Identifying the Need to Overhaul
This is the story of recognizing the deficiencies of a system's foundational element, how it's stressing operations, and uncovering how deep the changes need to reach.
Background
Strategic research shows there's a problem
Targeting is a foundational feature of a DSP (demand-side platform). For a managed-service business, the research splits between the clients and the internal users.
Clients rated targeting capabilities as a top differentiator.
Users rated targeting setup as the top complaint.
Investigation
Directional user research
We conducted comprehensive directional user research sessions both remotely and during a two-day in-person conference to initiate our investigation.
Takeaways:
The tools were inconsistent.
The same targeting needed to be set up in too many places.
They couldn't edit and reuse previously created targeting objects.
Estimations were inconsistent and too general.
Inconsistent knowledge of specific options.
Engineering stakeholder interviews
We started with our front-end developers and worked our way into the groups who owned the back-end services.
Takeaways:
Engineering leads could speak to individual services but less to how they all combined to produce a final output.
There was a desire to sunset some options to simplify the underpinnings of targeting.
There was agreement that some concepts presented on the front end did not match how the backend handled them.
Problem spotting
Disconnect with reality
Even tenured users' interpretations of capabilities did not align with the configurations they would use for setup, resulting in issues when post-campaign reporting failed to support the pre-sales agreements.
Incomplete representation
The back-end model was insufficient, and the front-end was trying to combine different functions into a single object. Users were unaware of the system's capabilities and what it produced.
No ability to reuse
The pre-sales and campaign stages use different tools and involve different user groups with no method to transfer a targeting profile from one to the other, creating redundant work.
Systems thinking
Breaking it down into inputs and outputs
I partnered with our lead front-end developer to map out how targeting gets configured. This arduous task is what helped us understand all the pieces that needed to be considered and exposed opportunities for improvement.
Making the targeting object whole
We thought if the targeting object held all targeting configurations, then it could become a single point of truth instead of some things being set in different places.
I felt confident we could rework the UI to expose all options, giving the user a clear and honest representation of how targeting worked.
Ideas
Confirming the problems
The targeting object was incomplete, with only basic inputs.
Advanced inputs are only in legacy tools or must be requested by filing a Jira ticket.
Limited access to previously created targeting objects.
The line item's relationship to the outputs was stronger than the relationship to the targeting profile.
The concept we socialized
Expand the targeting object to represent all targeting configurations.
Each tool represents targeting as a whole.
Allow access to previously created targeting objects in all tools.
Strengthen the relationship between the line item and the targeting object.
Results
Things we shipped
We successfully shipped iterations that unified the UI for setting up targeting in the pre-sales and execution tools. Additionally, we added features that enabled users to utilize existing targeting profiles in each tool. These changes reduced the time campaign managers spent setting up sold campaigns by 20%.
Future considerations
Engineering groups argued against creating a monolithic targeting object but praised the concept of modular and reusable objects.
Create several smaller, more focused objects
These objects can be defined and applied in various combinations
This approach offers more flexibility and reusability
This effort also resulted in creating and filling product roles tasked with coordinating the large scale collective efforts of engineering groups like overhauling targeting.
Looking back
I've never learned more
This extensive project dove deep to find the root causes of user issues that were surfacing in many of our enterprise tools.
It employed various UX research methods and required challenging conversations with most operational roles. I collaborated with our lead on the front-end team, initiated cross-functional team meetings, and had to persuade product owners to buy in on the substantial problems we uncovered.
If I had to do it over, I would have worked harder to get engineering groups to acknowledge the layers of problems we were uncovering to build a better case for a complete overhaul. They helped me understand the back end, and we received a lot of head nods when we first started discussing solutions. However, they recognized the topic's immense scope and were reluctant to get involved any more than they had to.
Users Aid in AI Advancements
Managing the frustrations of users enduring the lack of tooling to fast-track a feature to use natural language processing
Providing a Vision
Unifying enterprise tools into a self-serve client-facing platform