Homejoy was a marketplace that connected customers with house cleaners. It shut down in July 2015 after running out of money while facing lawsuits over whether cleaners were employees or contractors. (YC; WIRED, July 2015)
The cause was disputed. CEO Adora Cheung pointed to the lawsuits as a barrier to funding. Former employees interviewed by WIRED also described heavy losses, weak repeat business, costly expansion, and cleaners taking customers off the platform. (WIRED, October 2015)
Pressure-test this opportunity
Explore the risks and possibilities with a prompt for ChatGPT, Claude, or your agent.
AI-researched. Check the sources before making a decision.
Learn the service before scaling it. Adora cleaned homes and required staff cross-training. That understanding improved product judgment, but consistent delivery still depended on local operations.
Separate trial from durable demand. Discounts acquired bookings without establishing normal-price repeat behavior. Preserve cohort terms and observation windows before treating growth as economic proof.
Match growth to the actual service. Cheung later said unit economics deserved earlier priority. Cleaning’s limited scale benefits weakened the case for racing into more cities.
Review operating and legal risks together. Classification disputes obstructed financing while retention and quality were under pressure. Resolve required legal review before approving expansion; record what evidence the decision used.
Overview
Homejoy made hiring a cleaner easier, but struggled to turn that convenience into a durable marketplace business. Siblings Adora and Aaron Cheung entered Y Combinator's Summer 2010 batch with Pathjoy, an online-services idea. Their cleaning website launched in July 2012. Homejoy screened cleaners, matched appointments and handled payments and support.[1][2][3]
Homejoy's company logo from its Y Combinator profile.
The service announced closure in July 2015. Worker-classification disputes complicated fundraising, while reported retention and service-quality problems weakened the case for further expansion. Cheung later said she would have prioritized unit economics earlier. Her account supports a failure of growth discipline alongside legal uncertainty, rather than an explanation based entirely on lawsuits.
YC labels Homejoy acquired. Contemporary reporting documents an operating shutdown and Google's hiring of part of the technical team. Those outcomes should remain distinct.[4]
Founding Story
The founders found the cleaning problem while trying to work in a messy apartment. Finding someone suitable required calls, comparisons and scheduling. Adora took the first jobs herself and then worked at a cleaning company to learn the service. In her 2014 interview, she said, “cleaning is a hard job.” That experience helped the software team understand what its providers did.
Homejoy also required office staff to test-clean homes. A 2013 YC post reproduced Cheung's account of cross-training all 30-plus headquarters employees. It described a belief that doing the work improved decisions; it did not establish measured financial returns.[5]
Adora Cheung in Y Combinator's 2013 profile of Homejoy's staff cleaning and cross-training practice.
The 2014 interview dates the website launch to July 2012; a September 2013 Forbes account dates launch to October 2012. Both place the cleaning service in 2012. The 2010 batch date describes the startup's lineage. The 2012 date describes the cleaning product. Cheung recalled a dozen pivots before that product, which gave the siblings practice shipping ideas but did not establish demand for their original online marketplace.
Timeline
Summer 2010: Pathjoy entered YC with an online-services marketplace.
July 2012: The consumer cleaning website launched, according to Cheung's 2014 interview.
April 2013: A Twilio founder interview listed nine metropolitan areas.[6]
December 2013: YC reported $38 million in combined Series A/B financing, bringing total capital to $40 million. Redpoint described a presence in more than 30 markets.[7][8]
2014: Homejoy expanded internationally; Cheung later described launches in London, Paris and Berlin.
March 2015: Cleaners filed actions alleging misclassification and unpaid compensation.[9]
July 17 / July 31, 2015: The company announced closure, then ended the service. Google hired part of its technical team.
September 29, 2015: A federal court granted entry of default in Iglesias v. Homejoy after the company failed to defend following counsel's withdrawal.[10]
October 2015: Aaron described his new FlyMaids venture and its acquisition of Homejoy customer/provider data; he said the site was being taken down after unclear testing.[11]
What They Built
A customer supplied location, home size and appointment time. Homejoy estimated the cleaning duration and matched the job with a provider. In 2014, the advertised service cost $20 per hour with a 2.5-hour minimum. Booking combined tasks that previously required separate calls: finding a cleaner, checking availability, arranging payment and seeking help when something went wrong.
Screening included applications, background checks, interviews and test cleanings, according to contemporary Forbes reporting. Those checks selected providers; each subsequent appointment still depended on punctuality, travel and work quality.[12]
The communications system mattered as much as the booking screen. In Twilio's interview, Adora described SMS notifications for appointments, problems and resolutions, alongside phone support. She identified Python/App Engine, EC2 processing, Stripe payments and SendGrid email. These tools coordinated people; they did not perform the cleaning.
Cheung later described demand forecasting, customer-value analysis and scheduling around both sides' preferences and travel. A city launcher first recruited cleaners and manually matched supply with demand. A city manager then took responsibility for local operations and profit and loss. Expansion therefore required a working local service, not merely a website available in another postcode.
Market Position
Redpoint's 2013 investment thesis was a trusted home-services brand extending beyond cleaning. Homejoy's advantage was reducing the work of arranging a service. Retaining the relationship required continuing value after the introduction: reliable availability, problem resolution and trustworthy delivery.
