Shipped B2C product

A/B Testing

Design strategy

Pricing

*Some details changed for privacy

Turning the junk removal pricing problem into a system customers trust.

Card 2
Card 3
Card 4

TLDR;

01

I redesigned how customers size a junk-removal job, and paid-job conversion rose 18%

I led design and research, from four research studies to three live A/B tests. After full launch, cancellations fell 13%, and the results led the business to rethink how it sets prices.

02

Getting pricing wrong had already cost us bookings in the past

Price transparency was our top complaint, but the last time the website showed prices, bookings dropped and the test was switched off within days. Leaving it alone wasn’t an option either: customers were still deciding when the crew arrived, and many booked jobs were being cancelled.

03

Customers couldn’t picture truck space with eight increments

We price by volume, so every price range looked arbitrary. In my research, customers added up single-item prices, which our pricing doesn’t do, and relied on the free on-site estimate to decide.

04

Three simplified load sizes beat eight

Leadership wanted to show all eight truck sizes, the way the company had always priced. In testing, customers understood small, medium, and large far more easily, so I designed those sizes with an animated truck and everyday job descriptions to help them picture their own load.

05

The test gave the business an uncomfortable answer, and now they’re actioning it

My design lifted conversion with or without prices, but prices made people stop reading. The design was working, and the price ranges were what still held some customers back. The finding led the company to rethink how it sets prices, now its own initiative.

Details

Role

Product Designer Lead Designer, Residential Conversion

Platform

Responsive Web · iOS · Android

Timeline

Sept 2025 - Feb 2026

Status

Launched system-wide 20% measurement holdback

Team

1 PM · 1 Tech Lead · 3 SWE · 1 QA

Toolbox

Figma

Google Gemini

Case study

01 | The problem

Customers were asked to price something they couldn’t picture.

The junk removal industry charges by truck space and no one thinks about their items relative to truck space.

Jobs with several items are priced by volume: how much of the truck your things fill, in eighths. The crew confirms the exact price on site, for free, with no obligation. The online experience wasn’t asking anyone to buy online, but asking them to let us come look where the sale will happen in-person.

Competitors turned our pricing into a warning

Comparison content across the category told customers that 1-800-GOT-JUNK?’s volume pricing gets finalized on site, so people should expect a surprise quote coming. That framing was successfully working against us before customers ever reached our website due to the rise of AI LLMs.

The last attempt to show prices online backfired...

When I first joined the team, the signals pulled in opposite directions. An earlier test had shown single-item prices, like a flat rate for one sofa, and bookings dropped within days. Yet price transparency was the top complaint, and about 1 in 5 visitors clicked a placeholder "Get price" button for multiple items. Most customers have more than one item for junk removal. I couldn't simply show prices again, but as a user advocate, I couldn't keep leaving customers to guess. The challenge was that whatever I design had to make pricing clearer without repeating that booking drop.

02 | The research

Locating where customers started losing trust

The obvious fix failed first

The pricing tabs customers wanted were buried near the bottom of a long page, and their high click rate showed people were hunting for them. We moved them up. It failed. Explaining that we charge in eighths describes our billing. It doesn't answer "what will this cost me," so surfacing it sooner just meant people hit the same wall earlier.

Customers added up single-item prices, which our pricing doesn't do

In my first study, 40+ participants did a think-aloud test. Several added single-item prices together for a fridge and a mattress.

Nearly everyone was relying on the “free, no obligation” estimate to book

Almost everyone willing to book said they'd only do it because the on-site estimate was “free, no obligation”. The website was passing its trust problem to the crew.

Explaining what was included made prices feel fair

When participants read what a price included, such as in-home removal and sweeping, several moved from "too high" to "fair."

How might we help customers picture their load before they see a price, so they can trust the number when the crews arrives?

03 | My solution

I designed a way to self-size a job online before seeing a price.

The old page explained pricing halfway down a long scroll. My new design starts with how pricing works.

The old page explained pricing halfway down a long scroll. My new design starts with how pricing works.

Three sizes instead of eight.

Internally, leadership wanted to show all eight sizes of the truck online with the prices attached per increment because that's how the crew had always done it in-person. During my testing, customers understood three sizes far more easily: small (minimum to 1/4), medium (1/4 to 1/2), and large (1/2 to full). A customer doesn't need to predict which eighth they'll land on, but they need to expect which third of the range they're in, so the final number isn't a shock.

The billing increments the crew used in-person became three sizes customers can recognize.

The billing increments the crew used in-person became three sizes customers can recognize.

A truck customers can see their own load in.

