One of the most common challenges faced by advertisers is lead quality. In a recent blog post, I outlined 16 potential solutions. One of them has something to do with your conversion event, which is something I’m trying now.
This is how it works…
Standard settings
When you run ads to increase site leads, you need to select a conversion event. In most cases, you will use lead or complete registration standard events. The meta will then focus on as many movements as possible.
The problem is that you want more than that action. Quality leads will take other important actions after subscription. Meta doesn’t know this, which leads to low-quality leads.
This is the basic weakness of meta-advertising optimization for delivery. The algorithm is literally focused on getting the required actions, which are defined by performance goals. Quality is rarely considered.
Solution
One solution is to change the optimization event to something that better reflects quality clues. How you define it will depend on your business and other factors (I will do it in one minute).
When they perform important actions, I tag subscribers in CRM. The easiest way to isolate good leads badly while retaining the number is to focus on the people who open my email and click on the link. I created a custom event for these operations and passed using the transformation API.
Then, you guessed it, I optimized the event – instead of lead or finished registration.
Require
I use 7-day clicks in the ad set as my attribution settings. I don’t want any viewing conversions, I need to have more than a day after clicking to take this action.
For this, there are some requirements for the events you are using. First, the event needs to be performed within 7 days of subscription to make it belong to the 7-day click window (early the earlier the earlier). Second, make sure you get enough volume to optimize correctly (ideally 50 or more per week to exit the learning phase).
If you can correctly define the mass action that occurs after a subscription that results in sufficient volume for proper optimization in less than seven days, this may be a great choice for testing. Otherwise, it might just create a bottleneck with very few clues.
Have you ever tried it?
