You are unable to choose keywords to target your ads - you can only controll negatives.
Only Title and Description from the product feed determine the context your ads can be shown.
You are unable to distinguish max CPC on keyword level - all keywords that trigger your product ad have equal bids.
To everyone who ran any search engine campaign, it is obvious that different keywords have different efficiencies, thus from the advertiser perspective - they have different value.
For sure, a general word such as "Fridge" will never reach such conversion rate as a word referring to a specific product (containing a specific catalog number), such as "Fridge Bosch KGN36ML3P".
In this scenario (search query with a catalog number), very often we are dealing with a user who is already at the end of the conversion path. He has already chosen a specific product, and the only step he makes is looking for a store where he can make a purchase or compare prices.
| Keywords | CR | Margin | OV | max CPC |
|---|---|---|---|---|
| Fridge | 0,1% | 5% | 1500 EUR | 0,068 EUR |
| Fridge Bosch KGN36ML3P | 2,4% | 5% | 1500 EUR | 1,64 EUR |
Determining max CPC depending on the efficiency of a given keyword
Analysing the conversion data above, to achieve the ROI of 10%, the optimal solution is to set low CPC = 0.068 EUR for the word "Fridge", and much higher = 1.64 EUR for the word "Fridge Bosch KGN36ML3P". Unfortunately, it is not possible in standard PLA campaigns.
Thanks to appropriate PLA campaign structure we are able to move keywords reffering to each product from your feed into one of three campaing types: one campaign contains only general search terms such as "Fridge", second campaign contains keywords with Brand names, e.g. "Fridge Bosch" and in the third campaign you will find very specific keywords, for example, "Bosch fridge KGN36ML3P".
This allows us to assign the individual bids to given keywords, for example:
| Keywords | CR | Margin | OV | max CPC |
|---|---|---|---|---|
| Fridge | 0,10% | 5% | 1500 EUR | 0,068 EUR |
| Fridge Bosch | 0,50% | 5% | 1500 EUR | 0,34 EUR |
| Fridge Bosch KGN36ML3P | 2,40% | 5% | 1500 EUR | 1,64 EUR |
Finally, when someone searches for the "Fridge Bosch KGN36ML3P", because of the very high bid, it is almost certain that the user will see your product in the top five ads displayed.
Unfortunatelly, creation of so-well organized campaign structure requires a lot of manual, repetitive work associated with adding appropriate negatives to each type of campaings to ensure the accurate choice of the ad to a given search query. If you have thousands of products, there are also thousands of negatives to be added and what is more - according to latest research, 15% of every day queries typed in search engines is completely new.
How to qualify them for the relevant Group?
To automate the camapaign management process, we developed a dedicated tool that uses Machine Learning to determine the correct assigment of keywords. The model uses your data to learn, to qualify keywords even more precisely, even though it already achieves amazing results, which is confirmed by the parameters below.
93%
Accuracy*
95,8%
Precision**
74,8%
Recall***
* Accuracy is the fraction of predictions our model got right
** What proportion of positive identifications was actually correct?
*** What proportion of actual positives was identified correctly?
Machine Learning helps to categorize new queries - those that have not appeared before and are not used in the campaign. In addition, our tool does it very quickly, all new queries are being analyzed within 24 hours.
In 2017 we implemented 3PLA to one of our clients from the household appliances industry. What is important, PLA campaigns on this account have been optimized over the years and have reached their maximum efficiency in the basic structure. The implementation of the new 3PLA approach, after just one month, improved the results as follows:
Change (November 2017 to October 2017):
3PLA approach can give you completely new opportunities to optimize your Shopping Campaigns!
How does it work, step by step?
1). Identify the product categories for which you would like to implement 3PLA.
2). Divide campaings according to 3PLA criteria.
3). Set up the 3PLA structure and add basic negatives.
4). Provide an API access to the adwords account for ML application to automate your work.
5). Evaluate your campaign effectiveness.