The original flat truck taught the unit, but volume is 3D. I designed an isometric truck that fills to each load size with familiar objects, so customers can map their own pile onto it. I illustrated most of the items myself and built the animation in Figma Make with our front-end developer.

Everyday jobs help customers confirm the size.

Then I described each size with a familiar situation for junk removal, like before a move or completing garage clean-out. Customers recognize their situation instantly, then check it against the truck.

Decision trade-off: less precision in online price accuracy, but a ballpark price estimate customers could grasp at a glance is better than nothing.

Three sizes can't capture every job the way eighths can, and I noticed customers skimmed the truck more than they read the details. Transparently, I accepted both. A quick, confident sense of their load mattered more, so they could feel confident to take the next step into booking. This would allow the crew to give an exact estimate at the door with the customer expecting it.

Does my new sizing design build confidence on its own? Does prices help or hurt on top of it?

04 | Testing results

My design lifted conversion, regardless of prices

Lifting paid-job conversion by 18% and booking cancellations fell by 13%

Both versions of my design beat the existing experience, and the version without prices lifted paid-job conversion by 18%. After a pricing survey confirmed the ranges (a balance between “willingness to pay” and our franchise prices) were plausible, we ran a mobile pilot in 12 franchise territories. We found that it was still holding some customers back. In research, most participants said they preferred seeing prices first. In the live pilot, prices made people stop reading: 40% of visitors opened a detail panel with prices showing, compared with 71% without. Because of this, we found my design was building confidence in the way we had hoped, and the prices themselves were what some customers weren’t comfortable with.

Decision: I strongly recommended launching without prices, as a first step

We shipped the design with the three load sizes, and took the price ranges off for the meantime. This way it would allow more customers to reach the sizing content and feel more confident in booking while the crew confirms the exact price for free at the door. We launched on mobile and desktop on February 12, 2026, keeping 20% of traffic on the old experience as a holdout. Compared with the holdout, cancellations fell by 13% across about 130,000 visitors from March to April 2026. Customers who understand how the estimate works arrive with calibrated expectations, and calibrated expectations mean fewer cancellations. It left the top complaint unanswered, but this work was crucial in allowing upper leadership to truly reflect on the current pricing model and the flexibility each franchise out of 150 has with the pricing model.

05 | Influence beyond...

My work played a core role in why the company is now rethinking how it sets prices.

Our findings led the company to revisit how it sets prices, which is now a dedicated initiative co-owned by several other departments. We also showed the business that building customer confidence before booking works, and we shouldn’t be afraid to compromise setting customer expectations with hiding information to just see high numbers in bookings. The price ranges were a barrier for some customers, but in a low-UX maturity company, this was a win in growing overall user advocacy throughout the business.

A price range can only go so far… How might we give every customer an estimate for their own load if they could take photos?

Instead of matching their pile to a range, what if customers could take a photo and see an estimate for what they actually have? They wouldn’t need to translate their junk into truck sizes, and the price at the door would hold fewer surprises. I designed a 0→1 AI solution where crews can check and correct the AI’s volume estimates, so we can build accuracy for a customer-facing version.

As of late 2026, I’m the lead designer and researcher on allowing customers choose parts of the service for what they need, where price certainty, optionality, and scheduling flexibility matters most.

Following the work, I’m researching customer situations around removal jobs, and advocating towards giving every item its own price, so the service can be unbundled towards various job types based on a customer need - such as, moving out vs. replacing furniture.

Details

Role

Product Designer Lead Designer, Residential Conversion

Platform

Responsive Web · iOS · Android

Timeline

Sept 2025 - Feb 2026

Status

Launched system-wide 20% measurement holdback

Team

1 PM · 1 Tech Lead · 3 SWE · 1 QA

Toolbox

Figma

Google Gemini

Details

Role

Product Designer Lead Designer, Residential Conversion

Platform

Responsive Web · iOS · Android

Timeline

Sept 2025 - Feb 2026

Status

Launched system-wide 20% measurement holdback

Team

1 PM · 1 Tech Lead · 3 SWE · 1 QA

Toolbox

Figma

Google Gemini

Details

Role

Product Designer Lead Designer, Residential Conversion

Platform

Responsive Web · iOS · Android

Timeline

Sept 2025 - Feb 2026

Status

Launched system-wide 20% measurement holdback

Team

1 PM · 1 Tech Lead · 3 SWE · 1 QA

Toolbox

Figma

Google Gemini

© Cathy Yan 2026.

Fuelled by iced lattes and ambition.

© Cathy Yan 2026.

Fuelled by iced lattes and ambition.

© Cathy Yan 2026.

Fuelled by iced lattes and ambition.

This study is password-protected.

Please reach out to me if you’d like access.