Building the UX practice

I joined WWT in St. Louis in February 2018 as a senior UX designer. I left late in 2018, was VP of UX Design at MediaCross, a St. Louis agency, in 2019, and came back to WWT in August 2019. On the map-data checker for a federal agency, I set up the team's DesignOps (design operations: the routines and tools that let a design team work to one standard every week). We held a weekly demo with research every two weeks, and built from a shared component library.

That was lead-level work with no lead title to go with it. In July 2020 I wrote the proposal for a Lead UX Designer role: draft 4 of the role description, dated July 10, set out four levels, Associate, Professional, Senior Professional and Lead, with the Lead mentoring the other three and running Lean UX workshops. The proposal was approved, and I was promoted into the role effective August 1, 2020.

In July 2020 I moved to Phoenix. I'd been asked to help open a new WWT office there, and I was one of the company's values interviewers, the people who talk with candidates about how they work with others. The pandemic changed that plan.

Instead, I adapted the map-data checker's DesignOps habits for new engineering teams in Phoenix, a group we called the Learning Team. My notes from its planning meeting are dated June 4, 2020. The teams built a small practice product to learn on, GO Fish, an ordering app for a sushi restaurant. It was a spike, a short project to learn what a kind of work takes before a client asks for it, and it prepared the team for fast-casual dining clients. The team didn't have a product owner yet, so I led those spike projects and mentored its four engineers through them. On September 8, 2020 we tested two versions of its prototype in ten recorded sessions on UserTesting, a remote testing service.

I didn't run it well, and an organizational change reshaped the group and then ended it. If I did it again, I'd start with one habit on one team, the weekly demo, and name who owns it, so it could outlast a change in the group around it.

Much of the practice grew around what clients asked us for. In May 2020 I coached a room in a Lean UX workshop for an electric utility, supporting a facilitator in training. In July 2020 we pitched DesignOps to an automotive logistics company, in a four-slide deck on what design operations is and the questions it answers, like how to define and measure design quality. That pitch was a forerunner of the Design Ops assessment, one of the six services in the catalog a year later.

The requests kept coming. When we ran the Lean UX workshop on our own sales team in 2021, the report opened with them: for a year and a half, sellers had come to the UX practice for slide decks, mock-ups and visualizations to help win custom software work, more often than a small bench of designers could easily answer.

WWT's UX practice already had a career framework: a workbook a colleague started in September 2019 that rates every role, from Apprentice to Lead Professional, on eight skills, each on a five-step scale from Developing to Shaping. What it didn't have was words. A cell said "Leading" and stopped there.

In the fall of 2020 I wrote two of the eight skills out in full. For each step of the scale I said what a person at that step actually does. Human Factors, dated October 15, follows the same five ways people take in information at every step, from noticing things to making decisions, and asks more of each at every level. Accessibility and Inclusivity, dated October 22, runs from knowing the guidelines (WCAG, the Web Content Accessibility Guidelines) to owning them on a project and including people with disabilities in testing.

That fall, WWT also compared usability-research platforms through a request for proposals: 5 suppliers bid by November 13, 2020, to be graded on a worksheet of 11 research methods and services across web, mobile, wearables and desktop. I wrote the comparison's UX content and its process documentation. Which platform it led to wasn't recorded.

Twelve people were promoted through the framework, me among them.

In June 2021, while the requests for help with pitches kept coming, I wrote the UX Practice Overview, two pages that tell a client what the practice does. Its six services came from UX Discovery Solutions, a deck dated August 19, 2020 that a team I was part of made. Each service opens with what the client is trying to do, in their words ("I have a product idea, and I want to make sure I build the right thing"), names the two roles who staff it, and says how long it runs.

Each one also says what tends to sink it. The deck lists the biggest obstacles for four of the services, and the same ones keep coming back: not enough time, no clear view of the real users, and research nobody has done yet. Read the other way, those are what a request takes for granted, so in 2026 I wrote each obstacle down as an assumption and named what the service tests first.

Its last section ties the services back to the ladder. Professional, Senior and Lead each come with the skills they bring, and the Lead is the one who runs workshops, leads discovery and speaks for the practice in sales. It's the one document where the ladder, the skills and the services meet.

