What gets measured gets managed: Measuring digital ad emissions
The importance of measurement.
As discussed in the first instalment of this blog series, the spotlight has turned towards measuring carbon emissions in the digital marketing industry, a pivotal shift at a crucial moment. However, for advertisers to begin engaging with decarbonisation efforts, they first need to identify where in the supply chain carbon emissions are being generated, and how much.
Great news. We have the tools to measure, and we are now aware of the impact our work is having, so we have no justification to not act.
Defining what “measuring” means.
Advertising is the lifeblood of the free and open internet, but it relies on a vast network of energy-hungry infrastructure to facilitate the delivery of each advert. Behind every transfer of data between an ad server and the end-user, there are millions of signals sent between technologies. Targeting, real-time bidding and tracking verification, all of which require electricity, in turn producing greenhouse gases. We can estimate the emissions produced by an ad by quantifying how much electricity has been used by each element in the chain of its delivery and the carbon intensity* of energy production at each step.
*a measure of energy cleanliness and is calculated by how many grams of CO2e are released to produce one kWh of electricity.
“We know our activity has an impact, and we know we have a responsibility to address it.
Carbon measurement enables us to see where, and how much, carbon is generated at each point in the supply chain which will help inform your reduction optimisations, but it’s no easy task.”
Abigale Borsberry - Head of Programmatic Media Operations
Mixed measurement methodologies but GARM & Ad Net Zero's new measurement framework on the way!
The automation and complexity of the supply chain can obscure how and where we create emissions, but while the industry wait for the recently announcement GARM/Ad Net Zero measurement framework, making like-for-like analyses and the setting of industry standards impossible. However, this is not an unfamiliar dynamic for the digital advertising industry. Advertisers often use a variety of third-party vendors and verification partners as their source of truth for delivery, viewability, brand safety etc.
It is a dynamic we’re adept at managing.
How you measure.
The two main approaches to measurement are based on either reported campaign data or real-time data, involving the use of tracking pixels to wrap ad tags, enabling you to measure the carbon output of adverts (and the surrounding adtech) in real time.
Both will use data from elements such as carbon intensity of the local energy grid, weight/duration of creative assets, day of week/ time of day, device, bandwidth usage, device and connection types, publisher site & placements and programmatic supply (SSP).
It’s important to note both approaches will be based on estimated outputs.
Variances.
Targeting parameters can lead to significant variances in the carbon output of an impression.
-
Each country has a unique mix of energy sources and infrastructure so we can see great variations in carbon intensity* between countries. For example, in 2022 Electricity Maps reported New Zealand, with well-developed hydro and geothermal infrastructure, as having a Carbon intensity of 96g Co2e/kWh, whereas Australia, which is much more coal-dependent, averaged 520g Co2e/kWh. Data recorded when testing with a measurement vendor demonstrated the significance of the variance. In the same campaign, New Zealand averaged a carbon output of 0.12kg Co2e/1000 impressions, whereas Australia averaged nearly twice as much, at 0.23kg Co2e/1000 impressions.
-
The time of day and day part can also affect carbon intensity as Energy Grid operators change the energy mix to meet demand. For example, Solar Energy yields drop during the night and winter months and during energy surges, such as the TV pickup phenomenon* in the UK, high carbon intensity alternatives, like coal are often used to bridge the gap.
* Phenomenon when many people simultaneously watching the same programme use commercial breaks to operate electrical appliances resulting in large, synchronised surges in national electricity consumption.
-
Devices and Operating systems have variances in energy efficiency. Information captured via a measurement partner in real-time underscored the importance of the observed difference. In the same Display Campaign, we saw Windows (Desktop) and Android (Mobile) both returning higher Co2e/1000 impressions than Mac OS X (Desktop) and iOS (Mobile).
The cost of measurement.
Carbon measurement does come with an associated cost. In tough economic conditions budgets are always put under strain to deliver the best return on investment (RO) possible and it is an unfortunate reality that carbon measurement is currently considered a luxury rather than a hygiene factor/standard requirement.
Goodhart's Law aptly warns us that "when a measure becomes a target, it ceases to be a good measure". If we make carbon emissions the sole target, there's a risk of unintended consequences. Essentially, the act of using emissions as a performance metric could inadvertently encourage strategies that undermine campaign effectiveness while artificially boosting the measured environmental performance. It's a delicate balance between environmental responsibility and ensuring the overall success of our digital advertising campaigns.