Designing Get Started to work at any scale

Redesigning DoorLoop’s Get Started flow so that a customer with 10 doors and a customer with 300 can both import their portfolio in one sitting, readable, editable, and accessible at any size, without a specialist doing it for them.

Role

Product Designer

Timeline

2 months

Platform

Web app

Interactive prototype

Overview

DoorLoop is a property management platform. Before a new customer can collect a single rent payment, they have to get their whole portfolio into the system: every property, every unit, every tenant and lease, every owner, plus banking details so DoorLoop can charge on their behalf.

Get Started is the flow that carries that. It’s the first working hour a paying customer spends in the product, and in 2026 it was the thing making them regret signing. I designed the data import half of it: properties and units, tenants and leases, and owners.

How Get Started works

Rather than adding properties one at a time into the live platform, Get Started collects everything in a setup mode first, a sandbox that doesn’t touch the account until the customer presses launch. Then it all goes live at once.

Onboarding specialists sit with customers through this, usually on screen shares, so a specialist is the first person to hear when something breaks.

Research

The most obvious indicator was already within the company. Onboarding specialists handle this process daily with actual customers, and multiple specialists reported the same recurring failures.

I joined a session between our product manager and Sebastian, one of the specialists handling the most Get Started accounts, to hear these failures firsthand. He walked us through one account in detail.

Account sold as 13 units; actual count was 25–27.

Sales had counted her buildings, not her doors. Everything downstream inherited that error.

• Get Started capped her at 13 units and blocked further progress.

• Upgrading her subscription didn’t lift the cap; she was still blocked at 13 units.

• Her mixed portfolio showed only residential options, so she created office properties outside the intended workflow.

• A specialist fixed her data, but the changes reverted before the next meeting.

• The confirmation screen showed three tenants, even though she had entered 27 leases.

• After launch, duplicate properties appeared and leases split, all because data was entered outside the workflow. Tenants had already been invited to pay against these records.

These issues were common. Other specialists reported the same caps, reverted data, and incorrect confirmation numbers. Recordings showed customers repeatedly encountering the same problems.

“I literally had to say, just click that. Don’t pay attention.” An onboarding specialist had to tell a paying customer to ignore the product’s confirmation screen.

This made me realize the confirmations weren’t just inaccurate—they were teaching customers not to trust the product right when we were asking for their portfolio.

The problem

Tenants and Leases is where customers stop, and nobody gets through it alone.

The pattern, from Sebastian’s account and confirmed by other specialists and the recordings we reviewed:

• Nobody breezes through. Four or five accounts on the new flow, every one needed help.

• Sessions don’t progress. On the old flow he could leave a customer to finish a few properties alone. Now he comes back to find them exactly where he left them, unable to proceed.

• Accidental typing traps you. Start filling a unit, change your mind, and the form demands a rent amount. Clearing it needs a three dot menu on the unit line that customers don’t find.

• The table is too dense to read. Hidden dropdowns, no clear line between property and unit. One customer added the same tenant twice because she couldn’t tell the units sat in different properties.

• Warnings without recourse. The flow says what’s missing but not how to resolve it, and past ten leases, sorting through them is its own task.

Specialists were doing the customers’ data entry for them, which is exactly what the flow existed to avoid.

Why it mattered

• This is the first hour of work inside a product the customer has already paid for.

• Friction here reads as a broken product, not a rough setup.

• One account in this state was close to canceling, and a specialist spent a call talking her down.

• Every account that needs hand-holding caps how many customers onboarding can take on

“You just bought a car, you take it out, and it stops, and they say, don’t worry, we’ll put a new battery in. You have a very bad impression right off the gate.”

Goals

User goal

Easily import an entire property portfolio into DoorLoop in a single session, with confidence that the information on the screen matches what will actually happen.

Business goal

Reduce the number of specialist hours required per account and prevent customer drop-off during their first week.

Scalability Goal

Enable the platform to support customers with very large portfolios, including those with hundreds of units, ensuring the process is practical and efficient, not just technically possible.

Design challenge

Create a screen that can display hundreds of unit rows in a way that remains readable for careful review, while still being intuitive and approachable for new users.

Key Decisions

Make rows the focus, not dropdowns.

Replaced the dense, collapsed table with a spaced, line-by-line view similar to existing leases and tenants screens. Now, each row clearly shows property, unit, tenant, and rent for quick review.

Centralize error handling.

Scattered warnings are gone. A persistent counter now shows how many fields need attention and guides users to each issue, so nothing gets missed.

Support large portfolios.

Typing works for small accounts, but not for thousands of units. Users can now upload existing data files. An AI step scans and autofills properties, units, tenants, leases, and owners. Large portfolio customers review and correct instead of entering every row manually.

Unified error system for all data sources.

Errors from both AI uploads and manual entries are handled in the same way. All issues are surfaced in a single system and fixed through the same counter, streamlining the experience for everyone.

The design

Built as a working prototype rather than static screens, so the table behavior, the bulk actions, and the error stepper could be tried rather than described.

Interactive prototype

Properties & units

The full step, with bulk add, inline editing, the error stepper and file import

Properties and units flow: Add property.

Rename bulk action.

Missing fields flow.

Tenants and leases step: quick view side menu.

Design system contribution

Scaling this flow to hundreds of entities required components that the platform didn’t have. Four came directly out of this project:

• Editable lists, for the row-level editing the new table relies on.

• Anchored popovers, replacing the hidden three-dot menus customers couldn’t find.

• Wizard stepper, the shared progress pattern now used across every data step.

• Mistakes navigator, the persistent error counter that steps through incomplete fields.

All four were built to be reused wherever the platform needs a similar pattern, not just inside Get Started.