Sixty-one percent of B2B marketing teams used some form of multi-touch attribution in 2023. By 2024, that number climbed to 76%. And yet, ask any demand gen leader whether their measurement actually tells them where pipeline comes from, and the honest answer is usually a shrug followed by a spreadsheet workaround.

The gap between "we have attribution" and "attribution tells us something useful" isn't a technology problem. It's a setup problem. Marketers built it themselves.

The Self-Inflicted Wound

Angelina Eng, founder of Enso Horizon and former IAB measurement lead, put it bluntly in a September 2026 piece for MarTech: "Marketers configure systems differently and create their own categories and definitions based on their needs. Then they blame the industry when the data doesn't reconcile."

Campaign naming conventions turn into strings of codes nobody outside the original team can decode. Tracking pixels get deployed inconsistently across platforms and agencies. Each choice looks small in isolation. Together, they make cross-channel measurement structurally impossible before a single impression serves.

Multi-touch attribution needs clean exposure and conversion signals. Marketing mix modeling needs stable, comparable inputs over time. Incrementality testing needs a valid holdout designed before media runs. If none of those requirements are built into campaign setup, the data can't answer the questions you'll ask six weeks later. That's not a vendor failure. That's a planning failure.

Customization Becomes Fragmentation

Eng described launching Project Eidos at IAB to develop a common taxonomy for ad inventory. Agencies responded: "We already have a taxonomy." Then they admitted they maintained a different version for every client.

Every custom label, every bespoke reporting category adds translation work before any analysis can begin. Analysts spend hours reconciling before they get to the part that matters: figuring out what actually drove pipeline.

For B2B SaaS teams, this fragmentation hits hard. Paid search already consumes about 33% of digital media budgets at a median cost per lead of $208.66 (B2B technology, per Starr Conspiracy benchmarks). When your measurement can't tell you which of those leads became SQLs, you're flying blind on the most expensive channel in your mix. Retargeting captured 42% of B2B display spend, and without incrementality measurement, you're almost certainly over-crediting conversions that would have happened anyway.

AI Will Scale the Mess

Eng's sharpest observation: "Give AI bad data and it will scale the problem, then deliver the answer with confidence."

When an AI agent tries to compare performance across five platforms, each defining "engagement" differently, the model produces a number that looks precise and means nothing. The confident dashboard becomes a confident lie.

Only 12.5% of A/B tests reach statistical significance with a positive lift, according to a 2023 testing report. Many optimization decisions are already noise-driven. Layer in AI agents making allocation recommendations on mismatched inputs, and you've automated the wrong answer at scale.

Three Things to Fix Before 2027 Planning

Audit your campaign setup against measurement requirements. Before launch, ask: can MTA trace exposure to conversion? Can MMM get stable inputs across quarters? Is there a holdout built for incrementality? Review naming conventions, tracking configurations, and attribution settings across every platform and agency. Fix structural gaps before spending.

Adopt a common taxonomy and stop treating customization as harmless. IAB and IAB Tech Lab have published standards for audiences, ad products, inventory, and measurement signals. They're not perfect. They don't need to be. Custom reporting views can sit on top of a shared structure; they shouldn't replace it. Require partners to disclose which standards they support and which fields they've modified. Put it in contracts.

Require independent verification, not claims of compliance. As of August 2026, the MRC's digital accreditation directory listed about 40 digital service listings from roughly 25 providers. Compare that to the number of platforms you fund. If you're relying on unaccredited metrics for budget decisions, that's a choice you're making, not a problem being done to you.

The Uncomfortable Part

Walled gardens deserve criticism. Platforms controlling their own data, inventory, and reporting rules inside closed ecosystems is a real constraint; over 60% of survey respondents have flagged it as a severe measurement challenge. But the walled-garden complaint has become a convenient distraction from the parts of measurement that marketers actually control: setup, taxonomy, verification requirements, and which partners get renewed.

You write the check. You decide whether naming conventions are enforced or whether every agency maintains its own dialect. The measurement stack isn't broken because the industry failed you. It's broken because the setup decisions that feed it were never designed to answer the questions that matter: what drove qualified pipeline, at what cost, and would it have happened without the spend.

Fix the inputs, and the outputs start telling the truth. Keep blaming the platforms, and next year the complaint will just have "AI" in the subject line instead of "signal loss."