In theory automated bidding systems are a great tool. You set your target metric, assign it to your campaigns and sit back and watch it manage your keywords to your set target. I mean what could go wrong? It is supposed to save you more time in day to make your accounts better by not having to worry about bid management. It sounds perfect doesn’t it? I’m telling you to be careful about using an automated bidding tool, and here is why:
1. Make sure that you have an account big enough
Automated bidding tools are made for huge accounts that have thousands of keywords and millions in ad spend. Their algorithms are designed to handle these types of accounts. If you have a relatively small account there is a good chance that the algorithm is not going to give you the lift and efficiency you are looking for.
2. Be aware of lag time and seasonality
If your business has any lag between taking an initial action on your site and the completion of the transaction, you have to take this in account with automated bidding systems. While these systems claim to take this into account, there is not an effective way to program an algorithm to understand things like, “seasonality,” or “multiple day conversion process.”
Most business has a high point and a low point and as an account manager you should know exactly when this time comes into play and you should account for it. In the case of an automated system they are going after your target metric and using historical data to factor the bid. If you historically have a strong holiday season, but business falls off tremendously in January, “man” version of you would start to decrease bids knowing the conversion rate and demand will decrease “Machine” version of you won’t, since it will factor in the conversion rate of the previous strong months. By the time machine realizes that conversion rates have tanked, you have already over-spent with no results to show for it.
3. Does your target metric really matter?
Your target metric is supposed to be the most you are willing to pay for that action. For this test we ran with an automated bidding system we have a bidding folder and a control folder. We set a test target cost per action and that was the most that we were willing to spend. The folders ran for 18 days and the bidding was supposed to give lift to conversions all while staying below the target cost per action. Here are the results:
They missed the test target cost per action. The control outperformed the bidding folder. I gave them the benefit of the doubt thinking that it takes a couple of days to get the algorithm going and it should hit its stride in the last ten days of the test. Here are the results of the last ten days of the test.
Again they missed the target cost per action and instead of getting a downward trend it increased pretty significantly. The algorithm did not take in account the conversion was dropping the cost per action suffered because of it. The lessons learned from this is that if you have worked in account for a period of time and you know how the account is going to perform over time you can and will beat an automated system.
4. When is it ok to use an automated system
The perfect account to use an automated system is an account that has an unmanageable number of keywords to keep track of on a daily basis. By having a massive amount of keywords it helps as a buffer for the underperforming keywords and makes entire ad groups and campaigns perform. For those huge brands that have thousands of keywords and ad groups there is no way one account manager can keep track of that much traffic and be effective. That is what automated systems are designed for so keep that in mind if you are looking to go automated.