Context
The Ad Gallery inside Disney Campaign Manager had been a quiet utility, a place to store assets and upload an occasional single video ad, not a surface built for serious creative work or the product's next chapter: AI-assisted ad generation.
Disney Campaign Manager · Product design · Sept 2025 – present
Building an Ad Studio for Disney Advertisers
01
Overview
Goal
Enable advertisers to create ads and assets, manage their creatives, and discover advanced ad formats, all from one place, while designing an information architecture that can absorb AI generation when it arrives.
Problem
Anything beyond a basic upload pushed advertisers into the full campaign creation workflow, even when they only wanted to build an ad and were not ready for line items, flights, or budgets.
Opportunity
Reframe the gallery as the Ad Studio: a creation-first workspace for formats, ads, and assets, so advertisers can work on creatives without launching a campaign first.
02
The Existing Workflow
Four phases, each one tightened the next.
01 · Discover
Audit the gallery
Audited the existing Ad Gallery to understand why it wasn't serving ad creation. Mapped the actions advertisers actually needed to take.
02 · Define
Reframe with IA
Reframed the surface through information architecture, both navigation and content organization. Defined the core distinction between ads and assets.
03 · Develop
Explore directions
Explored three IA directions for the studio itself, then a second wave of explorations for how AI generation could live inside it, three discovery models and three creation surfaces.
04 · Deliver
Land & annotate
Landed on a creation-driven IA for the Ad Studio and a hybrid draft generator for AI. Shipped annotated designs and a framework for unified AI creation.
03
Why the Workflow Was Breaking Down
Before redesigning anything, I needed to be specific about why the Ad Gallery couldn't carry ad creation. Four problems stood out.
- It was an asset repository, not a creation surface. The page was optimized for storing and retrieving assets, not for starting, configuring, or managing ads.
- There was nowhere to learn about formats. Advanced ad formats existed, but the gallery gave advertisers no way to discover what they were or how they worked.
- The only creation path was a side panel, too simple for the job. A lightweight side panel can't hold the complexity that advanced formats require. It worked for a single video upload and broke down for anything richer.
- Management actions were limited and unclear. It wasn't obvious how to manage your assets, link them to anything, or understand their relationship to campaigns.
The throughline: the gallery was being asked to do a job it was never designed for. Advertisers wanted an ad creation playground, and what they had was a closet.
04
Mapping the Ad Studio
To turn the gallery into a studio, I started with information architecture, looking at it from two angles.
01 · Navigation IA
How people move
How might we make it easy to access and reuse existing creatives? How might we make it easy to move between campaigns and creatives, and to link creatives to existing campaigns?
02 · Content IA
How things are organized
How might we make it easy to manage assets and retrieve information quickly? How might we organize the page hierarchy to support the key workflows, creating, managing, and browsing ads and assets?
05
Principles for the studio
Working through the IA surfaced four principles. Every exploration was measured against them.
-
Ad creation should be a reusable module, not a separate experience
Instead of building a standalone creation flow for the Ad Studio, ad creation should be a reusable module. Advertisers get a consistent experience whether they enter from a campaign or from the studio, and we maintain one system instead of two.
-
Ads and assets need to be clearly differentiated
Advertisers needed an unambiguous mental model: assets are standalone creatives that power ads; ads are the finished, deployable units. Separating them clearly, and making it easy to move between them, became a foundational requirement.
-
You should be able to start creating from anywhere
Creation shouldn't be gated to one entry point. Whether an advertiser is looking at their assets or their ads, starting a new ad should be one action away.
-
Design for AI-assisted generation from the start
We knew AI generation was coming. Rather than retrofit it later, I wanted the IA to anticipate it, so the structure could absorb AI without being rebuilt.
06
Exploring the Creation Model
With those principles set, I explored three IA directions for the Ad Studio. Each optimized for a different idea of what the studio fundamentally is, a library, a launchpad, or a set of quick entry points. The challenge wasn’t just layout; it was balancing retrieval, creation speed, and how much density advertisers could carry before the surface stopped feeling like a workspace.
