Many users tend to analyze keywords/search terms aggregated over time and at the campaign level. This type of analysis doesn't suit our algorithm bidding model :
- bids change every day
- the algorithm reacts pretty fast
- ASINs are managed at ad group level
How should you look at search terms performance
Here is our recommended way of analyzing our search term bidding model:
- search terms analysis should be done at ad group level
- if you want to check the investment on a search term, it should be an analysis of its variation over time
- you can check the bid modification in the History section of the Amazon Advertising Console
Bid decreases if one of these scenarios happen:
- spike of spend but poor conversion for the search term on a specific ad group
- last 14 days ACOS of the strategy is higher than the ACOS target
Why do we keep these bad search terms?
Our research shows that when you compare two periods of time and measure the amount spent on "bad search terms", the result is always the same. Spending on "bad search terms" is insignificant. Furthermore, the performance over time is almost the same, whether you keep the "bad search terms" or remove them from the active campaigns. So, you don't have to worry about seeing search terms that you consider "bad" in the m19 campaigns. The AI makes sure it is not impacting your performance but will be able to ramp these search terms up if they start driving sales.
How does it work? Basically, the algorithm will decrease the bid of those "bad keywords" to a point where the bid is not high enough to compete with other ASINs but stay high enough to catch any conversion opportunity.