April 3, 2018
Automation is all the rave in the digital marketing world. Marketers can use machine learning for bidding, ad copy, display campaigns, app campaigns, etc. The list goes on. Even if a digital marketer isn’t using a smart bidding strategy or have smart display campaigns, marketers are still benefiting from machine learning as Google still uses all these data points to know if a searcher will click on your ad. As awesome as automation is—and it is truly awesome as I as a marketer have seen how it can improve campaign performance—it cannot be taken as an end all be all, the solution to all performance problems and completely taking out human rationale. One smart bidding strategy to dive into is Target CPA and how, if a campaign is being held back by budget, this automation can actually take a toll on your performance rather than improve it.
Recently we tested a Target CPA automated bidding strategy with an education client that is tightly tied to strict CPA goals for specific locations. At first thought, this was a great solution to drive conversions without daily manually monitoring costs per leads. That was tedious and could easily slip. So, we set each campaign, targeting different locations, to their respective CPA goals. We knew that the machine needed time to learn, so when we initially saw CPAs fluctuating we weren’t too concerned. However, as we were nearing the 14-day mark (Google’s recommended learning period) we were still seeing higher CPAs than our set goals.
In reaching out to Google, our Google rep provided extremely helpful insights into what was happening within the campaigns. Specifically, two campaigns were having two different sets of problems. Campaign 1 was consistently running into budget constraints. Target CPA bidding strategy is not able to work in the most optimal way when campaigns are restricted by budgets or when there isn’t any room to grow in impression share. What was happening in Campaign 1 was not only was it constrained by budget, but it was also a lower volume campaign. If a campaign is constrained by budget, but is a high-volume campaign, Target CPA can still optimize and improve performance if there is enough volume/traffic. Campaign 1 had a daily budget of $400/day and a daily conversion volume between one and five conversions. Some days when the campaign would only get one conversion, the CPL may go as high as the daily budget limit – in this case $400. When this happens, Target CPA bidding tends to look more like maximize for conversions rather than optimizing toward the CPA goal. The machine thinks one conversion is better than none for the day.
From here, we had two options to take. We could 1) make the switch to maximize for conversions and manually control and manage the costs per leads, or 2) continue with Target CPA but remove the budget constraints. We could do this by turning off keywords or ad groups that were not converting but spending to allow more spend to go toward what was converting.
Before discussing which of the options we took, Campaign 2 showed us a different story. Campaign 2 was more stable than Campaign 1; it had no budget restrictions, a much higher daily budget with a much higher conversion volume. In this case, it had the lower CPL that stayed relatively consistent throughout the month. For Campaign 2, daily CPLs still fluctuated slightly higher than what our goal was for the month, but Target CPA would optimize toward the most conversions while trying to stay within a close range of our CPA goal. This was not exactly a problem, as we were trying to get the most conversions at a certain goal; however, the drawback was that if we weren’t hitting the goal CPA anyway, we should switch to maximize for conversions and get what leads we can even if we raise our costs of leads by 10%. (This particular campaign was considered a ‘high risk market,’ meaning it was very important to hit lead goal). So, for this campaign, Target CPA was not getting us the amount of conversions we could get by staying within CPA range – meaning we were limiting ourselves in leads when the leads for this market were more expensive than we originally projected. Our option for this campaign was to decide which goal was more important: the costs per lead or the amount of leads.
For Campaign 1 and Campaign 2 we switched from Target CPA to maximize for conversions. Even though both campaigns had presented us with different problems, Target CPA was limiting our campaigns—ultimately our monthly goal—by restricting spending or restricting volume. Maximize for conversions allowed each campaign to optimize towards the most volume, which was the priority. Target CPA is a great automated bidding strategy when campaigns are well trafficked and have room for growth within impression share and within costs. When trying to decide how to increase performance of your campaigns, consider automated bidding strategies but be sure to look at these strategies in depth before universally applying them to campaigns.