From opinion
to evidence
Building a system for continuous, direct customer access in order to get close to customers, replacing internal assumptions with research at scale.
Overview
Customers were out of reach.
When I joined AWS Training & Certification as the founding UX research hire, there was no research infrastructure. No panels. No recruitment pipeline. No tooling. No shared understanding of what customer research looked like in practice.
Product decisions were shaped by internal opinion. What the team thought, what leadership assumed, what felt right based on experience.
My job was to change that. Not by running a few studies and presenting findings, but by building the machinery that would make continuous, disciplined customer research possible at scale and training an organization to use it. The metrics below reflect cumulative impact across five years of operation.
Problem
Research didn't exist as an operational capability.
The organization didn't have any research infrastructure or a shared language for what customer evidence looked like. Three gaps made the status quo unsustainable.
Teams had no way to recruit customers. When input was needed, sales might suggest a contact. Most decisions moved forward without customer input.
A study with six participants took roughly 50 hours. Research couldn't keep up with the volume of product decisions.
80% of the UX team had never conducted research. There was no shared methodology, and findings were anecdotal and inconsistent.
I didn't build everything at once. Early wins, such as a small panel and basic recruitment pipeline, established credibility and created demand. The training program, tooling, and governance model followed as the organization's appetite for research grew.
Solution
So, I built the machinery.
The infrastructure came together in phases, each one unlocking the next.
Together they formed a single, self-reinforcing system.
I built three segmented customer panels from scratch: internal employees for usability testing, external customers for primary research, and external non-customers for comparison. Recruitment came first. Without a panel, nothing else was possible.
Outcome
750K+ opt-ins, 70% take rate, one-day average recruitment.
I selected, onboarded, and governed a third-party research tool, built self-service templates and an intake prioritization rubric so product managers could run basic evaluative research independently, and established a customer listening program for continuous access to customer voice.
Outcome
Average study time cut from 50 hours to 12.
I designed and scaled a tiered training program: literacy sessions for sales, methodology workshops for UX practitioners, and self-paced modules for the broader organization. Training ran in parallel once teams had something to research against.
Outcome
24,000+ employees trained across 52 countries, 80% completion.
Results
And customer input drove more and more decisions.
0
increase in studies
0
less time on studies
0
panel opt-ins
0
employees trained
0
listening sessions
The cumulative result was behavioral.
The way the organization made decisions changed. Research was no longer something that happened occasionally when someone could arrange a customer conversation. It became an operational capability, always available, consistently structured, and connected to the teams who needed to act on what it produced.
Reflection
Success outpaced the tooling.
The original tooling required SQL knowledge to access the participant database, which created bottlenecks. The panel grew faster than anticipated, outpacing early engagement mechanisms.
Over time the ecosystem was streamlined from 20 tools to 8, the database was migrated to a CMS to broaden access, and AI-assisted workflows were introduced to help teams synthesize and summarize sessions at scale. Each iteration made the infrastructure more self-sufficient and less dependent on the research team to operate it.
And bigger things became possible.
With panels, training, and operational infrastructure in place, the organization could execute research that had previously been out of reach.
The study detailed in the next chapter, from evidence to action, is a direct product of this foundation. Without a recruitment pipeline, a trained organization, and a culture that had learned to expect and act on customer evidence, a study of that ambition and duration would not have been possible.