Order Flow Redesign

How might we rethink our order flow to make booking a move online effortless and accurate?


About the redesign

Our online booking process was an à la carte model which put most of the onus on customers to estimate their move. It was up to a user to figure out how many movers they needed and for how long.

As a tech-enabled moving company, we knew we could do this in a smarter way by asking the right questions upfront and learning from data over time.

Constraints: simultaneous redesign of visual system/pattern library


  • increase conversions for online bookings (in relation to call-in orders booked by phone)
  • create an order flow to accurately scope a move without requiring customers to input extraneous details of their residences and inventory
  • utilize historical data to anticipate and recommend how much time and labor were appropriate for the job
  • make booking a move online effortless and delightful



After gathering input from the team and conducting initial customer surveys, I outlined our collective assumptions around why users might abandon the current order flow—documenting how the redesign should address each concern.

Move time estimations

In order to give customers the best move configuration (headcount x duration) recommendations, I first made a reference sheet of options to see what happens when each move exceeds its allotted time. While we had a simple per-man-hour pricing model, there were stiffer monetary repercussions for customers, the higher the headcount. We also needed to consider the operational implications of placing more headcount on a job, thus expediting move duration and vice versa.

Machine learning

I partnered with data science to explore how the new order flow might be instrumented for feeding their early-stage machine learning data model. We also ran analyses on customer satisfaction (NPS) and concession (refunds) data against move time accuracy to understand the correlation in missed expectation-setting.


Using the $1 Prototyping method, I worked quickly to sketch initial flow variations, mapping them out on the walls of our office. I invited engineering, marketing and customer support teams to visit the walls at their convenience and provide input in between formal design reviews.

Explorations for entering inventory using a glyph-based alphabet as a delightful-spin on the most tedious part of booking a move—listing everything in your home:

Usability testing

We partnered with a newly-constructed apartment community to host an event where new residents could participate in testing prototypes and followed by a 15 minute interview about their recent move in exchange for a gift card.

  • completed ~30 interviews and usability tests (most had not used Bellhops for recent move)
  • gathered a better understanding their mental model and decision-making process for selecting a service provider (needs/motivations)
  • better insight into how big of a competitor DIY moving was with this demographic
  • recorded and internally published video/audio of all interviews and testing sessions for anyone in the company to access
  • summarized the findings and circulated across teams

Finalized design for MVP

Responsive hi-fidelity designs were prepared for each major breakpoint from 320px mobile to 1440px width desktop. A Sketch–Invision–Zeplin workflow was used for a smooth handoff to engineering.

I worked simultaneously with the frontend team to ensure assets, typography styles, and color palettes used were saved as reusable components and compatible with the new pattern library.


  • worked closely with PM and Product Director to prepare a detailed technical spec describing logic required for proposed order flow calculations: TRAVEL DISTANCE; TIME CALCULATION; INVENTORY CALCULATION; LABOR CALCULATION; AVG. COMPLETION TIME CALCULATION; APPOINTMENT CAP CALCULATION

Monitored key metrics

  • conversion funnel
  • connected partial-sessions with call-in order completions by using a unique identifier that could be verbally relayed over the phone and tracked back to online drop-off
  • used FullStory to analyze user sessions and for initial bug reporting

Ran experiments

  • funnel-based A/B tests by switching and eliminating steps in the order flow
  • continuously revisited the accuracy of our pre-calculated estimations to adjust/fine-tune



  • established pre-calculated moves as recommendations in the order flow which gave us an advantage over the competition
  • findings around factors (other than inventory) that contribute most to completion time
  • initiated and implemented operational process changes involving workforce training and protocol