Six weeks later, its Lean UX workshop was in a pitch.

In the second half of 2021, our pitches to six clients, one in each of six industries, were built on the catalog's services. Pitches to a beauty company in July and again in September, a digital health company in August and an automaker in September offered the Lean UX workshop, each time as a 4 to 6 week effort that opens on the client's words: "I have an idea, and I want to make sure I build the right thing." A retailer's pitch in December offered the same workshop. A plan for a university, drafted from September to December, paired a smaller version of it with a 12-participant usability lab over 12 weeks. And the design questions we wrote that fall for a state government's request for proposals asked about the product's users, their context of use and what they really need, the words the catalog uses for human-centered design.

The ladder went into staffing. A bank's planning engagement, which started on October 25, 2021, was staffed with a UX Mentor, the catalog's title for its Lead role. Four days later the bank told us it was changing direction, and by November 11 it had chosen to buy a system instead of building one. The work went on hold. That same October, a worksheet we used to plan combined offerings with the Product Ownership practice listed all six services and the whole ladder of roles, to find where the two practices overlap in pre-sales.

In 2021 I was one of the UX practice's representatives in a working group with WWT's Product Ownership practice, planning services the two could offer together. Our roadmap ran from May to November 2021 in four steps, from agreeing on each role to working together during delivery.

By September 2, 2021 we had finished 2 items, a kickoff and a shared description of each role, and the group had set aside working together during delivery for the rest of the year. It stalled late in 2021, and nothing from it is dated 2022.

What the two practices already shared shows in the Product Ownership practice’s playbook, a workbook a team I was part of kept from April 2020 to March 2021. It sets a product owner ten goals for a project, from understanding the business to forecasting. On two of them, understanding the user and fitting the product to the user, it tells the product owner to work with UX, and three of its tools were already in our Lean UX workshop.

The Lean UX workshop then turned into two engagements we won. I led both, and each was in the hundreds of thousands of dollars.

The first was for an insurer. We ran the 5-day workshop with the team behind one of its sales platforms in December 2021, and the book of what the room made is dated January 21, 2022, with my name first of its three authors. By April we had proposed two more phases to build what the workshop found, a web app for coaching the insurer's salespeople, with their reports shown simply.

The second was for the electric utility whose 2020 workshop I had coached. This time I led it, and its final book is dated July 28, 2022.

The workshop we sold had a track record before the catalog. It treats every design idea as a guess until it's tested, and our version was packaged in an offering sheet dated October 1, 2019: five activities in a fixed order, with the first one done before anyone's in the room. The standard workshop ran 5 days.

We first delivered it for a federal agency in October 2019, then for the electric utility in 2020. The insurer's workshop at the end of 2021 and the utility's second, in July 2022, are the two we won.

I wrote the room scripts and content for the whole Lean UX workshop: the words a facilitator says to open and run each activity across its 5 days. The scripts below are how I run the workshop today. In the 2019 version, breakout rooms drew proto-personas that we merged at the end, and Day 4 was a Design Studio, where everyone sketched. Today the Jobs to be Done framework, which describes the job a customer is trying to get done instead of who they are, replaces the proto-personas, and the sketching serves a plan for testing.

Every example below follows one company I made up for this page: a regional field-service business that sends heating and plumbing technicians to people's homes and books the visits by phone. None of it comes from a client.

Goal: one paragraph the product owner owns, saying what's broken and how we'll know it's better. Time: two 45-minute calls a week ahead, then 20 minutes on Day 1. Materials: the planning guide's four-sentence template and the Problems box of the four-box canvas.

I open with: "You know this product better than anyone who'll be in the room. I'm going to keep asking how you know, and you're allowed to say you don't."

Steps: in the first call (45 minutes), the product owner fills in the four sentences: what the product does today and for whom, what's going wrong, what it costs, and the measure that would show it's fixed. Before the second call, they check each sentence with one person who does the work. In the second call (45 minutes), we cut anything that names a solution. On Day 1 (20 minutes), they read it to the room, which asks questions but doesn't edit.

The room coach watches for a solution hiding in the problem ("we need an app") and a measure nobody can count yet. It hands the next activity a problem the whole room has heard in one voice.

