Nigel Gilbert has spent more than two decades at the forefront of digital advertising. In the latest of NDA’s interview series, Decoding the Future of Advertising and Marketing, he explains that the “build, buy or partner” debate is back. But he thinks we’re asking the wrong questions again…
AI and agentic technologies are radically changing what businesses can realistically bring in-house, from customer intelligence and measurement to content automation and media activation. But just because technology makes something possible doesn’t necessarily make it commercially sensible. The question for leaders isn’t simply whether to build, buy or partner. It’s understanding which capabilities are worth owning, where external expertise creates more value, and how to retain control over the decisions that matter.
We’ve been here before
“Build, buy or partner?” also isn’t a new question in media & marketing.. This is really a second run at it…or maybe the third. The early 2010s took the first pass, and the primary use case for “build” back then was in-housing: contract the tech, and bring media buying in-house and run it directly.
Only a small number of businesses had the scale and the specific problem that building could actually solve. And the model had a design flaw. It assumed replacement of the agency rather than augmentation, and that’s what capped how far it could ever go.
Doing it properly meant:
- Data: Owning, integrating and governing the (largely media) data informing decisions.
- Measurement: Connecting activity to meaningful business outcomes, rather than relying on platform-reported performance.
- Strategy: Turning customer and performance intelligence into decisions about audiences, investment, offers and channels.
- Creative: Knowing what content and creative to deploy, for whom, where and when.
- Activation: Maintaining the specialist expertise required to brief, buy, optimise and execute across an increasingly complex channel mix.
- Customer experience: Connecting those decisions across the customer journey rather than optimising individual channels in isolation.
Most brands weren’t equipped to take on all that internally, and much of the work ultimately moved back to agencies.
Why is Build, Buy & Partner Back?
So why is the exact same question back again? Because AI and agentic are changing both what businesses can realistically own and what is worth owning.
These emerging technologies have made it highly addressable, and this time, martech is properly in the consideration set. Customer intelligence, audience segmentation, content automation, and of course omni-channel activation, can all now be connected by those who are prepared to understand how.
I have left measurement and attribution out of that list, because I believe those to be the real prize of transformation, not a feature of it.
Get the recipe right this time and the prize is real: tangible insights, and the actual internal and external agility to do something with them. In-house teams that are elevated rather than overwhelmed, and partners you can actually collaborate with rather than simply pay. Not to mention the ability to build out unique demand, not volume that you rent from platforms.
The danger of owning the technology but not the capability
There is a scenario in which businesses repeat the mistakes of the first wave, only this time with more sophisticated software and they end up with the worst version of both worlds:
It leads to a situation where a business has bought the technology, outsourced the expertise, and kept only the accountability. Here, no-one can explain how its own system works, can’t challenge a recommendation, and can’t change direction when conditions shift. It’s a governance problem that, historically (and hopefully not again) folks have mistaken for a capability one.
Should orchestration even really be the word we reach for here? It’s not wrong exactly. Scott Brinker’s 2026 State of Martech count puts the category at 15,505 tools, and the real fight among vendors has moved from who unifies the data to who controls the arbitration layer that decides the next action for a customer, omnichannel of course.
It’s why Salesforce, Hightouch and Databricks are all racing to own the “agentic CDP” category this year, and Forrester puts 83% of B2C marketing decision-makers as actively implementing agentic AI right now. So orchestration is pretty important as it’s a mechanism that can create a singular view of all the metrics that matter.
Governance is then the decision about who’s pulling which lever, and why.
What should businesses actually own?
So which parts are actually worth owning? Honestly, it depends: on category, on maturity, on how much a business is genuinely willing to build and sustain. But the logical end state looks broadly the same: data sovereignty and security, so the truth about your customers and your performance sits inside your own environment rather than someone else’s.
A genuinely holistic view of the customer and of performance, not five partial versions scattered across platforms and partners. That’s what lets you own the calls that actually shape outcomes: budget allocation, offer mix, and orchestrating an experience, rather than just assembling a media plan.
Activation is a different matter. Once that strategy is genuinely yours, expert activation across channels, according to it, is exactly the kind of work worth buying or partnering for.
A model like TAU is a useful illustration of what that split can look like in practice. Their intelligence layer, built on top of the clients’ stack, pulls 30–40 live data sources into one governed view instead of a stack of disconnected (or outsourced) dashboards. The decisioning logic stays trained on the client’s own planning logic, with source, freshness and ownership tracked throughout.
All the important adtech and media platforms show that owning the brain in fact means connecting the existing activation pathways, and adding more as required. Clients report scaling budgets by 40% with the same team, purely because the system gives them enough confidence in what’s actually working to commit further, often still very much working with partners in various areas.
In many ways this is where the payoff compounds. Once your insights are unified and genuinely your own, the governance and control that creates brings efficiency and performance gains in its own right, but it also buys something that gets underrated: agility.
There are lots of new, innovative companies worth watching right now, agentic platforms like Apostra, emerging measurement standards like Origin and Lantern, decisioning specialists like TMX Group and plenty more arriving behind them. If you’ve got the right stack underneath you, you can actually test and learn with each of them directly, rather than reading about these innovations and wondering if they could have an impact.
And because there will always be more of these, setting things up the right way now matters more than ever: not to chase any single launch, but to keep enriching a singular source of intelligence that your entire business can begin to benefit from.
So, Build, Buy or Partner?
I don’t think the answer is choosing one. The mistake during the first wave of in-housing was assuming that greater ownership of execution automatically translated into greater control. AI gives businesses another opportunity to reconsider that relationship.
If you take a capability in-house, you need to truly own the parts that count, not just a technology stack sitting on top of them.
Half-owning this is worse than not owning it at all, because it creates the illusion of control, along with a long list of direct contracts to manage for no real added value. If you partner instead, the test is whether you can see where, how, and how much is being deployed on your behalf, in enough detail to challenge it. A vendor or service partner relationship without that visibility is really just a volume-based service.
And whichever approach you take, retain enough intelligence and governance internally to understand what’s happening, challenge the recommendations being made and change direction when circumstances demand it.
