How to bring buyers closer to the impression
BY HAZEL BROADLEY, BEELER.TECH
On a dreary, wet morning, a reader opens an article while standing just inside a coffee shop doorway. The weather has changed their plans. Their phone is in their hand, the page is on screen, the content is fresh, the location is obvious, and the opportunity for an ad presents itself.
Then the ad market gets involved, and much of that richness is compressed into a one standard transactional object – a bid request.
That has been programmatic’s quiet bargain for years.
A consumer opens a page, starts a video, launches an app, or streams a show, and a tiny window opens for the market to decide what that opportunity is worth. The publisher knows a lot about that moment, but most of it gets flattened into a standard signal before anyone can act on it. And the things that made that impression worth buying – the context, the timing, the intent – get harder to see by the time the request reaches demand.
Sell-side story
Until now, the industry’s mistake has been assuming the best answer comes after the impression has traveled downstream.
That assumption made sense when third-party cookies did the heavy lifting, the sell side had very little time to act, and the buy side held most of the decisioning power. But it holds less now. The open internet is losing easy signals, buyers are buried under avoidable volume, and publishers are watching their own knowledge get repackaged somewhere else.
Sell-side decisioning changes where the market tries to solve the problem. Instead of waiting for the DSP to interpret a partial picture downstream, it moves intelligence upstream, where inventory and publisher signals still have the most shape. The DSP is still important, but when the first decision is better informed, everything downstream benefits.
As Daniel Ahlbert, CEO of Netric, says:
“For publishers, it’s a chance to take a more active role in how inventory is valued, packaged, and sold, instead of watching the most valuable signals get flattened before they can be priced properly. For buyers, it offers fewer, less noisy, and more relevant opportunities without forcing them to rebuild how they buy.”
The first decision has always been made in the same place
The basic programmatic flow is familiar enough that most people barely see it anymore. The publisher’s ad server identifies an available opportunity, the SSP reviews the details, the bid request is sent to DSPs, and the winning ad is served back to the consumer. The whole thing happens so fast that it feels inevitable.
But it’s not. It’s a design choice.
Historically, the most sophisticated decisioning lived on the buy side. DSPs had the budgets, campaign goals, optimization models, and buyer-side definitions of success. SSPs were treated more like infrastructure and handled the compliance, auction mechanics, routing, and supply access before the “real” strategy happened elsewhere.
That split made sense when the sell side had apre-bid window of roughly 10 milliseconds to act which made deep evaluation at speed seem unrealistic. Cloud infrastructure, edge compute, parallel processing, and containerized models have changed the practical limit of what’s possible in that window.
The sell side also has a broader view than any single buyer through a DSP after filtering, throttling, and supply-path preferences have already shaped the view. That vantage point can be pivotal because the best opportunity may never reach the right buyer if the first filter is too blunt. When the earliest layer applies useful logic, the rest of the chain gets a sharper candidate set.
To me, that’s the simplest way to understand sell-side decisioning. It doesn’t ask the industry to throw away the programmatic stack and start again. It asks the industry to stop wasting the richest signals by waiting too long to use them.
Preserve the value that scale can strip
The bid request is a great scaling mechanism in digital advertising, but it was never designed to carry every nuance. Standardization always involves compromise. A publisher understands the full context of an impression, the reader, the placement, the fit. The bid request carries a fraction of it.
The internet has already lost other valuable signals through regulation, platform changes, browser policies, and the decline of easy identifiers have all reduced what buyers can rely on. If the market also fails to use legitimate publisher-side context at key moments, it weakens the very supply it’s trying to monetize.
Sell-side decisioning gives publishers a more controlled way to activate what they know. The most valuable signals can be applied at the source instead of sprayed into the bidstream, with the publisher deciding what’s exposed, under what rules, at what price..
Weather intelligence is a useful example because it shows that ‘signal’ doesn’t have to mean personal identity. Weather conditions influence what people need, where they go and how they buy. When those signals are recognized on the sell side, buyers can respond to real-world moments through a single deal ID, instead of piecing together fragmented data across multiple platforms. In one case, The Weather Company helped a major coffee retailer use weather and product signals to drive more than 100,000 incremental store visits.
The same principle applies to other forms of intelligence. It’s faster to move the decisioning logic to where the data lives than to push every signal downstream. That’s getting more urgent as agentic AI shifts from talking point to working tool, evaluating content and opportunity in real time.
