Key takeaways
- AI strategy in media starts with data ownership. If the company lacks useful, contextual data, AI will not create durable advantage.
- Do not outsource creativity and strategic thinking. AI should support higher-value work, not replace the thinking that differentiates the business.
- Prioritise use cases with measurable financial impact. Forecasting, pricing, yield optimisation, and decision support offer clearer ROI than generic automation.
- Adoption is not value. Tool usage should not be confused with business outcomes, especially when hidden operational costs are rising.
- Governance should enable experimentation. The best AI governance protects ethics, security, and compliance without shutting down innovation.
- Build the business case like a capital investment. Define the baseline, dependencies, and expected operational change before buying the tool.
The question at the center of this discussion panel- whether media companies must become tech companies in the AI era – is no longer theoretical. For publishers, agencies and media groups, AI is already reshaping audience targeting, ad operations, forecasting, creative workflows and governance. The real issue is not whether to adopt AI, but which capabilities must remain inside the business if it wants to protect margin, retain strategic control and stay relevant as a partner to clients.
In a discussion panel hosted by Adam Gaca, guests Dane Buchanan, Global Chief Data & Analytics Officer at M+C Saatchi Performance, Miryem Salah, Director Digital Data & Transformation, MD IT Hubs at VodafoneThree and Tim Lum, Global Chief Intelligence Officer at VML argued that the answer starts with data, but does not end there.
Their credibility comes from working directly in media, analytics, marketing transformation and performance operations, where AI is not an abstract trend but an operational constraint.
The conversation cuts through the hype: AI can drive measurable value, but only if media leaders define the business problem first, build the right data architecture and decide deliberately what to own, what to automate and what never to outsource.
AI in media strategy starts with data ownership and contextual depth
The conversation makes one point clear: AI is only as useful as the data environment behind it. For media companies, that means the first strategic question is not “What model should we use?” but “Do we own enough of our data universe to make the model meaningful?”
Dane Buchanan put it bluntly: “one AI that’s not necessarily well without data, uh ensuring that either you own your data universe or you’re able to partner with it to make sure that you have enough to train the model”. That statement captures the central risk for media and advertising businesses. Generic AI tools can generate output, but without proprietary data and context they cannot create durable advantage – they become interchangeable software.
This matters because media companies operate in environments where context drives value: audience signals, campaign history, pricing behaviour, inventory patterns, content performance and client-specific workflows. If those inputs are fragmented across systems or controlled by third parties, AI will not produce a strategic edge – it will simply accelerate mediocre processes.
The stronger argument in this conversation is that data ownership is not just a technical concern; it is a commercial one. A media company that cannot connect its data to decision-making is vulnerable to being disintermediated by platforms, vendors, or better-prepared competitors.
Read more about AI in media:
Why “owning the data” is not the same as storing more data?
Ownership here means more than retention. It includes:
- access to the right historical and real-time signals,
- clean architecture,
- enough contextual metadata to train or guide models,
- and a governance model that makes the data usable, not just archived.
That distinction explains why several speakers warned against treating AI as a bolt-on. If the data layer is weak, AI sits on top of noise. If the data layer is strong, AI can support forecasting, planning and decision-making in a way that compounds value over time.
Get recommendations on how AI can be applied within your organisation.
Explore data-based opportunities to gain a competitive advantage.
Media companies should not outsource their creative and strategic thinking
A second major theme in the discussion panel is that media businesses should resist the temptation to outsource the parts of the work that define their identity. Technology can speed up production and automate repetitive tasks, but the panel repeatedly returned to one concern: if the company gives away creativity, it gives away differentiation.
Dane Buchanan said the part not to outsource is “as much as possible that element of creativity. Creativity comes in a number of different ways. It comes in creating contents, but it also comes in the mindsets, the thinking mindset around what do I do with this?” That insight is especially important for media businesses that are under pressure to adopt AI everywhere at once.
The logic is simple: if AI is used only to produce more content faster, the result is often more volume, not more value. But if it is used to support better editorial decisions, sharper audience segmentation, or faster prototyping of ideas, it becomes an enabler of higher-level work. It pushes teams to ask better questions:
- What problem are we solving?
- What outcome are we improving?
- What part of this workflow truly benefits from automation?
Miryem Salah and Tim Lum reinforced this point by stressing the need to keep strategic control over the systems that shape client value. In performance media especially, ad operations and analytics are not commodity functions if they are connected to measurable insight and client retention. The business must preserve the measurement layer and the advisory layer, because that is where trust and margin are built.
