CTV Advertising: 2026’s Urgent ROI Problem

By 2026, Sarah Chen, the Head of Media Buying at Zenith Digital, was feeling a new kind of pressure. A major CPG client had just greenlit a huge increase to their Connected TV (CTV) advertising budget, almost doubling it from last year. The problem was, Sarah couldn’t give them a single, credible report on what they were actually getting back for that money across all the different CTV platforms. “We’re spending more, but I can’t tell them exactly what they’re getting for it,” she told her team, pointing to a spreadsheet of mismatched metrics from Hulu, Roku, and Amazon Fire TV. For Sarah, the hunt for a universal standard in CTV measurement wasn’t just another industry buzzword. It was her daily operational headache. How was she supposed to prove these campaigns worked if everyone was speaking a different language?

Sarah’s situation wasn’t unique. For years, old-school linear TV was simple enough: Nielsen ratings were the accepted currency for reach and frequency. But with the explosion in CTV adoption, that simplicity is gone. Viewers scattered from broadcast to dozens of streaming services, and every service built its own proprietary data collection and reporting dashboard. By early 2026, Reuters was projecting CTV ad spending to hit $35 billion globally, but a huge chunk of advertisers felt like they were throwing money into a black box. This chaos meant Sarah’s team burned hours manually taping reports together, basically comparing apples to oranges, or, more accurately, trying to make sense of impressions from Hulu next to completed views on Roku without a clear way to track a conversion.

The real problem is identity resolution. On the open web, you had cookies and other trackers for a certain amount of user identification and ad targeting. CTV is a different beast. Most CTV devices don’t support cookies, and the walled gardens run by the big streaming platforms lock down their user data. “We can see a household IP address, sure,” Sarah explained in a strategy meeting, “but knowing if it’s John, Jane, or both of them watching, and then tying that view to a purchase? That’s the holy grail, and right now, it’s a blurry mirage.” This mess made it nearly impossible to de-duplicate reach and frequency, let alone attribute conversions with any accuracy. Even with Zenith Digital’s heavy investment in a sophisticated Trade Desk integration for programmatic buying, the platform couldn’t provide a truly unified view when the data coming in from the publishers was so inconsistent.

One of the more promising things Sarah was tracking was server-side ad insertion (SSAI). Instead of the ad being loaded by the viewer’s device, SSAI stitches the ad right into the content stream on the server. It improves the viewer experience by cutting down on buffering, and it gives publishers a more controlled environment for measuring ad impressions. “With SSAI, the publisher has more control over what’s delivered and how it’s measured,” Sarah noted, “which *should* lead to more consistent reporting. The challenge is getting all publishers to adopt a common standard for how they report those SSAI impressions.” Some smaller publishers were slow to adopt the tech because of cost and complexity, but the big players were getting on board, since they knew advertisers were demanding better data.

The industry’s failure to agree on a universal measurement currency was another major source of Sarah’s headaches. Was the key metric an impression, a completed view, or an engaged view? Every platform had its preference. For example, some platforms count an impression the second an ad starts, while others demand a certain percentage of the ad be watched. This makes comparing campaign performance across partners almost impossible. “It’s like trying to compare the speed of cars when one reports in miles per hour and the other in kilometers per minute,” Sarah quipped in frustration. “The numbers look different, and you can’t just convert them without knowing the underlying definition.”

Several industry bodies were trying to fix this. The IAB (Interactive Advertising Bureau), for instance, had put out updated guidelines for CTV measurement, pushing for standard definitions of impressions, viewability, and completion rates. But guidelines aren’t mandates, so adoption was spotty. Another key player, the Media Rating Council (MRC), continued its tough auditing process for measurement vendors, aiming to accredit solutions that met its criteria for accuracy and transparency. Sarah made MRC accreditation a baseline for considering any new measurement partner, but even then, she found that accredited vendors could have slightly different data interpretations.

And then there’s privacy, which complicates everything. With rules like GDPR and CCPA, and platforms like Apple pushing their App Tracking Transparency, the old ways of identifying users were on shaky ground. This reality spurred interest in privacy-enhancing technologies (PETs), especially clean rooms. Clean rooms, for example, create a secure space where advertisers can match their own first-party data against a publisher’s data without either side having to share raw, personally identifiable info. “We’re exploring clean room solutions with a couple of our larger clients,” Sarah mentioned. “The promise of getting better audience insights without violating user privacy is obviously compelling, but the tech is still new and a pain to integrate.”

What Sarah’s CPG client really wanted was simple: to connect CTV ad views to someone buying their product, whether online or in a physical store. That requires solid attribution modeling. Because there’s no consistent user ID across platforms, deterministic attribution (a 1-to-1 match) is mostly impossible. So, you’re left with probabilistic modeling, using clues like IP addresses, device types, and household demographics to make an educated guess. It became the fallback. It’s better than nothing, sure, but it’s loaded with uncertainty which is a tough sell to a client who lives and dies by hard numbers. “They want to know if their ad on Pluto TV led to someone buying their cereal,” she explained. “I can give them correlations, but correlations aren’t conversions.”

Unified ID solutions were another approach getting real traction. Initiatives like Unified ID 2.0, an open-source project, were trying to build a new, common identifier for the open internet (including CTV) that was based on user consent, typically using encrypted email addresses as a privacy-safe alternative to third-party cookies. Sarah knew this could be a breakthrough, but getting everyone, publishers, advertisers, and ad tech, to adopt it was a monumental challenge.

So Sarah and Zenith Digital were fighting this on multiple fronts. They started by only working with partners who were totally open about their measurement methods and were actually participating with groups like the IAB. They also beefed up their own data science team to try and stitch together all the messy data sets themselves, using statistical models to normalize the different metrics and build smarter attribution models. And they jumped into pilot programs for any new tech they could, like clean rooms and unified IDs, just to stay ahead of the curve.

Sarah figured a real universal standard would eventually emerge, probably as a mix of new tech, industry-wide agreement, and pressure from regulators. It wouldn’t be one magic bullet, but a framework that lets different systems talk to each other and report data consistently. Until then, her team had to keep adapting and pushing their media partners for more transparency and standardization. The CPG client was still demanding clearer ROI, but they respected that Zenith was in the trenches with them, wrestling with the complexity instead of just presenting glossy, meaningless charts. The finish line wasn’t in sight, but Sarah felt they were finally running in the right direction.