Oracle

Aug 2021 - Aug 2024

Project Overview

Project Overview

Problem

Recently acquired by Oracle, CrowdTwist has a suite of products they offer. In particular, their Segmentation tool allows marketing analysts to create filtered audiences of reward members to target promotions towards. There are many issues with the current user experience such as the limited amount of data a marketing analyst has access to from a front-end perspective. Currently, customers contact CrowdTwist to make more specific audiences for them. This can only happen via SQL and with limited human resources with a large customer base, this causes major bottlenecks.

Solution

A drag-and-drop experience where marketing analysts can place data together like Lego blocks to create their audience filters.

Impact

We were successfully able to launch the product and gather insight from customers and their ability to create audiences on their own and decrease the number of calls to Oracle over the product.

My Role

I led the end-to-end design for the new segmentation product.

What I Did

  • Market research (User goals, Success Criteria, Shape of Data)

  • User interviews (Identified pain points and target experience)

  • Low-fidelity and High-fidelity wireframes with usability testing

Query Builder Background

The segmentation tool boils down to a query builder. A query builder is like a playground of blocks that can be stacked together to produce an entity of rules.

Current Segmentation Landing Page

Current Segmentation Landing Page

CrowdTwist has many enterprises under its belt, which means each of these companies are building segments everyday. Take a look at a customer example of how many segments are built.

Current Segmentation Builder and Detail Page

Current Segmentation Builder and Detail Page

As you can see, just viewing the page itself was difficult. So, I took this as an opportunity to start from scratch with user goals and success criteria.

Before you read more, here are some industry terms (this will help)

Segment: Filter of data used to create an audience for promotions to link on
Segment Group: A group of segments attached together for promotions

User Goals

Success Criteria

Shape of Data

Shape of Data

The goal was to understand the scale and magnitude of segmentation to get a grasp on how huge this product is today and how much more it can offer once users could build their own rules.

The goal was to understand the scale and magnitude of segmentation to get a grasp on how huge this product is today and how much more it can offer once users could build their own rules.

As-Is & Future Scripts

As-Is: Segmentation Table

Today's experience with the landing page

Future: Segmentation Table

The goal experience for the landing page

As-Is: Segmentation Rule Builder

Today's experience with the rule builder

Future: Segmentation Rule Builder

The goal experience for the rule builder

Field Research

I combed through the internet and looked around for any related versions of segmentation builders and came across a few that had my interest and a few that I knew wasn't going to work out.

Adobe's Segmentation Builder on their Experience Platform was a good start. It was very clean and modern, but it was still heavily relying on coding. I took a look at Salesforce's segmentation, however, it worked very much like PowerBI and that wasn't matching what I was aiming for.

Funny enough, the best one came unexpected.

Oracle had a project relating to segmentation many years ago, but turns out it was abandoned after I contacted the PM and he gave me that message. It was a concept, but it had the strings of what I was looking for in building out this rule builder.

I took some inspiration from Adobe and Oracle and mixed them together and ran with it from there.

I combed through the internet and looked around for any related versions of segmentation builders and came across a few that had my interest and a few that I knew wasn't going to work out.

Adobe's Segmentation Builder on their Experience Platform was a good start. It was very clean and modern, but it was still heavily relying on coding. I took a look at Salesforce's segmentation, however, it worked very much like PowerBI and that wasn't matching what I was aiming for.

Funny enough, the best one came unexpected.

Oracle had a project relating to segmentation many years ago, but turns out it was abandoned after I contacted the PM and he gave me that message. It was a concept, but it had the strings of what I was looking for in building out this rule builder.

I took some inspiration from Adobe and Oracle and mixed them together and ran with it from there.

Lo-Fidelity Wireframes

Landing Page - Segments Tab

All of the pages you see here are based on Oracle's design system. With the heavy lifting of the page shells taken care of, I just had to focus on the layout and content. We had a search function and relevant search suggestions that could be used to look for segments. Status allowed users to know if a segment was running or if it had been archived. We showed the most vital information to users like the total audience members, how many offers were attached to a segment (called associated offers), and what kind of segment it was.

