The Discipline Nobody Owns

Posted by Mike O'Brien on July 15, 2026 in Postcards from the digital frontier | Leave a comment

Marketing after certainty: leading when AI, data and old-school friction and trust collide

For most of the last two decades, a brand’s marketing spend passed through separate hands. A brand agency held the narrative. A digital agency ran the media. A direct or CRM agency owned the database. Each discipline had its own specialists, its own vocabulary, and, more importantly, its own incentive to be sceptical of the others. A digital media buyer proposing a channel because “we should be on it” would eventually meet a brand planner asking what audience insight justified it. A CRM team over-segmenting a database would meet a brand team worried about coherence of voice. The separation was inefficient. It was also, without anyone designing it that way, a check.

That check is disappearing, and it is disappearing for reasons that sound entirely rational. Agencies have spent five years converging into full-funnel, always-on “growth partners”: brand shops moving into experience design and owned channels, digital shops absorbing the whole customer journey from discovery to retention, direct and CRM agencies pulled into the centre of the model as first-party data becomes the only reliable signal left. Margin pressure has pushed the same convergence from the commercial side, with automation and in-housing forcing agencies to productise their offer rather than sell discrete campaigns. Both trends are genuine responses to genuine problems: privacy regulation, platform fragmentation, and the collapse of the neat above-the-line and below-the-line boundary that used to organise the industry. Neither trend was designed to remove scrutiny. It has removed scrutiny anyway.

What the convergence actually cost

The evidence for what happens when marketing logic is separated from budget control was established well before agencies converged, and it has not gone away in the years since. The 2020 ISBA and PwC supply chain transparency study, the first attempt anywhere to map programmatic spend end to end, found that only around half of advertiser spend reached publishers, with a further fifteen percent unattributable to any participant in the chain at all, an amount the study’s authors labelled the unknown delta. Subsequent research by the ANA, the WFA, R3 and Ebiquity confirmed the pattern was not a one-off: across numerous studies over several years, only forty to sixty percent of digital dollars invested programmatically were found to reach publishers as genuine working media.

It would be wrong to present this as a static, unimproving picture, and a piece arguing for evidential rigour should not do that. The 2023 follow-up study found real progress: the match rate between what left a demand-side platform and what a supply-side platform could account for rose from twelve percent to fifty-eight percent, and unattributable spend fell from fifteen percent to three. The industry, when it looked hard at the problem, was capable of fixing a meaningful share of it. The relevant question for a marketing leader is not whether this specific technical opacity has improved. It has, in places. The question is what kind of organisational structure produced the opacity in the first place, and whether that structure has itself changed or merely moved.

The answer is that it has moved. Where budget scrutiny once depended on structural friction between disciplines, it now depends on the internal discipline of a single, converged partner whose commercial model rewards continuity of spend over interruption of it. A productised, always-on growth agency is not incentivised to ask whether a channel allocation reflects genuine audience insight, in the way a separate brand planner once might have been. It is incentivised to keep the always-on machine running, because that is what the retainer is built to reward. The convergence did not create the temptation to under-scrutinise spend. It removed the structural contradiction that used to force scrutiny to happen anyway.

The capability gap sitting underneath it

The second half of this problem is not structural. It is a straightforward absence of marketing fundamentals among the people now running national-scale, always-on budgets. The Chartered Institute of Marketing’s Digital Marketing Skills Benchmark, run in partnership with Target Internet across more than seven thousand marketers, found analytics and data proficiency had fallen further than any other core skill, to a score of twenty-nine percent, with almost seven in ten marketers scoring in the lowest two quintiles for their ability to analyse and improve their own work. That is a foundational gap in the single capability a converged, always-on model depends on most: the ability to read what the always-on machine is actually doing with the money.

The more serious finding in the same body of research is not the skills gap itself but its distribution. The study found a marked disparity between marketers’ confidence in their own abilities and their actual measured skill, most pronounced in content, usability, social media and analytics, and a related benchmarking exercise found that marketing directors demonstrated a weaker working knowledge of PPC, SEO, ecommerce and analytics than junior executives with only one to two years of experience. This is not a junior training problem that seniority will eventually solve. It is a seniority-inverted one. The people with the authority to challenge an always-on spend plan are, on average, less equipped to interrogate it than the people executing it, which means the capability gap and the structural gap identified above are not separate problems. They compound each other. A converged agency model with weakened internal friction is being overseen, in many organisations, by client-side leadership least equipped to supply the friction the model no longer provides on its own.

What this is not

It would be convenient, and wrong, to read this as an argument that agency convergence is itself a mistake, or that the answer is a return to siloed, single-discipline agency relationships. The forces driving convergence, privacy-led first-party data requirements, the collapse of channel boundaries, the genuine value of a coherent full-funnel view of a customer, are real and are not going to reverse. It would be equally convenient, and equally wrong, to treat this as a story about bad-faith agencies extracting value from unsuspecting clients. The evidence does not support a story about deliberate opacity manufactured for profit. It supports a story about a structural check quietly disappearing as a side effect of changes made for legitimate reasons, unnoticed because nobody was responsible for noticing it.

That is, in fact, the more uncomfortable version of the argument, not the more comfortable one. A villain is easy to manage. A structural gap that nobody designed and nobody currently owns is not.

What a leader does with this

The response is not a governance checklist bolted onto the existing model, and it is not a demand that agencies un-converge. It is a recognition that when a structural check disappears, someone inside the client organisation has to become the check, deliberately, rather than assuming the market will supply one. That means three things in practice.

The first is treating measurement and supply chain literacy as a leadership capability rather than a specialist one. The CIM data is specific on this point: the gap is worst, proportionally, at the most senior level, which means the fix cannot be delegated downward. A marketing leader who cannot independently interrogate a media plan is not exercising oversight over their converged agency partner. They are trusting it, and trust is not a governance model.

The second is building the ring-fenced habit of independent audit into the always-on model itself, not as an annual exception but as an ongoing discipline, in the way the ISBA taskforce’s Programmatic Financial Audit Toolkit was designed to make routine rather than occasional. The 2020 to 2022 improvement in match rates did not happen because advertisers trusted the system more. It happened because a specific set of advertisers demanded to see inside it and kept demanding it.

The third, and the hardest, is accepting that convergence has moved the location of a cost the industry has not fully priced. When brand, digital and direct sat apart, inefficiency was visible and annoying. Now it is smoother, better presented, and harder to see, which is a worse trade than it looks. An organisation that mistakes smoothness for soundness has not solved the extraction problem the old model had. It has simply made it quieter.

Sources referenced

ISBA / PwC, Programmatic Supply Chain Transparency Study, 2020 and 2023 follow-up.

ANA, WFA, ISBA, R3, Ebiquity and others, cited via ANA Programmatic Media Transparency RFP, April 2021 (40–60% working media range across multiple studies).

Chartered Institute of Marketing / Target Internet, Digital Marketing Skills Benchmark.