AdRoll Audience Builder

Case Study

AdRoll is a Martech SaaS org that builds tools and products to empower marketers. As the product evolved to include more channels besides just ads like email and SMS, there was a need to create a new segmentation tool in order for AdRoll to evolve into a robust marketing hub. This case study is highlighting the new AdRoll Audience Builder.

AdRoll historically has just been an Advertising platform, however, recently they’ve been making the shift to include more channels like Email and SMS in order to become a full marketing suite.

I have been focused on overhauling their legacy Audience product in order to accommodate the new business initiative.

The legacy Audience product was designed to support only ad campaigns. It was only able to create audiences based solely on site visitors and customers’ pixel-based activity. The pixel activity allows AdRoll to create audiences containing anonymous individuals. By expanding the product capabilities to include a Customer Data Platform and various partnership integrations with Ecomms and ESP’s like Shopify and MailChimp, AdRoll will now be able to segment both anonymous and identified contacts in one audience. Giving AdRoll a unique and potentially advantageous position in the market.

Business Objective:

AdRoll’s business objective was to harmonize the current fragmented audience-building experience in the platform in order to reduce frustration and confusion for the user. The business needed the audience builder to allow users to create lists of contacts based on both pixel, ecomm, & esp data, which includes events, behaviors, attributes, and other various conditions. These audiences needed to be targetable in all campaign types (Ads, Emails, & SMS).

This required AdRoll to create some new BE and FE infrastructure, and ultimately to build a Customer Data Platform.

Previous Audience Building Experience in AdRoll:

The Research:

Activities included, but not limited to:

Research Synthesis:

After interviewing the subjects, I synthesized the research, mapped out the user journey of creating an audience for their marketing campaigns, and identified pain points, data points, and opportunities.

We learned the Marketing role is quite wide and they have many jobs to be done, these JTBD can change frequently depending on their individual and brand goals which ended up making us consider a variety of behavioral segments and thinking styles.

Our Users:

Targeted Marketing Behavioral Segments for AdRoll.

Our Users - Mental Model:

 User Journeys (Audience Building):

Pain Points:

-Inside AdRoll users have to navigate and understand the different types of audiences they can create; pixel vs. CRM and how they can be leveraged depending on how they were created.

-User’s often don’t have any clue who to target.

-Sharing contacts between software is a huge pain without an integration.

-Oftentimes users want to A/B test audiences because they aren’t confident their campaign strategies will compel their audience enough for conversion and currently have no way of doing this.

-Users have no way to use their eCommerce platform data with pixel, ESP, and contact attributes to create a single flexible audience that is able to be targeted in multiple types of campaigns.

Opportunities:

-Create a single solution for audience building inside AdRoll.

-Leverage our integration partnership data for events and attributes in combination with our pixel data for hyper-targeting of both identified and anonymous contacts. This will expand a customers’ reach up to 10x+.

-Accommodate the current AdRoll user behavior and allow the Audience builder to be accessible from inside campaign creation as well as in Audiences for flexible working styles.

-Allow for editing of existing lists made with the Audience Builder so users can quickly adapt their strategy if they are not getting the response they expected.

The Problem:

Brand Builders and Operation Orchestrators want to create audiences based on customer behaviors and attributes to match their brand’s campaign strategies. They currently do not have a way to do this inside AdRoll which leads them to create pixel-based audiences and only launch ad campaigns in the platform. The user is then forced to navigate to different platforms for any Email or SMS channel engagement, even though AdRoll currently does offer an Email solution.

Hypothesis:

If we can do the following things:

• Create a single audience-building solution for our users whose audiences can be leveraged in all campaign types.

• Allow users to leverage contact events and attributes for advanced and hyper-targeting.

We think:

• Users can get more specific with their audiences in order to match their segmentation to their campaign strategies and get greater ROI.

• Campaign types other than ads will be engaged across the platform.

• AdRoll value and satisfaction will increase for the user.

KPI’s:

Our primary success metric is user value, rather than focusing on which channels will get an increase in engagement after launch, and by how much, the real value in this product is making the user’s lives easier by giving them the tools they need to get the most return and engagement with their brand campaigns.

Increase audience creation by 20%.

Driving more engagement to AdRoll’s email product offering would also be a bonus.

Competitors we assessed:

Design Principles:

The Solution:

Audience building is the ‘reach’ part of a customer’s product lifecycle, and also within the execution part of a Marketers journey as building an audience is often something you do early on in the process of campaign creation. Your ‘reach’ to prospects and customers is very dependent on a solid segmentation strategy. Our solution enables hyper-targeting via behaviors, events, and attributes allowing a marketer to get closer to their campaign strategies and achieve a more specific and focused reach which should lead to more conversions.

Design - Technical Modeling:

This project required some intense technical modeling to define our necessary conditions and operators and to negotiate what was MVP. Each condition selection dictated the available operators, making the modeling expansive. The initial model is too big to share via image, so here is a snippet.

Design - Lo-fi Iterations:

In our lo-fi exploration we explored many different styles of list building and different contexts to accessing the list builder across the platform and then ran a usability test on a few different ideas to narrow in on a solution.

 High Fidelity Key Screens:

After usability testing we iterated again and created a high fidelity solution. Here are a few key screens.

 Design - Prototype:

Design - Engineering Deliverables:

This solution could have thousands of screens if I were to design every possibility of conditions and operator selections. Instead of working harder I decided to work smarter and just map out the entire solution for the engineers in the UI elements for easier reading. With this map and the high-fidelity solution of the primary UI/UX patterns, error states, and thorough documentation of the use cases they had everything they needed.

Beta Launch:

Once the development of a select number of events and attributes was complete we chose 300 Shopify users for our beta launch.

We’ve been observing users usage with a tool called Fullstory and monitoring for any bugs along with continuous QA, as well as checking in with the beta users for user satisfaction and feedback.

Results:

So far out of the 300 Shopify users, 76% of them have created at least one audience to use in a marketing campaign.

Next Steps:

Following up with our Beta users with a customer satisfaction and feedback survey that will help us start to measure our first KPI, user value increase.

On April 15th, 2022 we will be launching to the general population of AdRoll users. At that time we can start to measure the true impact of the audience builder against our key performance indicators.

Designing some audience reporting to these individual list pages.

Now that we have one solution for audience building the next venture for AdRoll is marketing automations that are only possible because of the backend and front end CDP work the Audience Builder required.

Project Learnings:

Work smarter not harder.

This particular project could have had me creating thousands of high fidelity mocks due to the nature of the design. Rather than spending my time doing that I found it more beneficial for both me and my engineering counterparts to map the solution out in a detailed user flow with the support of some key screens and use-cases.

If you don’t have product requirements - make your own.

Now this isn’t ideal, and obviously we as designers want the Product Manager position to prepare all the requirements before design even begins but we as a team were moving so fast this exercise became a collaborative one, and I believe it gave me a deeper understanding of the technical infrastructure.

Keep it simple.

This actually is a very technical solution but the interface feels light and seamless, that was done on purpose to remove the cognitive burden from our users. I tried to reduce as much noise as I possibly could leaving only what was absolutely necessary in the design.