Handy was a direct rival. It reportedly offered former Homejoy cleaners $1,000 signing bonuses after the closure announcement, showing that established provider supply remained valuable.[13]
The category continues. Handy's current cleaning page offers booking and customer support. Taskrabbit lets customers choose reviewed cleaners, schedule work and pay through its platform. These services challenge the idea that Homejoy failed because software-mediated cleaning had no demand. Their existence does not prove that Homejoy's own prices, customers or operations were sustainable.
As of October 2026, Homejoy's old domain and goodbye-blog route redirect to Homeaglow. Its June 2026 terms describe a separate company and recurring membership, including an initial six-month commitment and early-cancellation repricing of the first cleaning. A domain redirect alone does not establish corporate continuity or a Google acquisition. Modern promotional comparisons must include the ongoing terms, rather than only an introductory headline.
Business Model
Homejoy collected customer payments and paid providers. Contemporary Forbes reporting described $20 hourly customer pricing and $12–$15 provider pay. That spread was not profit: support, marketing, payment costs and unsuccessful appointments also required funding.
The December 2013 financing report distinguished a Google Ventures-led Series A from a Redpoint-led Series B, announced together at $38 million. Earlier funding brought the reported total to $40 million. No audited standalone financial statements establish revenue, contribution margin or cash runway.
Later customer-acquisition promotions changed the economics. Forbes reported first cleanings at $19 or $19.99 through deal channels, compared with more than $85 for a normal 2.5-hour visit. Former employees estimated 15–20% of customers booked again within a month; another source described experiments reaching 30–40% in some markets. These conflicting accounts do not establish one comparable retention rate.[14]
A promotion could pay for discovery without paying for a lasting customer. To evaluate that trade, an operator needs the same cohort's discount, observation window, normal-price repeat bookings and service costs. Aggregate booking growth cannot answer those questions.
Traction
The nine markets listed in April 2013 grew to more than 30 in Redpoint's December account. An April 2014 interview described more than 100 employees and over 1,000 contractors across 31 cities. These are dated reported measures of reach, not proof of active usage or profitability.
Homejoy also started a foundation supporting veterans and military families, documented in a November 2013 YC post.[15] That initiative shows a broader company mission; it supplies no evidence that subsidized customer acquisition was sustainable.
Post-Mortem
Growth outran economic proof
Cheung's 2018 retrospective is unusually direct: she would have prioritized unit economics earlier. She said growth and user experience ranked ahead of economics, while cleaning offered weaker network effects and scale benefits than businesses such as Uber or Airbnb. This is the founder's assessment, not an audited comparison.
The mechanism is clear. A customer may enjoy a deeply discounted service without paying its normal price again. Repeated promotional cohorts can keep bookings rising even when older cohorts decay. Expanding cities increases the need for capital before those cohorts demonstrate that they can fund continued service.
Delivery quality affected both retention and leakage
Wired's reporting describes cancellations, inconsistent service and customers hiring strong cleaners directly. Former operations manager Anton Zietsman said the company feared that training contractors to fixed standards would create legal exposure. This was an account of Homejoy's operating dilemma, not a universal legal rule banning contractor training.[16]
The marketplace faced two different losses: a poor appointment could drive a customer away, while a successful introduction could produce a direct customer-cleaner relationship. Better screening alone could not solve both. Homejoy needed enough continuing service value to keep either relationship on the platform.
Litigation and financing compounded the operating problems
Fast Company reported that Cheung called four classification lawsuits the “deciding factor.” That explanation concerns the financing decision; it does not erase the economic problems she later acknowledged. The reviewed accounts do not quantify the lost funding, defense costs or eventual liability.
Plaintiffs' counsel describes Diana Ventura's claims for misclassification, overtime and other compensation. The separate Iglesias court order reproduces Homejoy's statement that it would wind down and expected an assignment for the benefit of creditors. The September order granted entry of default, and directed a later motion for default judgment. It does not itself decide classification after a contested trial. No final merits or damages disposition was established here.
Closure, hiring and later assets had different outcomes
Computerworld reported Google's confirmation that it hired a portion of Homejoy's technical team. Homejoy stopped bookings and its service ended July 31. The evidence supports talent hiring alongside closure, without established terms for buying the operating company.
Aaron's October Hacker News statement adds a separate asset sequel: he said FlyMaids acquired customer and provider data. Customers in the discussion questioned the reuse, and he acknowledged unclear testing and took the site down. The statement establishes his account of the transfer, not a finding that every privacy allegation was true. A founder sequel and asset transfer do not reverse the original service closure.
The smaller rebuild opportunity is also contested. Amplitude already compares cohorts and retention; Mixpanel defines and measures return behavior. A separate decision registry must prove that linking those reports to operational and counsel evidence improves expansion decisions enough to justify another tool.
Key Lessons
Hands-on work improves product understanding. Adora learned the service and required staff to test-clean. Use that knowledge to improve delivery as well as booking.
Trial and repeat demand need separate evidence. Introductory price, normal-price behavior and observation windows must remain visible within the same cohort.
Growth assumptions depend on the service. Cheung's retrospective identifies weaker scale benefits than she expected. Validate each city's economics before extending the playbook.
Quality and worker models belong in one decision. Service consistency required operating controls, while classification disputes created financing risk. Review the actual model before expansion.
Track what survived precisely. Hiring employees, transferring data and preserving an operating service are different outcomes with different obligations.
