Product Update: Visual Builder for Machine-Learning Segments

Introducing a tool that will provide a new dimension of understanding when creating segments with machine-learning attributes.

The visual builder uses an interactive display to show the predicted reach of your audience segment according to the machine-learning model at work. Simply drag the cut-off slider to select the value that fits your objectives: do you want to create the smallest audience possible while retaining the majority of actual buyers? The visual builder allows you and your marketing team to make truly educated decisions.

Visual Builder predictive segmentation_2

Nice! But what exactly am I looking at?

An excellent question! The visual above shows a representation of the output our “likelihood to buy” predictive model will generate depending on how strict you tell it to be.

The X-axis of the bar graph shows a scale of numbers from 0 to 0.9. This is an index used to give a number value to how likely to buy each visitor is predicted to be. Closer to 1 means higher likelihood, closer to to 0 means lower.

The red bars represents the percentage of your total audience. The blue line represents the percentage of likely buyers captured within the audience at each point on the scale.

As you move the scale from left to right, narrowing the audience selection toward more likely buyers, you can see the size of your audience decreasing in the left-most pie chart above the bar graph. At the same time, the middle pie chart shows the percentage of likely buyers captured within the resulting segment.

In this case, you can see that just narrowing the audience to people with a likelihood to buy greater than 0.1 reduces the audience by over half, while still reaching 95% of likely buyers. It immediately gives you the ability to avoid wasting advertising dollars on non-converting visitors.