How iRobot built the Roomba from military robots to consumer icon

Executive overview

iRobot spent 12 years building robots for the military and industrial clients before cracking consumer robotics. The company had no business model, no marketing expertise, and survived on government contracts while chasing the universal demand people had been voicing from day one: a robot to clean their floors.

The Roomba succeeded not through a grand plan but through accumulated capabilities — mine-hunting navigation, toy manufacturing economics, and an accidental celebrity endorsement from a Pepsi ad.

The core insight: survival-mode contract work built the exact skills needed to make a mass-market robot vacuum — years before the founders knew that's what they were building.

Early years: surviving on contracts

• iRobot founded in 1990 by Colin Angle, Rod Brooks, and Helen Greiner with no external funding for eight years
• Revenue came from selling research robots to Boeing, Mitsubishi, and similar clients at $5,000–$10,000 each
• Six and a half years passed before the company started a month with enough cash to make payroll
• A co-development model with large corporations — working at cost in exchange for shared IP — funded R&D without giving up equity
• The SC Johnson partnership explored autonomous floor-cleaning robots for commercial janitorial use, planting the seed for Roomba
• Engineering culture ran deep: "marketing" couldn't be said without "weenie" immediately after

Pacbot and the defense business

• The Pacbot — a tracked, pizza-box-sized robot — was deployed by US forces in Afghanistan and Iraq to identify and disarm IEDs
• Thousands of units sold, eventually generating around $50 million in revenue from the defense side
• Soldiers sent postcards saying the robots had saved lives; one unit came back with pencil ticks marking each IED it had neutralised before a bomb finally destroyed it
• Defense work was a bill-payer, not the mission — the mission was always consumer robots

Learning consumer manufacturing through Hasbro

• A three-year deal with Hasbro to make robotic toys taught iRobot low-cost manufacturing and consumer product development
• The product, My Real Baby (1999), sold 150,000 units and covered costs for three and a half years
• Key lessons: eyes that move left-right instead of opening and closing look unsettling; babbling speech synthesis requires careful filtering to avoid accidental offensive phrases

Building the Roomba

• Consumer demand for a robot vacuum was universal from iRobot's first day, but the technology and economics weren't ready
• Early prototype dragged an electrostatic cloth — consumers valued it at $40–50, less than the cost of the battery
• Adding a vacuum raised willingness to pay to hundreds of dollars, but required inventing a highly efficient suction mechanism with minimal power draw
• Solution: counter-rotating brushes for large debris, plus a narrow-gap "squeegee vac" that accelerates airflow through a thin slot — vacuum broken into two separate systems
• Navigation was deliberately simple: a round shape ensured the robot could always turn around in its own footprint
• Launched at $199 through Brookstone in 2002 after Sharper Image passed; the Cheerios demo — crushing cereal into carpet in the buyer's conference room — closed the deal at every level of the company in 40 minutes
• 70,000 units sold in the first three months; 150 press articles published immediately after launch

The Pepsi ad and marketing epiphany

• After selling 70,000 Roombas in Q4 2002, iRobot ordered inventory for 300,000 units in 2003
• Sales flatlined; only 50,000 sold by Black Friday 2003, leaving 250,000 robots in stock
• A Pepsi ad starring Dave Chappelle — made without iRobot's knowledge — showed a Roomba chasing him around a house; sales tripled overnight
• All 250,000 units sold in six weeks after the ad aired
• Realisation: iRobot knew almost nothing about marketing and needed to hire people who did

Quality crisis and brand investment

• Roomba was engineered to last 150 hours, matching the European standard for upright vacuums — far too short for a robot used daily
• Units began failing after six months; iRobot replaced them all for free
• The replacement programme was expensive but generated more customer loyalty than if customers had never experienced a problem

Scaling and hitting a ceiling

• iRobot went public on NASDAQ in 2005 with around 250 employees
• Revenue approached $100 million within two years of launch
• Growth stalled around 2008–2009 as early adopters were exhausted; the mainstream customer wanted control and confidence, not a robot that operated invisibly
• Response: add systematic navigation, object recognition, and interfaces that let customers direct where the Roomba cleaned — efficacy, control, and safety
• Roomba eventually captured roughly 70% of the global robot vacuum market; 30 million units sold by 2020

Competition, COVID, and the Amazon deal

• Tariffs starting in 2020, COVID supply-chain disruption, and surging shipping costs all hit simultaneously
• Chinese competitors offered products consumers liked at lower prices; iRobot's market share dropped
• Amazon offered to acquire iRobot for $1.7 billion in 2022
• The European Commission and FTC blocked the deal; the acquisition was abandoned in 2024
• Within a year of the deal collapsing, iRobot was sold to a Chinese company
• Angle's view: the FTC blocked a deal that would have benefited consumers, then handed the technology — built over 35 years in the US — to a foreign competitor

Lessons on leadership and building

• Angle's wilderness-guide background made the CEO role feel natural; the hardest part was putting down the soldering iron
• Core leadership habit: continuously identify everything you're doing and give it away to someone who will do it better
• Delegation inflection point: hiring a fresh graduate who produced better 3D CAD on day one than Angle could — and accepting never to do it again
• Don't plan to the second decimal point 12 years out; focus on the next two quarters and keep a north star
• The next company is in consumer robotics, using generative AI to control robots