💰 Cut support costs by $7 million
This new product was getting leads, but struggling with Sales due to its install and configuration experiences.
When paired with other team-members' work, our efforts pushed us from two customers to over forty in six months.
My work on this project directly led to a $7 milion reduction in support costs.
Role
Research, project management, UX, & stakeholder management
Market
Containers-based infrastructure
Allies
BJ Scheid, Brenton Elmore, Danielle Justilien, Hannah Parton, Katy Hagen, Ken Fisher, Stephen Webster
Duration
~6 months (February-August 2022)
Problem
⚠️ This case study's work is under NDA. Identifying information and IP have been redacted.
Have you ever joined a team under a set of expectations that, turns out, needs to be replaced ASAP?
In late 2021, I believed I was joining a new product team to lead its new-born User Research practices, starting off with normal usability benchmarking.
Said product only had two customers, and since it had released in November, this seemed un-noteworthy. After those benchmarking studies began, however, those two customers seemed... foreshadowing.
Discover & define
Discovery work began during Benchmarking, though technically that wasn't supposed to be happening.
My original asks were to run basic usability testing on dashboard redesigns, even though there wasn't any data on the product's released dashboard.
So I ran the requested study, but included comprehension questions. What did our participants think this new product could do? Why? How? What comes to mind when they compare their interpretations against their day-to-day work?
The result was clear and immediate: these people had no idea what this product was or what value they might get from it.
Perhaps I recruited from the wrong set of people, I wondered, but their backgrounds matched those gathered from market surveys run before I joined the team. Which, turned out, were based on user personas... created in 2017.
As my boss Brenton Elmore put it after sitting in on one of those interviews, "This product was created for a user base that doesn't exist anymore."
Ominous data
> Synthesis and Analysis Board from the Dashboard research project.
In and around the "Story 2" section, you'll find answers to those open-ended questions. The first signs that something was wonky in the bottom right hand corner, where one category reads "Clarify the relationship between [this product's fundamental functionality]."
Many thanks to Hannah Parton, Ken fisher, Danielle Justilien, and Brenton Elmore for their collaboration during this work!
Results
At the beginning of 2022, it cost ~$7 million to fly Support engineers (11-15) to a customer location to help them set up, install, and configure this product in their environment. This usually took 2-4 months to complete.
The first round of changes from this project released in 2022 Q3. By December, install and configuration support was taking place in ~2-3 weeks online, keeping that ~$7 million additional expense per customer in company coffers.
Sales leads quickly started yielding a higher conversion rate following "Try before you buy," periods. Revenue jumped by ~62.5% by the end of 2023.
Apologies, but changes to profit margins cannot be disclosed.
Develop & deliver
Once we course corrected on users and buyers, the fun stuff began (and by fun, I mean complex).
Structural changes to existing offerings are not cheap, and some stakeholders sought redemption/control/general-anxiety-reduction by insisting on a replacement structure of their own design.
To respect and validate their efforts, I included these options in a comparative tree-jack studies, along with the original, control structure.
Which... ended with the control performing better than said designs. It took repeated deep-dives into data analysis with said stakeholders to show how user needs and wants were detectable in the results of that study when compared to previous results.
Incrementally, we shifted those designs closer and closer to users' preferred structure.
We followed up with in-person tree-jack studies to gather qualitative data and provide direct observation to hold-outs. These sessions both convinced the whole team to accept that alternative structure, and provided prioritization data on which structural changes to implement sooner vs later.
Several follow up rounds of higher fidelity navigation and use testing later (including core CRUD+PR workflows), 6 months had passed. We packed up the new version and released it.
By month end, one customer had updated to the new version and had no support requests, while the other requested another in-person assist.
Test options
> Facilitation work space.
Changes to information architecture and resulting navigation are quite expensive to implement; if you're going to do it, you want to do it right.
I facilitated sessions with 5 stakeholders to create multiple IA options to test with, as well as what tasks to ask participants to locate.
