Designing an intuitive way to set up cleaning routines
Roomba users can create routines directly from the home map instead of navigating lists and settings. By anchoring setup in the home itself, routine creation shifted from a manual task into an intuitive, spatial experience that prioritizes simplicity while surfacing helpful recommendations without taking control.
Status
Shipped 2025
Contribution
Principal Product Designer
Product
Roomba Home (Mobile App, iOS & Android)
Audience
Roomba Users, 2025+
Skills
UX Strategy & Research
UX Design & Prototyping
Design Leadership
Cross-functional Collaboration
PROJECT OVERVIEW
In 2025, iRobot launched a new robot lineup with advanced mapping and intelligence, requiring a full redesign of the Roomba Home app. I led the design of its flagship feature, Routine Builder, a map-first experience that lets people set up cleaning routines directly from their home map rather than navigating lists and settings.
The goal was to align the app’s core promise — a clean home — with how people think about cleaning. Routines needed to feel easy and intuitive to use, with recommendations available as quick starting points rather than imposed defaults.
I designed and shipped Routine Builder across two major releases, the MVP in March 2025 and a Quick Start refresh in August 2025. The initial launch established the core interaction, while real-world use drove refinements that improved usability and adoption.
CONTEXT
In 2025 my team was presented with a unique opportunity: a clean slate to design a new Roomba Home app entirely around modern robot capabilities without carrying forward legacy hardware or software constraints.
This allowed us to shift from a robot-centered model to a home-centered experience grounded in people’s spaces and cleaning habits. Rather than iterating on the classic app, we treated this as a fresh start, designing around homes, maps, and routines from the ground up.
Early research and competitive analysis pointed to a consistent behavior: people wanted to start cleaning from their map. To explore this in practice, I sketched early concepts focused on making that entry point feel simple, flexible, and aligned with how people already think about their homes.
KEY INSIGHTS
As we tested map-first concepts, a consistent pattern emerged: people didn’t want more control, they wanted the simplest path to a clean home.
01
Starting from the map was most intuitive
Though a challenge for engineering, selecting rooms directly on the map was how all participants expected to begin cleaning.
02
Choosing where to clean mattered most
Most people thought first about where to clean, then how. Very few showed interest in setting room order or cleaning preferences up front.
03
Profiles provided enough control
Of the people who engaged with preferences, most were satisfied with cleaning profiles like "deep clean" – only one explored advanced options.
04
Recommendations worked when optional
Recommendations resonated when they felt helpful and non-intrusive. Most reacted negatively to anything preselected or forced, preferring guidance they could choose to use.
One user captured the sentiment perfectly:
"Don't make me jump through hoops just to vacuum my kitchen."
I embedded Routine Builder directly into the Home tab, anchoring the experience around a single, interactive map. This made the map the starting point for creating and running routines.
From there, the map became the primary surface for choosing rooms, previewing what a routine would do, and making quick adjustments before starting to clean. Instead of moving through lists and settings, people could see and shape the outcome directly in their space.
This approach required close collaboration with engineering to ensure the experience felt responsive and fluid, but it allowed us to deliver an interaction model that matched how people naturally think about cleaning their homes.
DESIGNING A MAP-FIRST ROUTINE BUILDER
BALANCING SPEED WITH CONTROL
Quick Start routines helped people move quickly without losing a sense of control. Speed mattered, but only when people could still see, understand, and adjust what would happen next.
Rather than executing immediately, Quick Start options opened directly into a pre-populated Routine Builder. This let people start from familiar patterns while reviewing and adjusting their plan before cleaning begins.
To ship a responsive, map-first experience within scope, we focused the MVP on helping people start cleaning quickly and confidently, and deferred additional complexity until after launch.
-
The home map shipped as the primary entry point, making room selection and routine setup fully interactive from day one, even with significant technical lift.
-
Quick Start cards provided a fast path into familiar routines, surfacing recent and recommended options without introducing additional decision-making up front.
-
Default settings did most of the work, with routine details and advanced preferences available but secondary. Cleaning profiles tested well, but were deferred to prioritize map performance and responsiveness.
MVP EXPERIENCE
BEYOND MVP
The MVP validated the map-first approach, but real-world use surfaced two clear friction points: cleaning settings were still too complex, and Quick Start cards competed with the map for attention. Post-launch work focused on addressing both.
