Interviews, insight & analysis on digital media & marketing

Marketers are no strangers to AI-led media buying – but are they using it to full effect?

By Matt Nash, UK MD, Scibids

It is fair to say that digital marketers are no strangers to AI, leveraging elements long before ChatGPT brought it into vogue in 2023. In fact, a recent report we compiled in collaboration with Censuswide found that 94% of brand marketers are currently applying AI in their digital advertising. Nevertheless, the recent AI craze has arguably affirmed the technology’s role in ushering in the next phase for our industry. Indeed, those who went to Cannes last month know how difficult it was to cross the Croisette without hearing those two letters.

However, while veterans of utilising pre-packaged AI tools for smarter, more efficient and more effective campaigns, digital marketers are still not making full use of its capabilities. In particular, when it comes to programmatic ad buying, too many marketers are relying on ‘off-the-shelf’ algorithms which offer basic functionality for optimising towards standard media KPIs that are often uncorrelated to real business outcomes. At a time when CFOs are scrutinising media budgets, this is not ideal.

We know that with the depreciation of the cookie, there is going to be less readily available data for advertisers. We also know that the data that is available is often stored in different and disparate systems so the challenge is how do you bring that data together, make sense of it, and finally make it actionable. Trying to do this manually is impossible, and the off-the-shelf tools are not suited to this either, meaning we need to find new ways of achieving this.

The market-leading, Demand-Side Platforms (DSPs) recognise this challenge and in recent years, in addition to their standard ‘off-the-shelf’ offerings, have opened up their APIs to make it possible for brands to bring their own intelligence into the programmatic bidding process for advanced optimisation.

Leveraging the most advanced features and capabilities of the DSP at scale here requires the use of far more sophisticated AI than is regularly available as standard. In particular, it demands the use of customisable algorithms, which can better cater to the unique business needs of digital advertisers. It’s therefore no surprise that interest in this technology is growing rapidly. But what are customisable algorithms?

Customising your media buying

Customisable algorithms are a custom set of bidding rules executed by DSPs but (dynamically) generated by an AI to deliver advanced media buying outcomes aligned to a specific business’s goals. Example metrics include qCPMs which often take into account external (to the DSP) data from measurement, attention and verification companies.

Because the algorithm is designed for a specific campaign, it ensures that not only are the outputs more tailored, but also that the inputs are as well. At a base level the AI utilises the non-user-specific semantic and contextual metadata from the DSP log files, which brands and agencies have previously looked over in favour of third-party data sets. This allows the advertiser to find customers more efficiently without being forced to rely on third-party cookies. Simultaneously, these algorithms enable marketers to introduce their own first-party data and consented third-party data sets.

Customisable algorithms represent the most valuable AI-based tool available to digital marketers currently, enabling a greater degree of transparency as well as more control over what is increasingly regarded as an opaque programmatic supply chain.

Looking ahead

Using standard media metrics as a tool for measuring success alone will no longer be effective. Brand marketers need more from their investments and their campaign data, which now must be based on business-specific KPIs. This is where custom algorithms come in.

The good news is that we found that the vast majority of senior UK brand marketers (91.7%) are informed about the opportunities for customisation within their programmatic strategy. This statistic gives us hope that the industry will begin to embrace customisable AI wholeheartedly, becoming more mainstream within media buying.

The biggest challenge for brands and agencies is how to leverage customisable AI at scale? Smart people, given enough time, can manually create one customisable algorithm script for a single campaign. However, unless you can hire hundreds of data scientists, it’s impossible to offer this vital layer of technology at scale across multiple campaigns in multiple markets. A powerful AI answers this challenge by automating the creation and deployment of customisable algorithms up to 4 times a day, every day, for any and all campaigns in an individualised manner. The AI progressively learns about the value that’s being created, adapts to changing market conditions, and pushes new, evolving rules throughout the campaign.

As well as freeing up teams to perform other high value tasks, digital marketers using a powerful AI to deliver customisable algorithms can streamline and optimise media investments and ad effectiveness, demonstrating the value of digital marketing to decision makers within their organisations at a time when it matters most.

*Scibids is a client of Bluestripe Communications, owned by Bluestripe Group, publisher of NDA