Pure “curation” is thinking too small
Curation is often where this conversation starts, but it shouldn’t be where it ends. A curated deal can combine inventory, data, and rules into something a buyer can activate through a DSP. That can be useful when the deal is built around a real outcome. The problem is that curation has become a crowded term. Sometimes it means thoughtful packaging. Sometimes it means a margin layer. Sometimes it means a wrapper around inventory that the publisher barely understands and the buyer can’t properly audit.
Sell-side decisioning is broader than curation. It includes curation, but it also covers the logic underneath: what gets sent, on what terms, at what price, and toward which buyer goal. It’s a decision layer.
That’s why media owners should be careful not to reduce the opportunity to “we can now make more deal IDs.”
As Daniel says, “deal IDs are useful pipes, but the value sits in what flows through them. A deal carrying better context, stronger signals and controls, and clearer intent, does more than package supply. It changes the way the buyer sees the opportunity.”
The market’s attitude has shifted because the newer version of curation is more useful than the old one, which often looked like another way to extract money between the buyer and the publisher. The more transparent version lets publishers, agencies, data providers, and technology partners build around outcomes together, with decisioning happening before the auction rather than after the value has leaked away. Adoption clearly shows the shift. For instance, Index Marketplaces saw adoption triple year over year by the first half of 2025, proving that better-shaped demand paths beat more deal IDs.
Waste starts before anyone calls it waste
A lot of programmatic waste is created before anyone realizes. If a DSP has to parse an enormous volume of bid requests that were never going to meet a campaign goal, the system burns time, money, compute, and attention on triage. The buyer may still optimize, but by then, the pipe is already full of the wrong opportunities.
Sell-side decisioning changes that sequence. If the exchange can evaluate which impressions are a good fit – i.e. likely to meet a viewability threshold, drive a visit, match a suitability standard, support a creative requirement, or align with a supply-path preference – before the request leaves the sell side, the buyer’s system sees less clutter. What’s more, for both sides, good inventory is represented more accurately.
Case study examples are starting to make the theory harder to ignore. One retail campaign using a predictive model deployed on the sell side reported a 75% reduction in cost per site visit. Meanwhile, a financial services company using algorithmic curation reported a 90% reduction in suppliers and SSPs, a 52% reduction in domains, a 56% lift in conversion rate, and a 26% decrease in cost per conversion.
While those numbers shouldn’t be treated as universal guarantees, they still show why the industry is paying attention. When the first layer of decisioning improves the candidate set, the rest of the stack doesn’t have to work as hard to rescue performance.
Viewability also supports the case. If a campaign has a viewability requirement, the market can wait for the demand side to sort requests, or it can apply the requirement before it sends the bid request. The latter route doesn’t replace the buyer strategy, but it does remove obvious misalignment earlier, and gives the buyer higher-quality opportunities to consider.
There’s also an efficiency angle worth mentioning. Fewer unnecessary calls, fewer irrelevant requests, less duplicated compute, and less sprawling supply-chain activity should reduce wasted processing. Of course, that doesn’t turn digital advertising into a perfect system, but it does make a bloated system less wasteful.
Publishers can’t monetize control from the sidelines
The sell side has spent years talking about control, but control that can’t be monetized is incomplete. Publishers can block, floor, approve, and package. They also need to turn their own knowledge into products buyers will pay for.
Sell-side decisioning gives them a stronger route to do that and the commercial evidence is starting to back it up. According to Index Exchange platform data, the top 100 media owners on its exchange saw 53% higher eCPMs through Index Marketplaces compared with the open auction. While that doesn’t prove every publisher will see the same result, it does show why control becomes more powerful when it’s tied to demand, packaging, and pricing.
A publisher’s strongest assets go well beyond impressions. They include first-party relationships, reader behavior, content expertise, editorial context, app usage, video environments, and knowledge about what performs. When those assets can be shaped into deal-based products, publishers can sell more than access. They can sell a clearer path to an outcome.
The old trade-off was often niche quality or broad reach. Sell-side decisioning can help combine differentiated signals with a wider pool of eligible inventory, so a buyer can reach a more precise audience without staying trapped in a tiny supply pocket. The operational lesson for publishers is to simplify the pitch. Centralize first-party signals into a small, portable set of attributes buyers can understand and activate. Pick a go-to-market story – high-quality media environments, or conversion-driven outcomes. Trying to sell both at once muddies the proposition and slows the sale.