Exploration 01
Search-driven
The studio as a powerful library, retrieval and filtering are the hero; creation is one path among many.
Strengths
- Strong for large libraries and power users who already know what they’re looking for
- Fast path to find ads, assets, and campaigns through search and filters
- Familiar mental model when the job is “get me to the thing”
Tradeoffs
- Treats creation as one of many search outcomes, not a first-class destination
- Creation entry points can compete with dense search chrome
- Less legible for advertisers whose primary job is to start work, not retrieve it
Exploration 02
Creation-centered
The studio as a launchpad, format templates and Create stay visible so advertisers can start from wherever they land, whether on Ads or Assets.
Strengths
- Best balance of the three, fast into creation while management stays approachable
- Templates across the top make format discovery visible without a separate hunt
- Left-rail Ads / Assets reinforces the two object types as peers
Tradeoffs
- Less optimized for retrieval at very large library scale
- More surface area competing for attention on first load
- Requires discipline to keep discovery, creation, and management equally clear over time
Exploration 03
Action-driven
The studio as a set of quick workflow entry points, create ad, upload asset, and browse templates sit forward in a denser, table-forward layout.
Strengths
- Fast access to the highest-frequency actions for operators who live in the list
- Scannable density for teams managing many ads at once
- Clear affordances for “do the thing now” without hunting in navigation
Tradeoffs
- Doesn’t scale as gracefully when the action set grows, starts to read like a toolbar
- Less room for education, previews, and format storytelling inline
- Surfaces actions but not the broader workspace story the studio needs long-term
Each direction solved a different slice of the problem well. Together, the explorations made it clear the studio needed to feel creation-first without giving up management, which set up the creation-driven direction in the next section.
07
Choosing a Direction
I went with the creation-driven IA. It gave advertisers the fastest path into ad creation while still making ads and assets easy to manage, the best balance of the four principles.
The recording below shows an advertiser browsing the Ad Studio, discovering advanced ad formats, learning more about what each format offers, and starting ad creation from that path. The screens that follow are the final creation-driven workspace (with interactive annotations) and the dedicated Assets view opened from the left rail.
1 · Creation-driven workspace
2 · Assets page
08
Designing for AI Integration
Because the Ad Studio is where advertisers prepare creative before they think about line items and campaigns, it was the natural place to introduce AI-assisted generation.
I built a Jobs-to-be-Done framework to understand how AI generation would change both the Ad Studio and the act of creating an ad from it.
When
an advertiser is preparing creative for an upcoming campaign and isn't sure where to start,
I want to
describe what I'm trying to do and see a range of generated directions I can pick from,
So I can
move quickly from idea to a working draft, then refine it in the composer with precision.
Two things became clear from the framework:
- AI needs a discovery space. Advertisers would need somewhere to learn what the AI could actually do, its capabilities weren't self-evident.
- AI needs an exploration layer. A space where advertisers could use AI to generate ad and asset explorations before committing to a direction.
09
Building the Discovery Layer
I sketched three concepts for what a discovery layer could look like, how an advertiser first encounters AI generation inside the studio. Each option tested a different bet: lead with the prompt, treat every entry as a peer, or let people move between AI-first and format-first without committing upfront.
Discovery 01
Prompt-first
The advertiser enters a prompt, generates ideas, then steps into the ad creation surface to refine with precision, with recent work and AI-generated starting points visible alongside the prompt.
Strengths
- Strong, opinionated entry for advertisers who already know what they want to say
- Makes the AI capability unmistakable the moment you land in the studio
- Surfaces recent ads, assets, and AI-generated starters next to the prompt
Tradeoffs
- AI can read as mandatory, risky for visitors who came in to do something specific
- Less parity between AI and non-AI routes when the prompt dominates the band
- Needs careful framing so “prompt” doesn’t feel like the only door into the studio
Discovery 02
Three equal entry points
Generate with AI, create from a format, or upload a VAST tag, all three sit at the top with equal weight, so no single path is visually dominant.
