#BizTrends2020: Machine learning set to burst the bubble of digital marketing

Andrew Smit, digital data specialist at Incubeta

As we enter 2020, the rise of the machine will once again claw back the mystery of ad spend as we are forced to relinquish (at least some) control to Facebook and Google.

Everything is simpler and more confused

Google and Facebook remain the two main players in the digital space. The two mega-platforms commanded the lion’s share of the $283 billion global online ad spend in 2018.

In South Africa online ad spend is expected to exceed TV spend for the first time in 2023, placing immediate pressure on marketing professionals to turn to technology to help them master the digital space.

It’s ironic that in a world where Google knows if you have walked into a mall, marketing departments are finding it harder to track and measure the performance of their digital campaigns.

The old saying that "where there's mystery, there's margin" is now relevant again. The increasingly strict privacy laws that restrict the mega-platforms like Google and Facebook from sharing personal data means we aren’t privy to the deep insight they have on our own customers. And, while it is true that the data we are getting from the platforms is good for optimisation, it is not necessarily correct for overall reporting.

There is no single source of truth

The rise of machine learning will require digital marketers to give more agency to the main platforms such as Google and Facebook to manage and price the campaigns. Because they have so much data that they are legally prohibited from exporting to a third party, digital leaders have to trust the mega-platforms to act in their interest.

The harsh reality is that companies can’t create algorithms that will come close to competing with those created by the machine learning software of these platforms.

Another consequence of the power of the machine within these platforms is that you cannot expect to track across platforms. You will have to track each platform independently and, my prediction for 2020, is that we will see the signs of multi-touch attribution dying out. The only source of truth we will have will be the overall performance of the client.

This is not necessarily a bad thing. We believe that if you aren’t moving the revenue needle for a business, you are not going to be able to prove your worth. If the client is not seeing meaningful results in their business, then what’s the point?

If the real measurement of advertising efficacy is to be believed, we will have to implement uplift studies. We will need to measure the effects by using tools such as causal impact to measure the fetch of turning a campaign on and off and, based on the results, shift or tweak parameters.

As the new decade plays out, and the machine consumes more of our data and learns more about us, transparency will increasingly become the bedrock of trust between client and marketing agency. And, for Incubeta at least, this is not a bad thing.