As robot capabilities expanded in 2025, we needed a scalable way to manage growing complexity. Cleaning profiles grouped multiple preferences into simple, goal-based options: Light, Standard, and Deep, along with Smart, an intelligent mode that adjusted behavior automatically based on dirt detection, cleaning history, and room type.
Testing showed people preferred applying a single profile to an entire routine rather than configuring rooms individually. We also learned that users thought about how to clean separately from the mode their device was running. For example, people saw a deep clean as something that could apply to vacuuming, mopping, or a combo routine.
Although profiles were initially scoped out of the MVP, alpha feedback made their value clear, and they became one of our first post-launch UX priorities.
Simplifying How to Clean
As the experience was stress-tested across homes of different sizes and layouts, we discovered that Quick Start cards were often competing with the map for space, undermining the primary interaction.
To address this, I rethought Quick Start cards as a set of pinned icons beneath the map, with expandable detail housed in a dedicated My Routines menu. This preserved fast access to familiar routines without interrupting the map-first experience.
Drawing from familiar patterns in apps like Apple Maps and Zillow, this approach kept the map visually dominant while allowing depth to live one layer deeper.
In parallel, I partnered with engineering to introduce smarter default Quick Starts, giving new users meaningful starting points even without prior cleaning history.
Quick Starts, Refined
BRINGING IT ALL TOGETHER
Routine Builder delivers on the core insight from early research: Roomba users want a direct path to a clean home. This experience makes routine setup feel intuitive and direct, without sacrificing control.
For everyday messes, cleaning starts directly from the map.
People can tap rooms or choose a familiar routine, quickly review what will happen, and start cleaning within seconds.
What was once a multi-step set up flow became a lightweight action that fits naturally into daily life.
This foundation set the stage for routines to grow from simple shortcuts into a more intelligent, personalized system over time.
In the moment, cleaning is quick and easy
As people discovered routines that worked for their home, familiar patterns became easy to repeat, schedule, or automate without requiring them to rebuild decisions each time.
What started as quick, one-off actions gradually turn into habits. By keeping recent and recommended routines immediately accessible, the framework supports consistency without planning or setup.
Over time, routines could move seamlessly into automations, allowing cleaning to happen reliably in the background, aligned with people’s schedules and preferences.
Turning routines into habits that build over time
IMPACT
01
Completed the shift to a home-first cleaning experience
Centering the app around the home map and repeatable routines aligned the experience with how people actually think about cleaning: by space, not by device. This shift clarified the app’s core purpose and made starting a clean feel more natural and direct.
02
Usability improved measurably post MVP
The redesigned app saw SUS improve from 63.6 to 82.5, moving from 'OK' to 'Excellent' usability in just under 5 months. Routine Builder was the core interaction update enabling this home-first transformation.
03
Established a foundation for intelligent automation
By grounding cleaning in the context of the home and recurring routines, the experience is positioned to support smarter recommendations and future home-level automation. This work created the structural foundation for intelligence to feel contextual, trustworthy, and incremental over time.
REFLECTIONS
-
Real user behavior aligned the team
Testing map-first entry early gave the team shared evidence to move past opinion and invest confidently in what mattered most. Seeing people gravitate toward the same patterns helped align product and engineering around a clear direction.
-
MVP tradeoffs prioritized focus over scope
Prioritizing the most valuable interactions early allowed us to ship meaningful features quickly, while creating space to learn and evolve the experience based on real use rather than assumptions.
-
Familiar patterns reduce friction, and make room for progress
Anchoring the experience in patterns users already understood lowered cognitive load and let us focus innovation on intelligence, automation, and scale.
WHAT'S NEXT?
This work grounded routine creation in how people already think about cleaning, making it easy to build routines they can trust and reuse.
Over time, those repeated behaviors form clear patterns that become a foundation for more intelligent guidance, without jumping straight to automation.
As additional data and intelligence come into play in 2026 — like seasonality, shedding pets, household activity, and advanced debris detection, and more — recommendations can become more contextual and supportive. The goal isn’t to replace human judgment, but to help people maintain better cleaning habits without having to think about the details.
Ultimately, cleaning should feel less like something to manage and more like something Roomba quietly supports, stepping in when it’s helpful and staying out of the way when it’s not.