The AI Engine in GatorMail may sound complicated, but it is actually very easy to use on your campaigns once you know the basics.
This article will walk you through where the AI can be used, how to set it up on your campaigns, and what the results mean.
If you would like to know our goals with AI and how it can benefit you, please read the following article.
Typically you would be sending campaigns to customers and prospects, but you could also be sending to partners and other audience types.
When you create a campaign you can teach the AI engine the type of audience you are sending to, so it can make more accurate predictions for your future campaigns.
On campaign creation in GatorMail you need to set the campaign strategy type.
This can be:
- Prospect or
You can test your campaign through the AI Engine at any time when in the Campaign Details.
When you initiate the campaign to Live an AI check will be automatically run to make sure your AI is constantly being updated against the prediction model for better accuracy.
You will find this under the GatorAI section in the Campaign Setup.
It will give you a Predicted CTR. To get more detailed results you will need to click into the Title.
The check can take up to 1 minute to execute as it gathers your campaign data (audience, subject line and email content) and then runs this against your historical data model.
Once complete you will be taken to your AI report, each AI report historically is stored so you can see how your recommendation changes affect your Prediction Score.
Initiating an AI Check
1. In the Campaign Setup, select the GatorAI tab.
2. Select 'Initiate a new GatorAI' to start a new AI check.
AI Reporting Dashboard
- These are your Campaign Details. This is the campaign being used to test for: the subject line, the email content and the audience.
- This section allows you to go back to all the campaign AI checks you have run, as well as showing you the HTML of the email being checked.
- Data Modeled - This is where you can see how many campaigns, emails sent and the number of months of campaign data that has been used to build the prediction model. If you do not have 6 months’ worth of campaign data then you will see you have are modelled against the anonymised global model.
- The Click-Through Rate section shows how many contacts we predict will click from your final audience based on the AI predictions from previous campaign data, your average CTR for campaigns, the live CTR of the campaign, and the industry CTR which you can compare your success against.
- This will produce a Campaign Prediction Report in which a graph is produced representing every campaigns actual CTR vs the Predicted CTR. This is where you will see your "Spikes of Hope" and "Flatline of Nope" occurring. You can expand sections of the graph by using the point and drag function of your mouse.
- The left bar on the graph represents those contacts that are highly unlikely to click based on their past click history. If you click the bar on the graph you can create a group or make an exclusion group. The thinking here is why send to contacts that are not going to click? This will increase your delivery to the clutter and spam folder as it diminishes your audience engagement for email providers. You can use these new groups to send campaigns to, perform split tests on, and more.
The right bar on the graph shows you the contacts most likely to click. You could send a different email split to both groups to see the higher click-through rate of the two audiences.
- Based on your subject line and HTML/TEXT email content, the AI engine will recommend some changes that you could make to your content based on your historic CTR. For example, past campaigns with more images and links may have had a better CTR.
Also under the recommendation section is the "Flesch-Kincaid Score" that calculates the reading age of your email i.e. how complicated or simple the content is for the reader.
What’s the chance your email content is complicated enough to stop the recipient from clicking through due to their disengagement?
Important Note: When you receive a recommendation, for example of 10%, then you will only potentially receive an improvement of 10% of the Predicted CTR. So if your predicted CTR is 1%, you might get an improvement of 1%.