Skip to content
artificial intelligence  •  Machine Learning

Faster content creation with machine learning – case headline AI

Jun 14, 2019

AUTHOR

Kimmo Ihanus

Growth Hacker at Avaus
kimmo.ihanus@avaus.fi

Have you ever been in the endless limbo, where the entire team is debating which headline to use, or which image will bring most conversions? And no one’s really sure what works, but everybody has their own opinion? 

 

At Avaus, we feel that marketing teams should focus on more relevant questions: What is the potential target group? What do they value in our product? Is the user experience optimal yet? How can we optimise the message for different segments and channels?

 

That’s why our growth hackers are actively experimenting to find new ways to do things more efficiently. In many cases where we humans think we know the answer, machines actually do the job better, when given enough data to practice with.

 

A good example of such a problem is the email subject line. Many marketers know that, usually, a good open rate doesn’t necessarily correlate with high CTR or conversion rates. Still, in many cases, marketing teams “sacrifice” significant chunks of the target group for pre-testing of different subject lines. If enough of historical data of previous campaigns is available we can avoid this by training a machine to do the decisions for us.

 

To demonstrate this we created a demo application that uses Avaus’ own email data to predict email open rates. Our own database doesn’t contain enough variations to make reliable predictions but you can play with it to see how it works here!

 

This solution is already in use for several customers. According to our tests, the algorithm is better at predicting open rates than the most seasoned email marketers – if the amount of variation given to the machine is high enough. The solution does not guarantee success but it frees valuable time so that marketers can focus on planning and designing more important things.

 

The same approach can be applied to other channels (SEM/SEO, digital advertising, Facebook and many others) which we are actively experimenting with.

 

Contact us to learn more:

 

Contact us