Example, invented for this page: "Our dispatchers book home repair visits by phone across three counties. Customers wait at home through four-hour windows, and too many visits end with nobody there. Each missed visit costs a technician an hour of driving. We'll know it's better when fewer visits need a second trip."

Goal: get what everyone believes onto the wall, then find the beliefs that would hurt most if they're wrong. Time: the 160 minutes after the problem statement. Materials: sticky notes in two colors, one for the business and one for users, and a large grid that runs from "we know this" to "we're guessing" one way and from harmless to costly the other.

I open with: "Everything you write today is a guess, including the good ones. One belief per note. Write fast and save the arguing for later."

Steps: 20 minutes of silent writing about the business (who pays, and for what), then 20 about users (who they are and where they get stuck). 40 minutes to cluster the notes and name each group, a 15-minute break, then 45 minutes placing every group on the grid. In the last 20 minutes everyone gets three dots for the costly, guessing corner, and I read the winners back.

The room coach watches for the loudest person writing for everyone, and for beliefs stated as facts ("customers hate calling"), which get the question "how do we know?" The riskiest five or so assumptions go to Day 2, where the room ties them to the job.

Example, invented for this page: a business belief, "homeowners will pick their own arrival window if we offer it online", and a user belief, "dispatchers spend most mornings rebooking missed visits". The room votes the riskiest: homeowners will trust a window enough to stay home for it.

Goal: name the job customers need done, the business result that would show it's done better, and what users need along the way, each tied to the assumptions ranked on Day 1. Time: Day 2, 180 minutes. Materials: the ranked assumptions, a job statement template ("When [situation], I want to [goal], so I can [outcome]"), a sheet for the objective and its key result, and a long strip of paper for the journey.

I open with: "Nobody wants a repair visit. They want the heat back on tonight. Today we write down the job people are getting done, and pin our guesses to it."

Steps: 20 minutes writing job statements in pairs, then 20 to agree on the main job, which becomes the objective. 25 minutes on the key result, a business outcome the room can measure, and 15 listing the user needs each assumption touches. A 15-minute break. Then 45 minutes mapping the job as a journey, from noticing the problem to having it fixed: each step of the process, and the people and things in it (domain modeling). 25 minutes placing each prioritized assumption at the moment in the journey where it would show, and 15 checking that every assumption has a moment.

The room coach watches for a job that's really a feature ("book online") and for a journey that only shows the screens we already have. The job, its key result and the journey, with the assumptions pinned to their moments, go to Day 3.

Example, invented for this page: the job is "When my heat goes out on a workday, I want a technician to come while I'm home, so I can get it fixed without losing a shift." The key result: a quarter fewer second trips in one county. Luis, a homeowner who works shifts, needs a window he can plan around, and Dana, the dispatcher, needs to know by evening who won't be home. The riskiest assumption, that homeowners will trust a window enough to stay home for it, sits at the moment Luis decides whether to trade a shift.

Goal: turn each prioritized assumption into a hypothesis written as a test, with a clear measure, and rank them. Time: Day 3, 180 minutes. Materials: the journey with its assumptions, a grid of value against risk, and one template: "We believe [this change] for [these users] at [this moment] will achieve [this outcome]. We'll know we're right when we see [this measure]."

I open with: "A hypothesis is a bet you can lose. If you can't say what you'd see if you're wrong, it isn't finished."

Steps: 20 minutes walking the journey and reading each assumption back at its moment, then 30 in pairs naming a change that would test each one. 45 minutes writing each one in the template with its measure, a 15-minute break, and 40 minutes placing them on the grid. The room spends 20 minutes picking the top three to test and 10 checking that each measure can be counted.

The room coach watches for measures that are really tasks ("launch the portal"), and for a hypothesis that has drifted from the assumption it was meant to test. The ranked list goes to Day 4's test planning, top three first.

Example, invented for this page: "We believe a text the evening before, letting Luis keep or move his window, will cut second trips. We'll know we're right when missed visits in one county fall for two weeks running."

