Agentic AI. Outcomes prediction. Supply path optimization. Privacy-first audience targeting. We have the building blocks for the next generation of digital advertising, but if we stack them on top of outdated attitudes towards data and transparency, they’ll all fall down. If the buy-side and sell-side want to seize the opportunity of new technology, they need to be on the same team.
At a minimum, advertisers need to share their conversion data for an advertising platform’s models to be able to optimize future campaigns. In contrast, publishers need to share their audience data so that advertisers can target against it. If both sides hold their cards close to their chest, only walled gardens will come out on top, leaving the future of the industry at the mercy of a handful of Big Tech titans.
If you’ve been in digital advertising for any time at all, you’ve heard these pleas before. Collaboration and interoperability have been headline slides at every trade show presentation for years now. And for good reason: being open with data that is useful for everyone (while protecting data that is confidential or subject to privacy regulations) yields better results for all parties.
Despite constant reminders, many have dragged their heels, and there’s no one factor to blame. Some fear being transparent for what it will reveal to the competition, others don’t have enough work to get their audience or reporting data in order and accessible. Meanwhile, AI technologies that could activate this universe of data either gather dust or are being utilized by companies that are soon going to race ahead in the market.
These technologies are already delivering many of the long-discussed digital advertising goals: AI models are optimizing for efficiency and carbon emissions; low-quality and fraudulent inventory is being filtered out; audiences are being assembled and segmented across domains and platforms using entirely non-identifiable data. The more oversight the models have, the more they will improve the supply chain, attracting more spend and lifting all boats.
AI agents are promising to be the next big growth driver, but their ability to deliver this growth in the long term hinges entirely on interconnectivity. Currently, AI agents are platform-specific, effectively an automation layer with an LLM-powered natural language interface that can perform a wide variety of tasks. Useful, but its growth will hit a wall before long if it remains confined to individual platforms.
For an AI agent to be able to take full advantage of what the supply chain has to offer, each component part between A and B needs to provide AI-friendly API access and make all relevant data readable by the agent, and all this needs to happen in real time. A tall order, but feasible on a technical level. The question is whether ad tech’s various platforms are willing to work together to achieve it.
Would connectivity lead to cooperation or consolidation?
A valid fear of a more interconnected ecosystem is that cooperation can easily tip over into consolidation, especially with the trend of more DSPs offering direct access to supply and more SSPs building buying platforms. If both ends can serve the other, it’s easy to see a future where all platforms consolidate into larger and larger end-to-end platforms. However, such a market would remain competitive, due to reputational rather than technical hurdles.
For an end-to-end platform to be successful, it needs to be trusted by both sides of the supply chain, who — though dependent on one another — have oppositional interests. Buyers want the lowest cost, and sellers want the highest price. If either side suspects the other is getting preferential treatment, an end-to-end platform loses its essential claim to neutrality.
There will also be fears that preferential treatment might extend to individual clients. Of course, bigger fish are always going to have an advantage in the market (if you can buy more or supply more, you have a leg up over the competition), but the effectiveness of spend or the value of inventory must be determined on an even playing field. All transactions must be treated equally.
Without a DSP or SSP on the other end to keep both sides honest, an end-to-end platform must be auditable by third parties via log-level data, without exposing confidential campaign specifics. As end-to-end platforms scale and start to resemble walled gardens, they inherit the same concerns around their ability to mark their own homework.
Being able to ease these concerns will be what differentiates one end-to-end platform from another and keep the market competitive. These concerns are also why DSPs and SSPs will continue to have a place in the market; being able to say to one end of the supply chain, “We exclusively represent your interests,” is a strong sales message.
Returning to AI agents, if the technology is capable of what has been proposed (which is by no means confirmed), then the agents could effectively become end-to-end platforms themselves, representing clients on an individual level. With full oversight over the supply chain, it could assemble the necessary mix of buy-side, sell-side, and intermediary technologies to achieve the best results at a campaign-specific level.
Very much hypothetical at this time, but it’s worth imagining radically different technological futures. It reminds us that everything we currently know of our supply chain and the old dividing lines of its market were also once hypothetical. Everything can change and has changed before.
Right now, change is coming fast, making it difficult to predict what will come out the other end. SSPs, DSPs, intermediaries, buyers and sellers, devices and platforms; we all share the same market, and it’s in all our interests to work together to build a market that can endure, innovate, and grow.