“If you don't have that, that is you know your bread-and-butter within the organisation, and you don't have the data, AI is just another piece of software.”
That sentence captures the strategic danger of shallow AI adoption. A media company can buy a tool and still lose relevance if the tool does not reinforce a distinctive operating model.
The hidden commercial cost of “AI everywhere”
The panel also warned that tool adoption can create a false sense of progress. When AI is spread across the organisation without a clear operating model, the business can end up with:
- duplicated workflows,
- spiralling costs,
- fragmented governance,
- and teams using tools in ways leadership does not understand.
That is not transformation, but an operational drift.
AI use cases in media: where the ROI is real and where it is overstated
The speakers were notably pragmatic about ROI – they did not claim that every AI use case delivers the same impact, or that adoption alone is a success metric. In fact, Tim Lum challenged the obsession with adoption rates, noting that usage by itself does not equal value.
The strongest examples in the conversation were in forecasting, pricing, yield optimisation, and decision support. Dane Buchanan referenced a machine-learning forecasting use case that improved inventory predictions by around 20%. He also described a more dramatic example from his past work, where AI helped improve yield on a single airline route by about 40% by identifying anomalies analysts would have missed.
That example matters for media leaders because it shows how AI creates value in decision-heavy environments. When people have to review thousands of decisions or signals, the model is not replacing expertise; it is narrowing the field to the highest-value actions. As Buchanan explained, the result was reducing a vast workload to “the top twenty you really need to look at” and focusing on opportunity costs and revenue opportunities.
For media companies, that translates into a few practical use cases:
- pricing and yield optimisation,
- campaign forecasting,
- inventory planning,
- anomaly detection,
- and prioritisation of client or content actions.
The panel’s position was not that every workflow should be automated. Rather, the business should focus AI where decisions are repeated, data-rich, and financially material – that is where efficiency becomes real and measurable.
£1B+ in bookings for the UK’s largest independent broadcaster with a new ad management platform
Why AI business cases need CFO-level scrutiny
The panel also highlighted a critical internal challenge: AI projects often fail not because the model is poor, but because the business case is weak or incompletely scoped. One speaker described presenting a case with “twenty times ROI”, only to face skepticism because the organisation had not yet built the trust or the framework to believe such returns were realistic.
That is a valuable reminder for any media executive – AI investment should be assessed like any other strategic capital allocation:
- What measurable problem is being solved?
- What baseline are we improving from?
- What operational dependencies exist?
- What data architecture changes are required?
- What hidden costs will show up later?
Without those answers, AI becomes a procurement exercise instead of a business strategy.
Governance in the AI era should enable innovation, not block it
The most nuanced part of the discussion was around AI governance – the panel did not argue for weak controls. They argued for enabling governance instead of policing for the sake of control. That distinction matters in media, where experimentation is essential but so is brand safety, ethical use, and operational reliability.
Tim Lum described governance as something that must cover the entire ecosystem: data access, implementation, who can use the system, what the AI is allowed to see, and how decisions are structured. In other words, governance is not a lock on the door. It is the design of the room itself.
“I tend to say that the best governance is the stuff that is more the enabling side of that than what does it necessarily.”
That approach is especially relevant in media because the risk profile is broad:
- black-box decision models,
- compliance issues,
- inconsistent regional maturity,
- uncontrolled experimentation,
- and teams building shadow AI solutions outside central oversight.
The panel noted that governance maturity differs by region and by digital transformation maturity. In more digitally advanced markets, teams are used to test-and-learn operating models and can connect experimentation to ROI. In less mature environments, “adoption” itself can become the metric, which is a mistake. Usage without value is not governance; it is noise.
For leaders, the message is clear: governance must protect the business while still letting teams build. The objective is not to suppress creativity but to keep experiments inside a secure and commercially coherent framework.
Conclusion
The discussion’s core argument is not that media companies must become software companies in a literal sense. It is that they must become deliberate about what they own, what they automate, and what they allow outsiders to mediate. In the AI era, competitive advantage will not come from simply deploying tools faster than everyone else. It will come from combining proprietary data, strong operational design, and a clear view of where human judgment still matters most.
For media leaders, that means the real survival strategy is selective transformation: own the layers that shape insight, client value, and strategic control; automate the rest with discipline; and build governance that lets the business move without losing itself. The companies that do this well will not just use AI. They will use it to defend margin, improve decisions, and stay structurally relevant in a market where software is becoming table stakes.
