimal number of times that a shopper is served ads for a given brand before they make a purchase. This is dummy data, but this is the type of graph that you can generate from this analysis:
Just by looking, you can see the purchase rate accelerates rapidly from 1-7 exposures, with a peak at 11 exposures. With more exposures, the purchase rate doesn’t increase significantly. We could conclude that the ideal number of exposures is 7-11 for the best chance of a conversion.
Using Amazon Marketing Cloud Audiences in Advertising
Amazon has two different advertising platforms; Sponsored Ads and DSP. The Sponsored Ads platform targets shoppers based on their current shopping activity. For example you can use the products that they are currently looking at, or the search terms that they type to target the ads. DSP is different in that you are directly targeting shoppers based on their demographics, previous purchasing behaviour, lifestyle factors etc.
Within DSP advertising, you can add your AMC audience to your targeting. Once this is done, your ads will serve to the audience you have chosen.
Within Sponsored Brand and Sponsored Product ads, you can use Amazon Marketing Cloud audiences to increase the likelihood that your will serve to the AMC audience you want to upweight. This allows you to place higher bids if the result is an ad exposure for someone in this specified group. For Sponsored Display ads, you can target your chosen audience directly, just like DSP.
Returning to the Example
We decided that the ideal number of ad exposures for a conversion was 7-11. We can now take this insight and turn it into action.
The first choice would be to build an audience in AMC, comprised of everyone who has had at least one ad exposure, has not yet purchased, and has had fewer than 7 ad exposures. As purchase is the goal, when someone has made a purchase, they are no longer relevant. The fact that these people have had one ad exposure already means that they have already been selected by your existing targeting as being relevant. When we add this new AMC audience to the targeting for advertising, we are maximising the chance that they receive further ad exposures and are more likely to convert.
In addition, we can build an audience of people who have had 11 or more ad exposures and haven’t purchased. At this level it is less likely that further exposures will generate a purchase. It would be better to spend the advertising investment on shoppers who are more likely to convert. So we could use this audience as a negative target in the DSP. This would ensure these people receive no more DSP ads for, say, 6 months, when their needs might have changed.
Amazon Marketing Cloud audiences update themselves
One of the huge advantages of building audiences in AMC is that it’s possible to build them so that they auto-update. If someone in our new Amazon Marketing Cloud Audience makes a purchase and we have constructed the audience correctly, then they will drop out of this Optimal Frequency audience ‘bucket’. Once they’ve had 7 ad exposures, then they will also drop out, as they’ve reached the optimum zone for conversion.
In addition, shoppers who reach 11 ad exposures are added to the negative targeting group. This will actively prevent further DSP ad exposures.
This means that once the criteria are set, then the audiences can be left to run themselves.
Feeding back into analysis
As a follow up to the example above, we could then evaluate the impact of the audience changes that we made. Going further, we could look to see if there’s any variation between the number of exposures required for a purchase for different creatives. Or a particular combination of different ad exposures that gives the best purchase rate for the least exposures. This analysis could feed additional audience creation and deployment, and so on.
Summary
Amazon Marketing Cloud enables deep and granular analysis of advertising interactions and results. The analysis itself is powerful, but insights are also actionable. This is because you can create audiences based on the results. You can create these audiences in such a way that they self-update based on the criteria selected. They provide direct targeting or upweighting for these audiences in your advertising.