Yahoo Ad Manager Plus was Yahoo’s first demand side platform (DSP)  that allowed advertisers and agencies to buy advertising across multiple channels that includes display, mobile and native and devices. The platform has undergone multiple iterations and is currently in the process of being 'refreshed' again to further streamline workflows and improve its usability.

 

V1 - YAHOO AD MANAGER PLUS

 

Team

I was the first UX designer on the team tasked with understanding the broader advertising landscape and demand side platforms in particular and to work closely with the product manager to outline a plan for what we would deliver over the course of the quarter. The team eventually grew to 5 designers with each designer in charge of separate feature sets.

Goals

Create a fully functional demand side platform that would allow advertisers to leverage Yahoo's extensive inventory and user information to programmatically buy advertising and optimize their campaigns.
Achieve feature parity with competitor products like DoubleClick and Turn.
Ensure smooth integration with other ad platforms within Yahoo.

RESEARCH

 Deep dive to understand the digital ad ecosystem by reading industry reports

In order to better understand the complex advertising ecosystem, I read quite a few industry reports on the industry focused on topics like Demand Side Platforms, Real Time Bidding, Programmatic Buying, Inventory Yield etc. The motivation was to understand where the DSP we were building would be positioned in the marketplace.

Comprehensive competitor analysis

That included reviewing systems like Google’s DoubleClick, Turn, Mediamath etc. I also reviewed Yahoo's existing ad offerings across display, video and mobile to ensure seamless integration between the platforms.

I created an Ad Glossary to refer back to (given the endless acronyms in the ad tech world) which was a useful tool to help onboard other designers to the team.

Formative User Research

I worked closely with the UX researcher on the team to come up with a series of interview questions and some tasks to identify

- User roles, both internal and external.
- Their current workaround to compensate for the lack of a full fledged DSP.
- Experience of our users with other DSP tools in the market.
- Workflow and usability pain points

DESIGN

Based off of what we learned from the initial background and user research, I worked with the PM to whiteboard process flows for each type of user role in the system. We also did a broader application level flow to identify where the new DSP lives in relation to the existing ad products. We had a lot of help from our subject matter experts to understand the complex feature sets that we needed to include within the DSP. The engineering team was looped in early in the process to help break down the tasks into bite sized weekly deliverables.

Top feature sets

Advertiser, Campaign and Line creation, Reporting, Creative workflows, Dashboard design, Audiences and Targeting features, Bulk edit, Notifications among others. I was the first designer to work on some of these feature sets and worked closely with other designers for some others.

The process was extremely agile and the requirements were based off of all our upfront formative research. My deliverables were usually process flows, wireframes and in-depth interaction design specifications and documentation. This work was on a parallel track with the creating and documentation of our most common design patterns. By the time the project got to this stage, we had a dedicated visual designer on board to help provide detailed visual design specifications.

Bulk Editor

The bulk editor was a mammoth undertaking considering the number of line items and attributes that need to be tweaked. My approach was to try and categorize the attributes into different buckets and provide an easy way to specify what elements could be tweaked. I also recommended a bulk editor 'mode' to maximize the space available and provide room for additional controls that may not have been important on a simple grid page.

The bulk editor designed was well received and pushed to production and is still live on the current DSP.

Fig 1. Bulk Editor - currently in production

Fig 1. Bulk Editor - currently in production

Campaign List Screen

Fig 2. Worked on the primary grid shown above and the side panel for additional information.

Fig 2. Worked on the primary grid shown above and the side panel for additional information.

Fig 3. Audience building and targeting in the DSP

Fig 3. Audience building and targeting in the DSP

Top design patterns

Multiple UI elements and patterns that I contributed to creating - table and form field interaction design and styling, data visualizations, filters, dashboard cards, contextual help among others.

This entire process was iterative with weekly check ins with our users to see if we were on track to deliver the right features and if the features they tested were understandable and easy to use. We also had a representative from the engineering group in these sessions to advise us on feasibility and LOE’s for recommended features. Post the primary flows, I worked closely with the visual designer on the team to turn my wires into more polished mockups with specs to handover to the engineering team.

Help Guide UI

Help guide UI explorations for in context help when using the system.

Help guide UI explorations for in context help when using the system.

AUDIENCE BUILDER 2.0

Before we started the overall refresh, I was tasked with improving the existing audience builder to make it more streamlined. There were usability issues in the current version that made it difficult for users to search for and use audiences for their needs.

The goal was to deliver a tightly integrated audience product suite with the ability to explore, build and target audiences. - Modular design to ensure seamless integration into existing demand side products..