Strengths
- Doesn’t force AI on teams who aren’t ready to adopt it
- Respects advertisers who prefer classic creation or VAST workflows
- Keeps format and upload paths as visible peers in the entry band
Tradeoffs
- AI loses a standout “discovery” moment, it reads as one label among many
- Higher cognitive load when every route competes at the same visual weight
- Harder to tell the story of what AI can do uniquely well at a glance
Discovery 03
Toggle between modes
Advertisers switch between AI-prompt-driven discovery and format-driven creation, with state that persists so they aren’t resetting the room every time they come back.
Strengths
- Familiar toggle pattern for moving between “discover with AI” and “build by format”
- Persistent mode reduces rework when people bounce between workflows
- Lower cognitive load than inventing a wholly new IA metaphor
Tradeoffs
- Introduces a mode users have to remember (“which side am I on?”)
- Can obscure first-visit education when the default is ambiguous
- Two mental models in one surface can split how you measure success in research
Together, the three directions clarified the tension between making AI discoverable and not hijacking every advertiser’s intent, which informed how much prominence AI should get next to formats, uploads, and existing creation paths.
10
Defining the Creation Surface
Alongside the discovery layer, I explored three models for the creation surface itself, how AI and manual creation actually coexist.
Surface 01
Hybrid Draft Generator
AI generates a draft → edit in the DCM composer.
Strengths
- Preserves the familiar DCM workflow
- Low disruption, easiest to ship
- Full manual control after generation
Tradeoffs
- AI can feel like a separate step before you reach the composer
- Iteration loops may be slower when generation and edit are sequential
- Less fluid for open-ended creative exploration than a dedicated AI surface
Surface 02
Dual Composer
Separate AI composer + manual composer.
Strengths
- Optimized experience for each workflow
- Strong AI ideation and iteration without fighting manual controls
- Faster AI experimentation when the surface is purpose-built for it
Tradeoffs
- Fragmented product experience across two composers
- Users must choose a workflow upfront
- Higher design and maintenance cost to keep both surfaces coherent
Surface 03
Unified AI Composer
One composer with manual + generative tools together.
Strengths
- Single creation workflow end to end
- AI feels native to the product rather than bolted on
- Most flexible long-term as capabilities grow
Tradeoffs
- Most complex to design and build well
- Risk of UI overload if density isn’t managed carefully
- Requires rethinking the current composer rather than extending it lightly
The decision
The Hybrid Draft Generator. It lets advertisers explore widely and generate a range of ideas,
then step into the composer environment to refine, without forcing a workflow choice upfront or requiring a
full rebuild of the composer.
11
Challenges
- Everyone wants the unified composer, but we're not there yet. The unified model is the most compelling long-term, but it's a significant lift and requires rethinking the existing composer. The hybrid model lets us ship, learn, and decide how to proceed from real signal.
- AI generation and the Ad Studio are on different timelines. The Ad Studio is launching soon; AI generation hasn't launched yet. The IA had to stand on its own and be ready to absorb AI when it arrives.
12
Outcome and Impact
The Ad Studio ships in June 2026. The IA established a defined, reusable creation module rather than a parallel experience, and gave the team a shared vocabulary (ads and assets as two clearly-named object types with peer surfaces) that's now used across PM, design, and engineering.
The AI explorations established a framework for how generative tools could live inside the Ad Studio without requiring a full rebuild of the composer. The hybrid model gives the team a near-term path to ship and learn; the unified composer remains the long-term vision the IA is structured to accommodate.
- Clarified how the Ad Studio fits into ad creation. A defined, reusable creation module rather than a parallel experience.
- Set the product up for AI creation and scale. An IA that can absorb AI generation without being rebuilt.
- Established a shared vocabulary. Ads and assets are now two clearly-named object types with peer surfaces, used across PM, design, and engineering.
The Ad Studio ships in June 2026. AI generation is still in concept phase.