1. Programmatic was built around automation, yet many teams still spend significant time manually managing and optimizing performance. Where is the industry still getting automation wrong?
The mistake is assuming that automation on its own equals optimization. It doesn’t. You can automate a bad decision just as easily as a good one, and you can do it much faster.
Programmatic teams still spend far too much time pulling reports, adjusting settings and reacting to individual performance changes because a lot of the technology was built to provide information rather than actually act on it. The opportunity now is to move from reactive optimization to automation that is tied to a clearly defined strategy.
That thinking is behind Limelight’s Adaptive Rules Center (ARC). Rather than asking teams to manually monitor every performance change, ARC lets them define their own KPIs, rules and trading logic, and then automate actions across supply and demand when those conditions are met.
The operator still determines the objective and the guardrails; the technology handles the repetitive execution at a scale and speed a human team simply can’t match. Good automation takes people out of repetitive tasks. It shouldn’t take them out of decision-making.
2. AI is now everywhere in adtech. At what point does “AI-powered” become table stakes rather than a true differentiator, and what should companies actually be looking for?
We’re already there. If everybody’s differentiator is that they use AI, then AI isn’t really a differentiator anymore.
That doesn’t make the technology any less important. AI is extraordinarily powerful, particularly in a market like programmatic where you are processing huge amounts of data and making decisions in real time. But the interesting question is no longer, “Do you have AI?” It’s, “What are you enabling me to do with it?”
That is why Limelight has focused less on treating AI as a black-box product and more on giving partners control over how intelligence is applied. Through ARC, partners can build custom logic around their own goals, KPIs and trading strategies, while the broader Limelight platform provides the data and flexibility needed to inform those decisions.
Companies should ask whether a platform lets them understand why decisions are being made, use their own data and measurement partners, and change course when their commercial strategy changes. AI should be an engine for your strategy, not the strategy itself.
3. How can ad networks and publisher networks use AI and automation to make faster decisions while still maintaining control over their strategy, data and outcomes?
The key is separating strategy from execution. A publisher or ad network should decide what good looks like: which partners it wants to prioritize, what performance thresholds matter, how it values different inventory, where it wants to experiment and how much risk it is willing to take. Automation can then monitor those signals continuously and execute against those parameters much faster than a person could.
ARC was designed around that model. Limelight partners can create rules based on the dimensions and KPIs that matter to their individual business and automatically take action when supply or demand over- or under-performs. The platform also gives partners margin control, deep analytics and the ability to integrate third-party measurement, so automation does not require handing over ownership of the operating model.
Control also depends on data access. Businesses should be able to see the underlying performance data, understand what the system is responding to and use that information elsewhere in their own technology stack.
The best automation therefore isn’t about surrendering control. It actually gives teams more control because it lets them encode their own commercial logic into the technology and apply it consistently at scale.
4. Where should humans remain in the loop as programmatic optimization becomes increasingly automated? Are there decisions that shouldn’t be handed over entirely to a machine?
Humans should remain responsible for intent. A machine is exceptionally good at looking at thousands or millions of data points, identifying patterns and taking action quickly. What it doesn’t inherently understand is why a business has chosen a particular strategy, the value of a commercial relationship, the nuances of a publisher’s audience or the broader context behind a decision. Those are human judgments.
That balance is something we’ve deliberately built into Limelight. ARC, for example, supports different levels of automation, including Recommendation and Autopilot modes, so teams can determine how much authority they want the technology to have depending on the use case. The point is not to automate everything simply because you can.
There is also an expertise element that sometimes gets overlooked in adtech. Limelight pairs its technology with dedicated Client Success support because even sophisticated automation benefits from people who understand the market, can identify opportunities and help partners apply the technology effectively.
The strongest model is technology handling scale and repetition, while people remain responsible for strategy, context and oversight.
5. Transparency has been a persistent challenge in programmatic. What level of visibility and data access should publishers and ad networks expect from their technology partners today?
Transparency has to mean more than giving someone a dashboard. Publishers and ad networks should be able to understand what is happening across their trading environment: where revenue is coming from, which demand is performing, which supply paths are creating value, where efficiency is being lost and how different decisions are affecting profitability.
That is an area we’ve prioritized within Limelight’s platform. Partners have access to deep analytics, custom reporting and alerts, as well as API access that allows them to bring their monetization data into their own environment. The platform also supports third-party measurement integrations rather than requiring partners to rely solely on a closed set of tools.
More importantly, the data needs to be actionable. Giving someone thousands of data points without the ability to interrogate them or act on them is technically transparent, but it isn’t particularly useful.
Ultimately, a useful test for any technology partner is simple: does the relationship increase your ability to understand, decide and act, or does it make you more dependent on someone else’s interpretation of your business?
6. With so much AI and automation marketing in the industry, how should companies distinguish between technology that creates tangible business value and technology that’s simply adding another layer of complexity?
Start with the outcome rather than the label. If a vendor says its technology is AI-powered, the next question should be: what changed because of it? Did revenue increase? Did fill rate or eCPM improve? Did auction success improve? Did the team reduce the amount of manual work required to manage the same level of business? Did it identify opportunities that would otherwise have been missed?
Those are measurable outcomes, and that’s how we evaluate automation at Limelight. In one ARC deployment, two account-wide automation rules contributed to a 289% increase in revenue and a 10x improvement in fill rate. In another use case, ARC automatically amplified high-performing demand within existing QPS limits, resulting in a 100% revenue increase and a 70% improvement in efficiency.
The point isn’t that one rule or one piece of automation will produce the same result everywhere. Every trading environment is different. The point is that technology should be accountable to a baseline and a business objective.
If you can’t clearly explain what the automation changed, why it changed it and what value that created, then there is a good chance you have simply added another layer to the stack.
7. As core programmatic infrastructure becomes more accessible, how should companies think about the decision to build technology themselves versus partnering with a white-label infrastructure provider?
This decision has changed considerably over the last decade.
When we started Limelight, we initially went looking for technology that we could use ourselves. At the time, the barriers were significant, high setup costs, large minimum commitments and an ecosystem that was much harder for a new entrant to access. Ultimately, we built the technology.
Today, much of the foundational infrastructure has become far more accessible. That means companies should be very clear about what they are actually gaining by building everything themselves.
If owning and developing the underlying technology is genuinely central to your differentiation, and you have the engineering resources, time and capital to maintain it continuously, building may make sense. But building also means maintaining integrations, infrastructure, compliance, product development and specialist headcount over the long term.
A white-label model can provide another route: use established infrastructure while retaining your own brand, commercial relationships, pricing strategy and operating logic.
The most important thing is not simply “build versus buy.” It is how much control you retain. Companies should assess who owns the customer relationship and data, whether the technology provider competes with them, how flexible the platform is, what the economics look like and what level of expertise sits behind the technology.
You shouldn’t have to build every pipe yourself in order to own your business.
