More than 15 years ago, Michael Learmonth published a report in Advertising Age that showed how few of the dollars spent in digital media actually went to publishers. Back in 2010, before programmatic really took hold of our industry, his research estimated that less than half of all dollars spent to advertise on premium sites actually went to those sites, thanks to the many intermediaries and third-party tech redirects that were implicit in any campaign.  

More recent research by industry trade groups, such as the UK’s ISBA (Incorporated Society of British Advertisers), puts that number at just over 51%, despite the growth of programmatic and the arrival of real-time header bidding. Interestingly, the research indicates that a significant portion of media buys—up to 15%—disappears completely during campaign flights. Here in the US, the ANA (Association of National Advertisers) has claimed for years that ad fraud runs as high as $80 billion a year in digital video alone. This 15% disappearance has been met with a collective shrug by media buyers in digital for more than a decade now, as companies like The Trade Desk, Amazon, Meta, and Google reap billions in fees while requiring buyers to use only their internal verification tools—or else.

Why have marketers settled for this for so long? Why has it taken so long for the industry to force platforms to submit to an audit that’s not run by the platform itself? Is this about enabling transparency? Or is it more about who gets the lion’s share of that 15% that nobody can account for? Or is it both?

Also Read: How Brands Turned the World Cup Into a Cultural Play

The “Incrementality” Trap

Is the recent announcement by The Trade Desk that they’re changing how they’ll pay audience partners with a new “incremental” model just another brick in this wall of occlusion?

If data company A and data company B both sell custom, curated audiences via The Trade Desk, and both perform exactly as expected, The Trade Desk soon will be able to choose which one they pay more and which one they pay less to based entirely on its own algorithmic scale of “incrementality.” You might think this is about performance—that is, which dataset drove more conversions. But it has nothing to do with performance and everything to do with differentiation as decided by a secret algorithm.

Payments will be tied to the incremental value a partner’s identity data provides—meaning, if a data partner contributes unique signals not already captured elsewhere in the demand-side platform’s (DSP) system. The shift is intended to deprioritize duplicate data and reward differentiation.

To see how this plays out in the real world, imagine two different business-to-business (B2B) data providers who have both mapped out high-value “Cloud Security Decision Makers.” Under this new model, if The Trade Desk’s internal system determines that it already has a similar audience footprint, Data Provider B is financially penalized and deprioritized—regardless of whether Provider B’s specific data actually converts better for the advertiser.

The Trade Desk promises to introduce new tools, including APIs and scoring mechanisms, to help partners understand how their data is evaluated under the revised model. However, those tools are not yet available.

When will they become available? Who knows? What will these APIs be based on? Again—who outside of The Trade Desk knows? The platform is essentially telling programmatic buyers that what’s good for Google is good for them: “Nobody else can check our homework except us.”

Meanwhile, advertisers are complaining that this is simply about cutting costs and making more from each campaign—controlling the economics of the open web’s programmatic segment even a little more tightly with each opaque turn.

Also Read: The Hidden $71 Billion Crisis in Digital Advertising

The Marketer’s Survival Guide: Taking Back Control

Marketers do not have to wait for the ecosystem to fix itself. If you want to stop letting platforms grade their own financial homework, you need to change how you buy and validate your audiences today.

• Demand Log-Level Data (LLD): Stop settling for slick, high-level dashboard summaries. Demand raw, log-level data from your DSPs. If a platform refuses to show you exactly which impressions your data ran against and at what exact cost, they are obscuring the truth.
  
• Audit Your Data Partners Directly: Don’t let a platform be the sole judge of data quality. Test your data partners via direct Private Marketplace (PMP) deals or data clean rooms, where you can verify the uniqueness and conversion power of the data yourself.
  
• Diversify Outside the Major DSPs: If a platform treats your budget with arrogance, shift a percentage of your spend to emerging, independent programmatic environments that offer open APIs and welcome external validation.

Also Read: Are We Using AI to Help Customers or Avoid Them?

The Next 24 Months: The AI Disintermediation

If a performance media buyer compensates media sellers only for inventory that performs, the seller has the right to reserve inventory for other kinds of buys, such as impression-based campaigns. The Trade Desk says, “Not so fast. We get to decide which data is adding incremental value in cross-media campaigns based on KPIs that only we will see.”

One wonders how long the buyer community will tolerate such arrogance. Buyers will continue to pour money into this segment and wonder where their dollars are actually going—but not for much longer.

This friction is setting the stage for a structural collapse of the traditional ad-tech stack. The tech giants believe they have a permanent lock on the market because their ecosystem is too complex to bypass. They are wrong.

Very soon, enterprise brands will deploy proprietary, AI-based buying tools. Instead of logging into a clunky DSP dashboard, navigating hidden fee structures, and accepting opaque “incrementality scores,” a brand’s internal AI will talk directly to a data seller’s API. The AI will look at your first-party CRM, securely match audiences, negotiate real-time pricing, and deploy the ad directly to a publisher’s server.

By automating the negotiation and execution layers natively, AI completely cuts out the middleman’s 15% disappearing act, rendering the traditional DSP moot. Many industry insiders expect this shift to occur in as little as 2 years.

The post How Long Will Marketers Settle for Opacity in Programmatic? appeared first on MartechView.

In a recent discussion with a marketing analyst, she noted that while brands are aware of AI tools and view them positively, adoption remains slow.

The primary reason is that brands often believe their data is not ready for AI tools.

This is surprising, as brands typically require minimal preparation to use AI. Many are held back by common misconceptions, which can be addressed as follows:

Myth #1: Not Enough Data

Many brands hesitate to adopt AI because they believe they lack sufficient data. This perception is understandable, as AI marketing often emphasizes processing large data sets.

However, brands do not need large data sets to benefit from AI-powered insights. Small and mid-sized businesses often have enough customer records to leverage AI for audience targeting or customer acquisition modeling.

Ideally, a brand should have 20,000 customer records to import into an AI system, though 10,000 records can still yield meaningful results. Brands with only a few hundred records may not benefit significantly, but established companies with extensive customer profiles are well positioned to begin.

Myth #2: Disorganized Data

Even when brands have sufficient data, they may hesitate due to concerns about data organization. However, many current AI tools can organize data for brands or agencies and return it in a usable format.

AI tools can enrich datasets and connect customer profiles, which is a fundamental capability. Brands may use AI solely for data organization before exploring more advanced features, allowing for a gradual approach to adoption.

Also Read: The Organizations That Survive Disruption Never Had to Recover From It

Myth #3: Cannot Afford New Staff or Technology

A major barrier to AI adoption is the anticipated demands of integrating new tools, including concerns about staffing, training, and infrastructure.

This is based in reality. Brands have spent the past two decades navigating the integrations of buying tools and data and customer management platforms so that they can take advantage of this data-driven age. 

While it may seem that AI requires similar integration efforts, much of the process can occur within a single application. This eliminates the need for additional internal technology stacks or extensive engineering, as the necessary infrastructure is managed externally.

With app-based AI tools, training and staffing requirements are minimized. While some workflow adjustments are necessary, these tools typically do not require hiring new staff or extensive training.

Fearless Adoption

These three myths are understandable, but if they persist, they will hinder brands from embracing AI-powered advertising. Most brands are already equipped to begin using AI tools, even on a trial basis. Overcoming these misconceptions will encourage adoption and drive further innovation in marketing.

The post Three Myths That Are Keeping Brands Away From AI appeared first on MartechView.