An interview with Federico Benincasa, Chief Product Officer at Pubstack
Programmatic advertising is no longer a supply chain. It is a maze.
An impression can bounce through a dozen optimisation layers, throttling rules and curation systems before a bid even appears. Valuable supply gets lost. Publishers lose visibility. And revenue depends on decisions made by systems they do not control.
But there is a turning point.
For the first time, publishers can use AI-driven traffic shaping and explicit inventory mapping to understand the true behaviour of the market and influence it. Not by adding more SSPs or tweaking waterfall rules, but by controlling how their inventory is routed, interpreted and valued.
We asked Federico Benincasa, Chief Product Officer at Pubstack, why he believes these two strategies will define the next generation of publisher monetisation, and why most publishers are still underestimating what's now possible.
Over the past years, the supply chain has become tangled. Instead of a straightforward buyer–seller dynamic, we now have a deeply interconnected network where impressions can circulate through a dozen platforms before a bid is even placed.
A second auction layer (via Prebid and server-side bidding) has introduced circular pathways and overlapping optimisation strategies. SSPs sometimes behave like buyers, DSPs apply their own throttling and valuation logic and both sides operate with partial information.
The result is a system where publishers often struggle to understand why certain impressions succeed or fail. Key pockets of high-quality inventory become algorithmically invisible, lost among redundant filters, missing signals and inconsistent metadata.
So if the problem is not too many SSPs or bad waterfall logic, what is it?
Most programmatic decisions happen outside the publisher's environment. DSPs and SSPs apply filters, valuation rules, throttling and quality checks, but publishers only see the outcome, not the underlying logic.
AI-driven traffic shaping can change that.
By analysing SSP response patterns at scale, AI can detect what buyers actually value and what they are systematically ignoring. This allows publishers to prioritise impressions with real market demand, while reducing the noise created by impressions with no realistic chance of winning.
For this to work, AI needs visibility at scale. Most intermediaries only see the small portion of inventory where they win, often around five per cent or less. That is not enough data for meaningful learning.
Publishers, however, see all auctions, all responses and all impressions, making them uniquely positioned to train AI models capable of detecting genuine buying behaviour.
Partially yes, it’s a valuable side effect. Repeated rejection is often a pattern linked to quality or safety issues.
When AI flags these patterns early, it reduces downstream friction, protects revenue and improves the overall stability of the supply chain. Over time, better-quality supply encourages buyers to reduce their throttling, creating a positive feedback loop that improves yield.
Because most meaningful publisher signals do not survive the programmatic journey.
Context, placement, engagement metrics, scroll depth, session quality, all of these are often stripped away or diluted as impressions pass through intermediary systems.
Without clear and interpretable signals, buyers default to generic heuristics.
High-value inventory gets misclassified, ignored or undervalued.
Inventory mapping reintroduces clarity.
By structuring supply around business-meaningful KPIs — whether contextual, performance-based or predictive — publishers create a stable taxonomy that buyers and their algorithms can understand consistently.
Predictive AI can generate opportunity-level signals, such as expected viewability or expected engagement, providing buyers with a forward-looking understanding of inventory quality.
This allows publishers to surface value that previously remained hidden and helps buyers optimise their models using reliable, consistent signals rather than noisy or incomplete data.
Because they solve different parts of the problem.
Traffic shaping interprets how the market values impressions.
It identifies which inventory to prioritise and which to downplay.
Signal-rich mapping ensures buyers can algorithmically understand the supply they receive.
It clarifies the meaning of impressions and aligns them with the metrics buyers optimise for.
Predictive AI amplifies both strategies by generating new, actionable signals.
Traffic shaping alone reduces noise but cannot reveal hidden opportunities.
Mapping alone improves clarity but cannot prioritise what truly matters.
Together, they give publishers control over both the quality and the perception of their supply.
Before AI, this combination was impossible. The scale, granularity and speed required are beyond manual or rule-based approaches.
Legacy ad servers and traditional supply platforms are not built for this. To implement AI-driven traffic shaping and explicit inventory mapping effectively, publishers need a platform that can:
This is not a small optimisation. It represents a foundational shift in how publishers can route, describe and monetise their inventory.
Programmatic complexity is not going away; it is increasing.
But with the right AI-driven tools, publishers can finally turn that complexity into a competitive advantage. They can control how their inventory is interpreted, how it is prioritised and how it competes across the chain.
For the first time in years, publishers can genuinely take back control.
The programmatic ecosystem is not becoming simpler.
Auction paths will multiply, signal fragmentation will grow, and optimisation layers will become increasingly opaque. But as Federico explains, this complexity is not a threat; it is leverage.
Publishers who can decode market behaviour and structure their supply around business-centric signals will finally control how their inventory is valued and monetised. The challenge is that legacy ad servers and traditional supply platforms were never built for this level of strategic control.
This is exactly why we built our Spark solution the way we did.
With the right platform, publishers can:
Spark is Pubstack's AI driven monetisation platform that brings real-time AI-driven traffic shaping and advanced inventory mapping capabilities. It provides the visibility intermediaries lack and the predictive power needed to activate both strategies at scale.
The complexity isn't going away. But with the right infrastructure, publishers can finally turn it into a competitive advantage.
.png)
.png)
.png)
