By Robin Bootle, Co-Founder & COO, Multilocal
In a world where data informs our every advertising decision, the efficacy of each action we take to get advertisers closer to consumers depends on clean, well-maintained data and good data hygiene. Yet we know from the ANA’s Transparency Study that a substantial portion of bidstream metadata cannot be fully validated. In our own work, we’ve found that nearly 35% of information passed through the bid stream is incomplete or incorrect.
This creates a problem that compounds throughout the supply path. Errors create wastage, wastage increases cost per outcome, and rising costs ultimately deter advertisers who may pull investment from the open web altogether.
This isn’t a demand-side problem. It’s a supply classification problem. And it’s fixable.
We Have Standards, but Not Standardization
Taxonomies are the systems that tell us what inventory is available and which audiences it reaches. Standardized taxonomies applied consistently across platforms are a prerequisite for clean data across the supply chain. This baseline is the only way we’ll be able to achieve meaningful audience segmentation at scale.
Frameworks like the IAB Contextual Taxonomy or Google Topics exist, but they are not universally adopted. Even when they are used, interpretation varies widely across supply sources.
An overreliance on technology at the input stage can also create data hygiene issues. Moreover, many platforms rely on “rented” taxonomies, assuming the data coming in is complete and correct. Rented taxonomies are optimized for speed and scale, but not accuracy. They also rely on someone else’s interpretation of categories, which brings us back to the need for standardization.
Opacity Isn’t a Good Business Model
Some suppliers are reluctant to move toward standardization because quality assurance at the input stage is hard, time-consuming, and expensive. But there’s another reason platforms resist standardized taxonomies. They benefit from opacity.
Opacity has historically been part of the strategy. It is seen as a moat around proprietary supply and audience assets, and some still believe standardization will reduce the commercial value of direct sales relationships.
But that moat neglects advertisers’ need for audience and optimization insight, and it will likely come at the cost of advertiser trust.
Calibration Is Where Human Judgment Still Matters
A standardized taxonomy is a necessary first step, but calibration is what ensures clean data. Calibration helps guarantee that inventory is classified correctly so that audiences, context, and quality align with what advertisers are trying to achieve. This requires human vetting beyond the domain level to uncover issues that automated systems may not fully understand.
Classification is often subjective. Take the automotive category as an example. A brand wants to advertise in environments that reach car shoppers. Publishers might classify their content using any number of labels: car, vehicle, auto, family car, SUV, Volkswagen, sedan, or Wrangler, to name just a few. Automated systems may miss these nuances. On the flip side, they may exclude related categories, such as motorcycles, that could still reach a valuable audience.
Leaving taxonomies entirely in the hands of automation creates blind spots, missed opportunities, and wasted spend. This is why humans need to remain in the loop—not just to validate outputs, but to continuously calibrate and improve the quality of signals over time.
Three Questions Advertisers Should Be Asking Right Now
As planning, targeting, and optimization become increasingly automated, platforms that invest in data qualification and completeness will be better positioned to deliver relevant, high-quality opportunities for advertisers.
In the meantime, advertisers should be asking everyone on the sell side (SSPs, publishers, data companies, etc.) three critical questions:
- Is your classification system fully automated or human-vetted?
If humans aren’t in the loop, the risk of misclassification and wasted media increases. Human oversight helps ensure that inventory, context, and audiences are aligned with campaign goals. - Is the data your own?
Advertisers need transparency around data sources. What does the platform truly own, and what is licensed or rented from third parties? If data comes from elsewhere, ask what due diligence has been done to ensure accuracy and completeness. - How is the data calibrated and kept current?
Ask what processes are in place to recalibrate and refresh classifications over time. Is there a feedback loop where humans validate and correct outputs so models improve? Also ask how frequently calibration occurs, because even high-quality data can become outdated quickly.
When Calibration Works, Everyone Wins
When the data is clean, wastage drops, performance improves, and user experience is better. The right message appears at the right time, in the right context. Advertisers win. Media owners win. The open web benefits from increased investment.
As planning, targeting, and optimization continue to automate, especially in an agent-driven future, the quality of inputs will matter more than ever. That starts with clean data and verified signals. Those who invest in human-led calibration will outperform those who simply hope a rented taxonomy has done the hard work for them.
