Mediahub’s ‘carbon-killing algo’ claims to solve attention-value paradox

Mediahub’s ‘carbon-killing algo’ claims to solve attention-value paradox
Erfan Djazmi said Mediahub's ARC algorithm could massively reduce carbon emissions when applied at scale

IPG Mediabrands agency Mediahub is rolling out a bespoke algorithm that it claims enables advertisers to optimise media-buying for more audience attention and lower-carbon formats at the same time.

Dubbed “Attentive Reach and Composition” (ARC), Mediahub’s automated attention optimiser promises to find an ideal trade-off point between two conflicting online media-buying challenges set by advertisers: maximise audience attention (which usually means investing in expensive and carbon-intensive formats like video), and reducing a campaign’s carbon footprint (which increases the risk of using less effective formats).

This means that ARC cuts the long-tail of ads that are not seen for long enough to drive impact. When optimising for attention, the usual outcome is a reduction of 10-15% in impressions. However, attention-optimised impressions tend to correlate with better brand and business outcomes.

According to data from Lumen Research, Mediahub, and GoodLoop’s carbon calculator, implementing this approach in the UK digital industry in 2022 would have reduced carbon emissions by 4.8 million tonnes, based on a digital marketing ecosystem amounting to £25bn annual investment. This is equivalent to removing all plastic bags (3 billion) from UK circulation each year.

“The digital ecosystem is a cesspit of ad and content clutter,” Erfan Djazmi, EMEA chief digital officer at Mediahub told The Media Leader. “Advertisers struggle to make programmatic work due to legacy publisher ad models built on dated metrics such as reach and viewability, for instance. Rather than wait for publishers to evolve their ad experiences, we hired software and data engineers to program our own algos, using Lumen eye tracking data.”

Media grabs attention but creative keeps attention

Mediahub’s team of developers connect Scope 3, Lumen attention data and DSP data with the use of a customised Python computer-programming script, which allows them to define a multi-factor bid across addressable media. This bid, the agency says, maximises attention while reducing carbon emissions, all while retaining an efficient bid price — activated in real-time through custom bidding algorithms — that correlates well to an advertiser’s brand and sales outcomes.

The risk of this approach, however, is that an advertiser using this tool may lose out on advertising on high-impact formats if they are deemed “sub-optimal” in terms of maximising attention and reducing carbon footprint.

To prevent this, Mediahub has also created a partner marketplace that trades solely on attention by setting a minimum threshold of attentive seconds at a fixed price, while still maintaining volume and price controls. The marketplace (a cluster of publishers and media owners) is described as an “outcomes-based trading model” that uses the Lumen APM (Attentive seconds per thousand impressions) metric instead of traditional volume-based cost-per-thousand trading models.

The algorithm has been successful in reducing cost-per-brand lift by 75%, Djazmi said.

He added: “If you optimise blindly to attention only your ‘brand affect’ goes up but you’re paying for more expensive inventory. If you don’t control price, you’re cost-per-brand lift doesn’t pay back with attention and it becomes cheaper to go with a viewability floor. That’s why we’ve ring-fenced the [display video] inventory to a sub segment of inventory that is seen with an efficient bid price.”

Data and software engineers based in Mediahub’s London office have been piping in scripts and algorithms that the agency is now activating in different markets for its advertiser clients, such as Arla in Australia, and Pinterest in Europe.

Media Jobs