To give more context: Real-time refers to segments whose filtered audience will be given a promotion immediately versus Standard refers to segments that need to be given time to update. It's based on the engine that CrowdTwist runs on for separating purchase data from one kind of segment to another. It was a major reason why customers couldn't build segments on their own.

Landing Page - Segment Groups Tab

Similar to the Segments tab, Segment Groups had similar information and only needed to replace the segment type with how many segments were involved in the group.

Segment Detail Page

The segment detail page minimized as much page scrolling as possible. The goal was to take all read-only information and place it into data chips at the top of the page. Then the rest of the info could be focused on actual details that users cared about: the rules of the segment, associated offers, and segment groups that are linked with it.

Segment Groups Detail Page

The segment group detail page focused on the same information as the segment detail page; however, we wanted a way for users to view each segment that is used in the rules so they can get more context as to what it is and how it connects with the group.

Segment Creation Screen

The creation screen allows for users to enter the name, description, and selecting which type of segment they wish to build from.

Rule Builder

The rule builder was meant to act as a sleek and minimal drag-and-drop builder. A user can drag any attribute from the right side menu and place it into an existing rule block or an empty one to create separate filters. They could also search for the attribute and drag it from the side menu as well. Users have the ability to duplicate and delete attributes, and if they select multiple attributes together they could group them to separate a subsection of filters to not confuse the and/or logic functions.

Usability Testing

We went through the lo-fi wires with a focus group of current CrowdTwist customers and the feedback was excellent. I did notice in the sessions that users rarely went to the Segment Groups tab and when I asked about it, their answer was that they didn't know it was there. After discovering its use, it gave users new ideas on what it could be used for in terms of intersecting audiences. Based on this, I went back to the board and made the experience more intertwined with each other rather than separated. After all, the tables, detail pages, and rule builders were very much the same.

Another thing I heard periodically was the utilization of the detail pages. Many of the users didn't want to see all of their associated offers and segment groups tied in the page since they can grow quite large. One customer had over 150 offers tied to a segment. You can imagine the need for an overflow was very much a must-have. So, I divided the section of the page up into key pieces and went on from there.

We went through the lo-fi wires with a focus group of current CrowdTwist customers and the feedback was excellent. I did notice in the sessions that users rarely went to the Segment Groups tab and when I asked about it, their answer was that they didn't know it was there. After discovering its use, it gave users new ideas on what it could be used for in terms of intersecting audiences. Based on this, I went back to the board and made the experience more intertwined with each other rather than separated. After all, the tables, detail pages, and rule builders were very much the same.

Another thing I heard periodically was the utilization of the detail pages. Many of the users didn't want to see all of their associated offers and segment groups tied in the page since they can grow quite large. One customer had over 150 offers tied to a segment. You can imagine the need for an overflow was very much a must-have. So, I divided the section of the page up into key pieces and went on from there.

High-Fidelity Wireframes

Landing Page - Segments and Segment Groups Merged Together

After realizing users didn't see the tabs, I merged them together under one table and that seemed to do the trick for users becoming more aware of their groups.

Segment Detail Page

I split the segment detail page into three separate sections so that users can read all of their information in a non-scrollable format and only if they wanted to see all of their information for a specific section, they could click at the bottom of the panel to be taken to a page dedicated to it.

Segment Groups Detail Page

Very similar to the detail page above, but here we didn't need to list the groups a segment belong to because, well, this is a page for a group of segments added in a rule together. Users can click on the hyperlinked rules and be taken to the related segment.

Segment Creation Screen

Since all the segment types were under one roof, having the creation screen merged together actually replaced the need to have two separate pages dedicated to one segments and segment groups.

Rule Builder

The segment builder didn't change much, but we did make everything more cleaner in the UI, and users could save their current builds as drafts in case they ever wanted to back out of it without losing their progress.

Impact

After doing another run through with the focus group with these wires, it was concluded that it was a much better product and the merging of segments and segment groups was a major positive feedback.

After deploying this product in a beta with users, we noticed a signifcant increase in segments being built and less contact calls to Oracle over creating segments.

Overall, this project was a very good learning opportunity to own a product from start to end and take shape on creating a wonderful user experience.