This post is the first in a series that will use data to highlight some of the dubious, wasteful and lazy practices that we observe while monitoring, benchmarking and reporting on the work of major clients' advertising agencies or other operators of their Amazon Advertising (AMS) accounts. Merchant AI sees the techniques used, and results of several brand name, as well as less well-known ad agencies who operate clients' accounts in various market places around the world. Overall, these accounts can have aggregate spend measured in the 10s of millions of US$ per month, and are supposedly operated by some best-in-class agencies. As this series will show, there is still much room for improvement despite some of the remarkable claims bandied around this space. Recognizing that some companies are already well regarded for particular product categories in certain markets, many consumers will use branded search to get straight to a shortlist of results. For example a customer may search for 'lego', 'iphone' or 'S23 phone' because they already know what they want. In these cases, the customer is said to be already "predisposed" to buy a particular brand or type of product. This evident in the conversion data when compared to generic (i.e. unbranded search) or competitor-branded keywords. For a 6 month period in late 2022 and early 2023, across 3 international advertising marketplaces, based on classifying keywords used in SP & SB campaigns for several categories, the conversion rate and reported ROAS were:
CVR ACOS ROAS Own Product Exact 7.7% 1.5% 69.7
Own Branded 1.6% 4.5% 22.4
Generic 1.1% 16.6% 6.0
Competitor Branded 0.4% 16.2% 6.2
Competitor Product Exact 0.1% 19.3% 5.2
On the basis of the Amazon-reported ROAS, branded search looks like a winner!
Perhaps because of these results, the ratio of spend was Branded 48%
However, looking at the conversion rates, a customer that was already predisposed to a particular product was 7x more likely to convert than one that was still undecided on the brand.
To reinforce the point further, the odds of capturing a customer predisposed to a competitor product is ~10x harder than capturing an undecided customer, and ~77x harder than the one that is already predisposed to your product.
This begs the question "was the branded search incremental?" if it was really incremental while bearing in mind there are diminishing in returns to advertising, we should be probably be spending even more money on branded search. The key question is "would the customer would have bought your product anyway? " The only way to know is to turn off advertising and see what happens to sales rates with and without advertising. We did this too, but the answer is not for this forum. To form your own opinion, if you assume, say, 50% would have bought anyway, then 50% of the 48% Branded spend, i.e. 24% of ad spend was probably wasted.
Please note: All of the above is based on the premise that a company already has organic visibility in a category. This is is true for established players. For a new entrant with no visibility the game is different -- you may have to invest on- or off-Amazon to kick start your traffic & hence sales.
So why does this happen? It might be due to incentives -- Amazon has no interest in breaking out the numbers to inform the incrementality discussion. In addition, ad agencies are often paid based on a percentage of ad spend. If the client does not have visibility into the detail, an agency can make their numbers look good and improve their commission by burying branded search terms into their campaigns. The limiting factor that puts a cap on such behavior seems to be the CFO who caps at some palatable ACOS % because she smells a rat in the numbers.
If you are interested in such insights, please follow is on LinkedIn.
If you need some automated help to routinize Amazon analytics, including *really* understanding your ad spend and the effectiveness of techniques used, please contact us.