The Ad Waste Buckets: A Nuanced Approach To Defining And Measuring Ad Waste

Research suggests that up to 43% of digital ad spend goes to waste in some markets, with several global estimates putting ad waste figures at nearly a quarter ($20bn) of programmatic ad spend – a massive amount, by any measure, and a clear wake-up call to the industry to prioritize quality rather than chasing low-cost, low-value impressions.

But, how should we define ad waste when it can take so many forms, from non-rendered, non-viewable, or low-attention impressions to made-for-advertising (MFA) sites and ad fraud?

At Picnic, we define ad waste as ‘any impression failing to drive a meaningful outcome’. In our view, an impactful impression must be genuinely seen and engaged with by real users, leading to some form of meaningful brand impact.

Anything other than that may be considered a wasted impression.

The Three Buckets Of Ad Waste Management

Defining ad waste is one thing, but measuring it is another. We advocate a nuanced approach, which categorizes ad waste into three distinct ‘buckets’ – each requiring a tailored strategy for measurement and mitigation.

1. Objectively measurable: non-rendered and non-viewable ads

At the core of our measurement strategy are tracking pixels, which serve as beacons in the fog of ad impressions. Non-rendered and non-viewable ads are objectively quantifiable, using the IAB’s definition of viewability, which requires 50% of pixels to be visible for one second or more, providing a baseline for ad waste metrics.

Yet it’s important to understand that a ‘viewable’ impression does not necessarily guarantee that an ad has actually been seen. Other, more nuanced conditions, focused on the user’s experience and ad attention, must be met in order to measure the quality of views.

2. Subjectively measurable: low-attention impressions and MFA

While we are yet to reach a consensus in the industry on how we should define and measure attention, there is a growing acceptance of the fact that a creative needs to achieve a certain threshold of a user’s attention to have a real impact.

Another ad waste indicator is the presence of made-for-advertising (MFA) sites that have become synonymous with low-quality media – those built purely to generate revenue, without thought for content quality or user experience.

MFA sites are often spoken about in binary terms – a site either is MFA or isn’t – but in reality, it is not that clear-cut. Our own MFA model is probabilistic in nature, ingesting relevant page signals to give an estimate of how likely it is that a given page is built solely to generate ad revenue.

3. The enigma: ad fraud

Ad fraud – the elusive phantom haunting the digital ad landscape, defies easy measurement. It is generally a catch-all term for online actions undertaken to deceptively represent views, clicks, or engagements with an ad, enacted by bots, malware, humans, or some combination of the three.

As the industry engages in a perpetual cat-and-mouse game, and fraud detection becomes more sophisticated, fraudsters, in turn, develop new ways to go undetected. Yet recognizing and discussing the issue is a key component in tackling ad waste.

Emptying The Ad Waste Buckets For A Cleaner, More Efficient Advertising Ecosystem

As we navigate the intricate landscape of ad waste, our priority should be to highlight and define the metrics that matter, paying attention to the various buckets of ad waste and focusing on how we can empty them.

The necessary strategic vision involves not only the more traditional and objectively measurable criteria but also a close understanding of less traditional metrics such as media attention and how MFA sites work – as well as a vigilant stance against the ever-evolving threat of ad fraud.

Together, we must apply insights from these nuanced measurements of ad waste to inform industry practice and develop actionable insights from that data.

By doing so, we can make a real movement towards maximizing the transparency, quality, and impact of digital advertising investments.

Also published in: AiThority