That also changes how publishers should think about sales. Handing a buyer a list of placements and hoping the DSP finds the value is not a strategy. Instead, understand which signals are unique, which partners can extend them responsibly, and which outcomes those signals are best suited to support.
Control also means being present during the curation process. Publishers have reason to worry when agencies, curators, and technology vendors package supply without enough transparency back to those creating the inventory in the first place. Collaborative curation should keep publishers at the table, with clear data permissions, pricing rules, ad quality protections, block lists, reporting, and fee mechanics.
Buyers want cleaner supply, not another workflow
Buyers don’t want a new interface for every new idea. That’s one reason sell-side decisioning has a realistic chance of adoption. In the best versions, the buyer still activates through the DSP, but the opportunity reaching that DSP but the opportunity reaching that DSP is already cleaner, and more aligned with the campaign goal. Agencies can still plan, measure, optimize, and report the way they always have. The only change is where some of the thinking happens.
That also opens the door for smaller, more specialized buying platforms to compete. If better-shaped supply is accessible through interoperable deal paths, you don’t need to rebuild the entire open internet inside your pipes to compete. The advantage shifts toward precision.
The same thinking sits behind the move toward containerized models and neutral compute environments. If bidding logic, agentic models, or decisioning tools can operate nearer inventory, the market can reduce latency and duplicated infrastructure. That can help buyers, publishers, and solution providers, provided the economics are transparent and the rules are shared.
To coin a favorite from Rob Beeler, “there’s nothing but hard work ahead,” to ensure transparency through the chain.
And, as Daniel points out, “moving decisioning upstream shouldn’t mean moving the black box upstream. The market needs standards, auditability, and clear fee mechanics, or the same trust problems will simply appear in a new place.”
The open internet needs intelligence without enclosure
The biggest strategic question is whether the open internet can compete with the simplicity of walled gardens without copying their centralization. Walled gardens have long benefited from tight control of data, inventory, identity, measurement, and execution. The open internet offers real advantages but it often makes buyers work too hard to access them. Sell-side decisioning is a way to close that gap without replicating the centralization that creates it. Publisher control, partner choice, and DSP interoperability stay intact. Why is this important? Because, as Daniel says, “the last thing we want to do is rebuild a closed garden under a different name”.
I’m impressed I haven’t dwelt on the word ‘agentic’ so far, but it needs to be said: agentic AI will raise the stakes. Agents can evaluate more signals and move faster than manual work, but they also need strong boundaries. A system that can make decisions at an impression level should be easy enough for everyone involved to understand the flow of money. That’s why standards such as the Agentic Real Time Framework, machine-readable supply, and clearer transaction rules aren’t optional extras. They’re what keeps an automated system accountable.
The same applies to emerging ad products around AI interfaces and new discovery environments. Publishers will need better ways to package content, expose context, manage rights, and prove value as buyer workflows become more prompt-driven. Sell-side decisioning can’t answer every question there, but it does improve the operating model: keep the signal close to its source, apply decisioning early, and make the result easy for demand to activate.
The open internet won’t win by becoming harder to buy from. It has to show its value, without giving up the independence that makes it worth buying in the first place. After all, intelligence helps only if it clarifies the market rather than hiding it.
Make the first decision count
Now let’s return to the reader in the coffee shop doorway. The moment had weather, context, intent, content, location, attention, and commercial possibility all wrapped together before the ad stack even touched it. But this time, the market acts on it before it gets flattened. The signal stays close to the source. The publisher gets credit for the value it created.
The practical promise of sell-side decisioning is simple: buyers get fewer irrelevant opportunities and more supply shaped around outcomes. Publishers get more ways to monetize quality, data, context, and trust without handing control to someone else. And technology partners get a scaled environment for models and data products that would be difficult to build alone.
The risk is just as clear. If sell-side decisioning becomes another opaque toll booth, publishers will lose confidence, buyers will question the economics, and the open internet will have recreated the problems it claimed to solve. But if it’s built around transparency, interoperability, publisher permission, and measurable performance, it can move value back to the place where it starts.
That place of value is the impression – and we now know, the impression knows more than the bid request lets on. The market has always understood that attention, context, and commercial opportunity meet there. Sell-side decisioning is simply the industry’s way of finally bringing buyers close enough to see it.