AUDIENCE SEGMENTS
The product, Audience Insights, transpired from our internal marketing analysts workflows where they use custom SQL to tap into health systems data to create hyper-targeted complex audiences for direct mail and email marketing. We found a need and opportunity to productize the process giving the ability to marketers, who did not have the technical skills to create custom SQL, have the ability to curate target audiences for campaigns.
While testing the Audience Insights solution with the marketing analysts, we found a feature that was available in the SQL workflow, also known as the classic system, that had not been solved: the ability to create nested criteria to produce segments within an overall audience.
In order to arrive at a solution, we had to understand how nested criteria and segments were used to create complex audiences.
We sat down with a couple of our marketing analysts for a white boarding sessions to understand how they structured the SQL query and we found that there were three important groupings that defined a complex audience:
Overall Audience - Defined as global criteria that every individual in the audience would be defined by.
Groups - Defined as criteria that further refines the overall audience.
Sets - Defined as criteria that further refines the group.
We came to understand this hierarchy by the following example audience that is used as a template in creating complex audiences with nested criteria:
Now that we had a better understanding of how an audience in SQL was structured by the marketing analyst team, we discovered that the Integrated Marketing Manager and Business Intelligence Engagement Specialist had two primary use cases in creating a strategy that included groups and sets:
As a marketer, I want to be able to create segments (including groups and sets) in order to target individuals by different marketing material that fit their demographics, needs and behaviors so that my campaigns are more finely tuned leading to higher conversion rates.
As a marketer, I want to be able to create segments (including groups and sets) in order to target individuals more accurately so that I can save costs on marketing material.
In addition to the use cases discovered, we were able to understand some of their pain points when requesting audiences in their current workflow and how the classic system presented the SQL. The classic system is how they receive the target audience and counts from the marketing analyst team based on the criteria for their strategy.
Some of the findings that we found during these discovery sessions with IMMS and BIES were:
Unanimous amongst users that they liked how the global count at the top was clearly stated with the deduplicating (household, individual, address) methodology clearly stated.
Additionally, duplicating audience segments would be extremely helpful to have present while running queries in Audience creation. This allows for having the most accurate numbers when exporting an audience to a direct mail list or syncing to 3rd party marketing systems.
Suggest renaming Audience Group and Sets to Audience Segments in order to be more inline with universal marketing standards.
The goal was to create an interface that allows marketers to tap into their health systems curated data and have the ability to define sub-segments within an audience for specialized, hyper-targeting based on criteria chosen by the user.
User/Business Value
By allowing sub-segment targeting within an audience, users will be able to create multiple marketing versions for a call-to-action and budget for campaigns more precisely while targeting the right consumers just as accurately if not better.
Challenge
We broke the challenge down into three different objectives. These ‘how might we’ questions drove exploration, experimentation and ultimately our solution.
Integrate segments to Audience Insights
HOW MIGHT WE integrate the segment hierarchy into the Audience Insights current workflow?
Cap and sort segments
HOW MIGHT WE allow users to cap and sort audience segments for budgeting?
Reporting on segments
HOW MIGHT WE show users how their audience segments performed in dashboards/reports?
Integrating segments into Audience Insights
Now that we understood how and what the uses for segments were, we were able to dig into understanding how we might integrate segments into the current Audience Insights workflow. In order to do so, we created a schema of how targeting audiences worked in the current UI. In the workflow diagram below, we were able to pin point that our workflow wasn’t completely different from how our internal teams set up audiences, we were just missing a layer in the hierarchy. This helped guide our requirement gathering and ultimately our solution.
Audience Insights Workflow with Segments
Sourcing Data from Ideas Portal
In addition to understanding the workflow, we utilized data that we gathered through user submitted feedback to help with rounding out our requirements. We first sorted through 50 ideas in our feedback portal and identified common pain points among our users that are on the legacy system. We based common pain points on the following criteria: mentioned multiple times, points and comments.
Feedback Portal w/an Idea
We identified two pain points to the current process and prioritized into our requirements for MVP:
Total Segment Count. When creating a segmented Audience, the Audience segment tab shows the overall audience quantities and the remainder quantities, but not the total quantities for segments. A user mentioned: “The number is probably the most important to have since any list created or audience sent to marketing automation will reflect the sum of the audience segments.” The total quantity number is necessary when creating an accurate report and budget for marketing campaigns.
Duplicating Groups. “When building out Audiences for segmentation there are times when the criteria may be very similar between segments and it would save time if a user could duplicate then edit an existing Group rather than start from scratch every time. For example, Orthopedic campaign, Group 1 = older, males, who are patients; Group 2 = older, males, non-patients; Group 3 = younger males, patients....etc. You would create Group 1, duplicate, then just edit the particular variable.”
Audience Segments Design Solution
After we gathered enough information on workflow, data and feedback on Legacy products, we were able to pin point our initial design requirements and find a solution. Outlined below is a draft of our user story and design requirements:
User Story
As a marketer, I would like to break down my overall audience into segments so that I can target individuals with tactics that align more closely with their demographic and behavior.
Design Requirements
Audience Segment Tab is added to the secondary navigation on the List Workbench
Users are able to add audience segments and pull criteria in utilizing the existing UI for the overall audience.
Users will also be able to break down audience segments by subgroups using criteria with ‘and’ statements.
Users can see the breakdown counts for audience segments and the total individuals for each segment.
Users are able to prioritize which group has more available global audiences to narrow down. Order matters.
Ability to delete and change audience names.
Nice-to-Haves:
Duplicate Audience Segments to make the workflow more efficient.
Prototype
User Test
To gain understanding and validate our initial requirements, we tested our design solution with 5 of our two primary users of the application: marketers and client support. The test that we conducted was based around 3 primary goals:
How do users want or intend to use a target audience builder within Audience Insights?
How well does the Audience Segments allow for users to create their segments for the campaign strategy?
How do users want to be able to report on Audience Groups?
For more information and details on the test, linked here is the test outline.
User answers were analyzed in a google sheets and common points were identified.
In our findings document, we found that there were 3 levels: Primary Uses Cases, Major and Minor findings. Primary use cases define and validate how individuals will want to use segments, major findings are where most users feel strongly that the UX is necessary for MVP and minor findings are not necessary but nice-to-haves for the feature. Below are the major findings documented and you can find:
3 out 4 users expected to drag the audience groups to change the ordering. One user expressed: "The use of arrows could be extremely cumbersome when you have 20 different groups that you have to move around."
The max number that they have seen for audience groups has been around 70 groups - the example they provided was that the client was using groups to target individuals by an encounter with a primary care provider. Location would be a lot of groups but not as many - biggest need for a lot of groups is provider.
3 out 4 users expressed the need for description this was primarily the workflow that IMMs are using currently for Marketing Automation. The user expressed: 'In HG CRM we have to name our audience groups - 1a ,2a, 3a, etc. so a sub description would be helpful. Need the description to provide the non-technical naming convention. Group name needs to be utilized for programming.'
UX Recommendation: Get clarification for Marketing Automation requirements from Chris Besting
4 out 4 users strongly expressed that they definitely think it is necessary to provide the deduplication method within the global filters in order to have the numbers reflected accurately as soon as possible. Additionally, they mentioned that they do not think they would use a deduplication method ever or rarely for email channels.
3 out 4 users expected that wherever individuals/people matching criteria was referenced that they would want that to dynamically change depending on the deduplication method chosen. Ex. Individual, Household, Address.
4 out 4 users expressed that being able to duplicate audience groups would be extremely helpful to their workflow since groups are only differing by one selection within criteria.
Final Deliverable
Final Prototype
Development Specs
