TL;DR
Programmatic ad spend was shifting hard toward deals, but no self-serve user could find a deal inside our platform without a nine-step workflow buried in campaign settings. I led moderated concept research with media buyers that reframed the project around three jobs: 1. Browse with confidence 2. Price with confidence 3. Activate fast. These directly shaped what shipped first (and what got deliberately cut). Within the first months post-launch, the marketplace reached double-digit account adoption from a standing start.
Why this mattered
Across the industry, spend was moving from open auction to deals, with projected to reach 83% of programmatic inventory. Our platform's deal spend had roughly doubled as a share of media spend in a single year. But the platform experience hadn't kept up: to even see which deals were available, a user had to create a journey, add a campaign, add a tactic, open targeting settings, and click through nine steps. There was no way to browse, no pricing context, and no way to save a deal except applying it to a live campaign.
The clearest symptom: Internal teams kept track of available deals through an Excel spreadsheet. That's an existential product gap, not a UI polish problem.
Research questions
The product team arrived with a concept: an e-commerce-style inventory marketplace along side a growing feature wishlist. My job was to narrow it. I framed the study around three questions:
How do media planners and buyers actually discover and evaluate deals today, and what mental models do they bring from other DSPs?
What information is decision-critical at the point of choosing a deal?
What does the path from deal selection to campaign activation need to look like to be adopted?
Approach
I ran moderated concept-testing sessions with media planners and buyers, walking through mid-fidelity prototypes of the marketplace: merchandising storefront, listings page, deal detail drawer, and the "add to journey" flow.
Moderated over unmoderated was a deliberate trade-off: the prototypes were early and partially populated with dummy data, and the decisions we needed, which data points create buying confidence, where inventory should live in the navigation, required probing "why" in the moment, not task-completion metrics. I triangulated with the current-state workflow evidence (session recordings of real users struggling through the existing flow) and competitive teardowns of how other DSPs merchandise inventory.
Sessions were recorded and synthesized in Dovetail, with every insight traceable to timestamped clips.
Learnings
1. Buyers browse like shoppers, not like database users. Participants immediately mapped the marketplace onto familiar consumer patterns, "I feel like I'm in a streaming TV platform a little bit, which is great", and said their first instinct would be to type a publisher name they already trust. The e-commerce framing wasn't a gimmick; it matched the real mental model. But it also exposed a gap: users who didn't have a publisher in mind didn't know what to search for, so browsing scaffolding (categories, curated rows, filters for CPM range, market, ad type) had to carry them.
2. Pricing ambiguity was the confidence killer. Every participant zeroed in on cost data. Floor price alone wasn't enough, "if you're going to use the Journey Canvas, you put in a dynamic CPM range. Well, what the hell should that range even be?" Participants asked for floor price ranges, historical clearing costs, and win rates, and flagged that the "win rate" label was ambiguous (win rate for which campaign?). One participant proposed renaming CPM to "Floor CPM," which shipped. The insight: the marketplace's job wasn't just discovery, it was bid-setting confidence.
3. Activation friction was the adoption ceiling. The strongest emotional reactions in every session were about the current workflow: "I shouldn't have to go into a live campaign, drag something out of the canvas, click into it 35 steps later, which nobody knows how to do." Participants converged on a hard requirement, one to two clicks from anywhere to inventory, and a direct path from a selected deal into a new or existing journey.
Results
In the first months post-launch, behavioural analytics showed the marketplace browsing surface reaching double-digit account adoption among external users from a standing start, with sustained week-over-week growth, meaningful for a workflow that previously had zero self-serve entry point. Deals administration surfaces saw visitor and page-view growth in the triple digits as internal teams migrated off the spreadsheet.