Goal: plan a cheap, quick test for each of the top three hypotheses, and leave with sketches and acceptance criteria for any prototype a test needs. Time: Day 4, 180 minutes. Materials: the top three hypotheses, a sprint board, 6-up sheets (six boxes to a page), thick markers and a timer.

I open with: "The cheapest test that could prove us wrong wins. If a phone call or a paper mock-up can teach us the answer, we don't build anything."

Steps: 45 minutes of sprint planning, where for each hypothesis the team lists ways to test it and picks the cheapest one that would still move the measure. 15 minutes to name what each test needs a prototype for. 30 minutes of sketching alone, six ideas to a sheet, then a 15-minute break. 30 minutes for each person to show their sheet and for the team to merge the best ideas into one sketch per test. The last 45 minutes go to acceptance criteria: plain sentences anyone can check that say when each prototype is ready to put in front of people.

The room coach watches for a test that quietly turns into a build ("we'll just code it"), and for people who don't draw, who get handed a marker. The merged sketches and their acceptance criteria go to Day 5, where they become storyboards for the readout.

Example, invented for this page: to test the evening text, the team builds nothing. For two weeks, Dana sends the texts by hand from a shared phone to customers in one county. The merged sketch shows the text with Keep and Move replies beside Dana's board, where unconfirmed visits turn amber. One acceptance criterion: by 8 p.m., Dana can see which customers haven't answered.

In 2021 I led the same workshop for WWT's own sales team, to show what we were selling on a problem of our own: why was selling custom software teams so hard? We interviewed five sellers in August, for 45 minutes each, and sent a follow-up survey to 21 people in September, which 3 answered. In the room we turned what we heard into a proto-persona, a storyboard and hypotheses.

The 34-page report I co-wrote with two colleagues proposed a pitch-tailoring team: designers who help each seller shape a pitch to its client, with a library of case studies and slides to build from. Its measure of success was plain. Sellers would win more business, and would want to keep paying for the team.

In May 2021 the practice set itself the goal of measuring and explaining the value UX brings, and I took it on. I wrote an interview guide of about 55 questions in four parts, screened six practitioners and interviewed five between June 21 and August 10, 2021. The same month the interviews began, I also took a UX course from the Nielsen Norman Group.

Four things kept coming back. Nobody could point to a written standard for good design work. The client owned the product's numbers, and the goals set at kickoff were rarely looked at again. When budgets got tight, research and design went first. And when a project went well, the story was rarely told. We followed the interviews with a wider survey, and 98% of the people we sent it to answered. Then we mapped our assumptions and held one-on-one follow-ups before finishing a report to leadership, which set the initiatives for the following quarter.

From August 2022, after the utility's workshop in July, I spent two years, to 2024, on a DesignOps engagement for an enterprise networking company, where I ran Lean UX and field-tested it on the portal its customers use to run their networks. What we delivered is in that portal's case study. Along the way, I trained a designer on the team in the same methods.

A network portal for an enterprise networking company→

We won two Lean UX engagements, for an insurer and an electric utility, and I led both. The insurer's led to proposals for two more phases. Twelve people were promoted through the career framework, me among them; mine was to Lead UX Designer effective August 1, 2020, in the role I had proposed. In Phoenix, I mentored the four engineers on the Learning Team through spike projects I led before the team had a product owner, preparing for fast-casual dining clients. The 2021 research ended in a report to leadership that set the initiatives for the following quarter. From August 2022 to 2024, I ran and field-tested Lean UX for two years on an enterprise networking company's network portal, in a DesignOps engagement, and trained a designer on the team in it.

Some of it was never measured. I can't say which of the other pitches turned into work, and I have no measure of what any client did with a workshop after we left.

What a discovery engagement set in motion is easier to show than what it sold. A team I was part of drew how engagements would run once discovery ended: about 14 people across ten roles, staying near that size for eight months or more while the work moved toward the data. The plan below is redrawn from that drawing, simplified and with no client named.

I'd ask before I wrote. The 2021 interviews came after the ladder and the catalog, and they should have come first, so each document answered something practitioners had told me.

I'd also keep a list of every pitch and how it ended, and give each service one number to track from its first sale. Today I can point to the two engagements its workshop won, and not to what the rest of the catalog sold.