Challenges & Approach - Audience Builder
The audience build section had some technical challenges that we needed to address before we could move forward with the design stage. It involved cleaning up the data taxonomy, building a better tagging system, identifying key performing segments etc.

Once we addressed the challenges, I came up with a design proposal that would account for a variety of use cases :
- Building a specific type of audience
- Building a ‘composite’ audience using existing audiences.
For the composite builder, I recommended a two panel layout that would effectively merge tasks associated with browsing, searching and building audiences. The two panel layout was widely appreciated and is currently being built as our default composite building solution. We also did a quick round of user interviews to see if the structure that was being proposed aligned with their mental model of audience construction.

We came up with multiple iterations based on these interviews and general design reviews before selecting the version that we are currently building. A more detailed account of specifics related to this complex builder tool cannot be included in this write up but is something I am happy to review in person.


V2-BRIGHTROLL DSP

After Yahoo acquired BrightRoll which is one of the largest video ad marketplaces today in 2015, the design teams at both companies have been working together closely to create the next generation demand side platform. The goal of this new iteration is to streamline workflows, improve usability and to refresh the visual design.

I've provided some examples below of areas that I am focused on from a UX design standpoint. A lot of this work is confidential to Yahoo and may not be reused elsewhere.

GOALS

Streamlined workflows - for critical & repetitive tasks.
Beautiful and usable
Leverage familiar patterns - Excel in a modern web application.
Make the UI fast, efficient & extensible - Keyboards shortcuts, spotlight search and global actions.

 

RESEARCH

The goal of this refresh was to create a new, market leading demand side platform with an emphasis on streamlining workflows for tasks that most users carry out multiple times on a daily basis. The research phase for the refresh was a fast moving and extremely thorough investigation into the behaviors of our users using the current DSP.

User research sessions

We conducted many user studies with both internal and external users for formative research. The goal of this early research was to identify, from a high level perspective, the issues with the current platform. Users commented on a variety of issues - the overtly hierarchical structure of the current platform, the information density being too much, issues with patterns that didn't scale over time with increased data, inconsistencies in the design elements used across the platform etc. Search across the site was another big issue for the users, as it was slow and didn't give clean, categorized results when used. Users had trouble navigating around the platform because of this.

Part of the problem stemmed from the fact that we didn't account for how our design choices in terms of patterns and screen layouts would scale over time.

Competitor analysis

In addition to these high level user interviews, we did a competitor analysis to understand how the market leaders in the space were solving these issues. Under consideration were tools like Google's DBM, Turn, MediaMath among others.

 

DESIGN

We conducted week long design sprints to tackle the high impact areas on the site (audiences, reporting, inline editing) before moving on to more detailed design sprints for individual features. The details for each of these sections was tackled using the vision brief approach over a 1-2 week sprint cycle. For additional information on the structure we used for our design sprints and vision briefs please use this link.

At the end of every week long sprint, we would present our work (either as a prototype or simply disparate screens) to our end users for some quick feedback and course correction. We also had a dedicated weekly check in with the product managers and engineering teams to review our work and to help them get started on the architectural work for implementation. As we formed a baseline around the general navigation, structure and features sets the visual designers on the team started working on concepts that we could put in front of our users.

This fast moving, sprint by sprint creation of the refreshed version of the DSP has been extremely challenging and rewarding at the same time. All the designers on the team understand the product better and we are always on the same page with the engineering teams.

Top feature sets

Advertiser, Campaign and Line creation, Dashboard design, Audiences and Targeting features, Notifications, Search Plus among others.

My deliverables here were primarily process flows, wire frames and in-depth interaction design specifications and documentation. Like the first version, this work was on a parallel track with the creating and documentation of our most common design patterns.

 

CAMPAIGN GRID

Campaign Grid

Campaign Grid

CAMPAIGN SETUP

SEARCH PLUS - SPOTLIGHT

The updated search experience is modeled after tools like Alfred and Spotlight as a mechanism to quickly navigate to desired elements. Additional features include the ability to create new 'objects' from any page and a more robust search result list.

DATA MANAGEMENT

I have been working over the last quarter to create a Data Management Platform from scratch within our DSP to help our data providers, advertisers and internal teams to easily track audiences and taxonomies injected into the system. The platform allows them to also easily diagnose errors that may have occurred during data transfer and options to fix them.

This project is a significant undertaking and more than can be covered here. I am happy to provide additional context if required.

The BrightRoll DSP is an ongoing project, with plenty more to come over the next year. It is going to be the DSP to beat!