Eight years ago, the marketing world was captivated by a singular, seductive narrative: the "Programmatic Buying Foundations" curriculum. As an instructor of this course, my pitch was straightforward—data and technology would unlock a new frontier of hyper-relevant, highly effective, and easily measurable advertising at scale.
Today, as we navigate the frantic integration of Artificial Intelligence into our marketing stacks, that old pitch feels like a ghost from a past life. When you read Kevin Indig’s recent Growth Memo regarding the "hidden hours" buried within AI workflows, you aren’t seeing the birth of a new challenge; you are seeing the dark side of that old programmatic promise resurface. The promise of efficiency was broken then, and it is broken now.
This isn’t merely a case of history repeating itself; it is a warning about where the current obsession with AI-driven marketing is destined to crack. We are falling into the same accounting errors, ignoring the same labor shifts, and blinded by the same false promises of "automation."
The Illusion of Efficiency: A Historical Context
To understand why the current AI craze feels so familiar, we must revisit the programmatic era. My course taught a five-step workflow designed to streamline media buying. The curriculum was bolstered by glowing case studies from industry titans like Mondelez, Campbell’s, and Ford India. The framing was consistent: efficiency first. The underlying, often unspoken, assumption was that if you automated the buying, effectiveness and measurability would naturally follow as a byproduct.
However, those who wax nostalgic about the "golden era" of programmatic conveniently leave out the fine print. That same course required entire modules dedicated to ad fraud, brand safety, and the impending nightmares of GDPR compliance.
While we told advertisers that automation would sharpen their targeting, we were simultaneously forced to teach them how to identify fraudulent inventory and navigate the lack of transparency that plagued the industry. We sold a tool that promised better measurement, yet we had to provide an entire unit on why, due to privacy regulations and cross-device tracking issues, measurement was actually becoming less reliable. We were solving for efficiency, but we were creating a massive new debt of labor: the labor of monitoring, fixing, and auditing.
Supporting Data: The Productivity Paradox
The argument that AI has revolutionized marketing work is increasingly under fire. The consensus is shifting toward a realization that AI hasn’t eliminated labor—it has merely shifted it.
The Developer Disconnect
A study by METR provides the most damning evidence of this disconnect. Researchers assigned 16 experienced developers 246 real-world tasks, split between AI-assisted and manual workflows. The developers expected that AI would increase their speed by approximately 25%. In reality, they were 20% slower with AI assistance. Most tellingly, even after the data proved they were less efficient, the developers still believed the AI had sped them up. This psychological bias is the bedrock of the current AI-adoption crisis.
The Rise of "Workslop"
Marketing is currently suffering from its own version of this gap. A survey conducted by BetterUp Labs and Stanford University, covering over 1,000 workers, identified the phenomenon of "workslop"—content that appears polished and finished but lacks the substance or accuracy required for real-world application. The study found that fixing these AI-generated errors takes an average of two hours per instance. At a large enterprise level, this equates to an annual productivity drain of over $9 million.
The Resource Drain
Research from Workday reinforces this, indicating that for every 10 hours saved by AI, roughly four are spent "re-doing" or auditing the AI’s output. Upwork’s polling of 2,500 leaders and workers further quantifies this, revealing that the "reclaimed time" isn’t being used for innovation or deep work. Instead, it is consumed by the administrative burden of checking AI outputs, learning ever-changing toolsets, and managing a higher volume of tasks than before.
The "Invisible Headcount" Problem
HubSpot’s latest State of AI report highlights the scale of the issue: the vast majority of marketing leaders report that their teams are already utilizing AI, with a significant number opting to build proprietary internal tools rather than purchasing third-party solutions.
This is where the trap snaps shut. In-house building is not a one-time project; it is a permanent, largely invisible maintenance job. When a team builds its own AI-driven content engine or predictive model, that tool requires constant oversight. It becomes an "invisible headcount." If the primary developer or "owner" of that workflow takes a vacation or moves to a different project, the entire system often grinds to a halt, reverting to manual processes that the team has forgotten how to execute efficiently.
We are repeating the programmatic error: we have converted manual media buying labor into fraud monitoring and compliance labor. Today, we are converting content creation labor into prompting, debugging, and babysitting labor. This work doesn’t show up on a project plan as "real work" until a stakeholder asks why, despite the influx of AI tools, content production hasn’t actually driven the expected growth.
Implications for Marketing Leadership
The failure here is not that AI is "oversold" as a concept. The failure is an accounting error. Efficiency is being measured on the wrong side of the ledger. We celebrate the hours saved on a visible task while ignoring the hours spent setting up, testing, and babysitting the system that saved them.
If your marketing organization is chasing AI efficiency without acknowledging the "hidden hours," you are not innovating. You are merely operating under a new, more expensive administrative burden.
Strategic Recommendations
To avoid the pitfalls of the programmatic era, marketing leaders must adopt a more rigorous framework:
- Impose Accountability: Every internal AI tool must have a designated owner and an expiration date. If a tool cannot justify its existence through measurable ROI—excluding the "time saved" that is actually being spent on maintenance—it should be shut down. We need industry standards for AI accountability, similar to the
ads.txtmovement that finally brought transparency to programmatic. - Redefine Your Metrics: Stop asking, "Did this AI workflow save us time?" Start asking, "How many hours this month were spent building, fixing, or maintaining AI tools instead of performing the core work?" As the METR study proves, your team’s subjective sense of speed is likely inaccurate.
- Protect High-Value Fundamentals: AI is efficient at churning out high-volume, low-depth content. It is terrible at the "slow" work: deep investigative pieces, digital PR, and original research. This is the work that actually builds brand authority and gets cited in the future. Ring-fence a fixed percentage of your team’s time for this high-value, long-term work before the AI-tooling, by default, consumes your entire calendar.
Conclusion: The Old Syllabus, New Cover
I taught the efficiency pitch once, and I believed in it. But I also taught the reality of why it failed. If your marketing team is charging headlong into AI integration without auditing where the "saved" time is going, you aren’t at the forefront of a revolution. You are simply re-reading my old syllabus with a new, shinier cover.
True efficiency in marketing has never been about doing things faster; it has always been about doing the right things well. If we continue to ignore the hidden labor costs of our technology, we are destined to repeat the history of the programmatic era: promising the world while silently drowning in the overhead of our own making.
