Platform leakage II: the good, the bad, and the ugly

In this post we look at three marketplaces for which leakage is a significant issue and discuss how well they address it: from the good (StyleSeat), the bad (CoachUp), and the ugly (Homejoy). All are similar in that (1) they involve consumers obtaining a personalized service from “professionals” that is provided physically (rather than online), (2) there is significant demand for repeated interactions with the same professional. Yet they have dealt with the underlying leakage problem in different ways, with very different results.


Founded in 2011, StyleSeat is a marketplace that helps consumers discover and book appointments with beauty salons and stylists, mainly in the U.S. By 2020, it had raised $40 million and intermediated $6.5 billion in revenue for its beauty professionals. According to its website, StyleSeat charges professionals a $35 monthly subscription fee (after a 30 day free trial) and a 3% credit card processing fee for regular appointments made via StyleSeat’s booking system. It also takes 50% of the first appointment revenue (up to a maximum of $50) from a new client which comes via StyleSeat. Finally, StyleSeat charges clients a $1 booking fee.

Thus, the fees charged for ongoing transactions with existing clients provide little reason to try to go around StyleSeat (basically just a $1 per transaction fee). This is important because clients typically go to the same professional repeatedly, and form a relationship with them, so they would usually be comfortable to deal directly if it made economic sense for them to do so. Most importantly, StyleSeat provides plenty of benefits for keeping the transactions on its platform (website or app). These include:

  • Ability to get paid through the platform rather than having to handle payments directly. This includes the option to receive the funds on the same day for an extra 50 cents per transaction.

  • Handling the scheduling of bookings 24/7 via an app.

  • Personalized reminders for clients about their upcoming bookings.

  • The ability for professionals to better manage clients who do not show up. This includes allowing professionals to determine how strict their no-show policy is, such as collecting a deposit from clients so they can charge a no-show fee. Such clients can also be blocked from future bookings.

  • The ability to handle last-minute cancellations and automatically notify clients who are most likely to want the spots.

  • Sending an automatic receipt to clients after each appointment that includes a link to leave a review. The review system can be used by professionals to highlight their best customer testimonials at the top of their StyleSeat page.

  • A tool for making notes on individual clients that enables the professional to keep track of important details about a client and personalize each appointment, as well as customize marketing campaigns or promotions to them.

  • A complete record (and breakdown) of past transactions and payments.

  • The ability for professionals to turn on smart pricing, which automatically increases prices during their most popular times as determined by a StyleSeat algorithm. Professionals pay 33% of the incremental revenue they obtain through smart pricing to StyleSeat.

Of course, in addition to these benefits, StyleSeat also helps professionals find new clients by boosting them in StyleSeat’s marketing and social media channels. But the above list of benefits is what is relevant for keeping existing clients using StyleSeat and avoiding leakage.  


Launched in 2012, CoachUp is a marketplace that allows athletes (serious or amateurs) to find and book experienced coaches for private lessons in any sport (basketball, football, baseball, swimming, etc.). Coaches set their own rates. CoachUp charges coaches a one-time $50 activation fee, a $10 annual membership fee, and a sliding scale of transaction fees for every session booked through its platform: 43% of the first session with any given client, 28% of the second session, 18% of the third, 13% of the fourth, and 6% thereafter. It also charges athletes a $20 “placement fee” for their first booking with a new coach (unless the athlete was referred by the coach or is registering for a camp or clinic). Remarkably, CoachUp was able to attract the endorsement of some very high-profile athletes, most notably Stephen Curry and Julian Edelman, who are both investors in and advisors to the company.

Leakage has always been a particularly serious issue for CoachUp because (1) each athlete typically wants to find one good coach to train with repeatedly, (2) training sessions are conducted in person, and (3) CoachUp charges coaches significant percentage fees as described above. It is therefore tempting for coaches to use CoachUp to find new students and then take them off the platform after one or several sessions.

In this context, one of the most important challenges for CoachUp has been to offer enough value and positive incentives after the initial match between an athlete and a coach, so that they prefer keeping subsequent transactions on the CoachUp platform. Unfortunately, the company has been slow to add features that achieve this over the years, particularly when compared to the much more successful StyleSeat. It is useful to list all the different ways in which CoachUp can (and only to some extent does) address leakage:

  • Ability to book and schedule sessions, manage clients, monitor athlete progress with notes and videos, assign exercises and workouts in-between sessions, etc., all through an app. The ability to book and schedule sessions has been there since the beginning, but amazingly enough, CoachUp only started adding the other (important) features in mid-2020, when the COVID19 pandemic forced it to offer virtual coaching capabilities via its app. And it did so via a white-label arrangement with a third-party software provider.

  • Liability insurance coverage for any injury that happens during sessions booked through CoachUp. This has been a key part of CoachUp’s value proposition almost since the beginning.

  • Access to specific sport facilities: many athletes and coaches do not have easy access to the required sports facilities (e.g. basketball courts during winter in New England) and CoachUp can help negotiate access with local gyms at discounted rates (e.g. book time blocks that are reserved for CoachUp sessions). Despite the obvious value this feature would add to both sides, as of May 2021 CoachUp’s website still only promises it: “we're working hard to compile a list of recommended facilities and offer discounted access to partner facilities in the near future.

  • Clear incentives for coaches: appear higher in search results, possibility of being featured prominently, ability to charge higher rates, higher chance of being chosen for group lessons or clinics (which result in higher revenues than individual lessons), preferential access to facilities, premium platform features and discounted equipment, etc. CoachUp does provide some of these incentives (higher prominence in search results and chance of being chosen for clinics, discounted equipment), but not all. The key is that all of these benefits should be provided based on the number of sessions conducted on the platform.

  • Clear incentives for athletes: access to discounted merchandise, access to (and possibly short sessions with) the elite athletes that are affiliated with CoachUp. For instance, CoachUp could offer the first 10 basketball athletes that complete, say, 50 sessions via CoachUp in a 6-month period a 30 minute training session with Stephen Curry. And similarly for football with Julian Edelman. That would create very strong incentives to book sessions through the platform. It is surprising the company has not yet implemented something along these lines.

CoachUp seems to try to incentivize coaches to keep transactions on the platform through a sliding scale of transaction fees: 43% of the first session with any given client, 28% of the second session, 18% of the third, 13% of the fourth, and 6% thereafter. There are two problems with this approach. First, the numbers seem arbitrary and unnecessarily complicated. And second, it does not fundamentally solve the leakage problem. Does the fact that the fee goes from 18% down to 6% really help convince any coach to keep sessions 3 and beyond on the platform if they are considering taking them off the platform? Probably not. Rather than implementing such a mind-bending schedule of transaction fees, we would recommend focusing resources on creating clear value and incentives for keeping transactions on the platform along the lines described above. Once that is done, there is no need for a sliding scale after the second session, i.e. the fee can be the same for sessions 2 and beyond (e.g. 50% for the first session and 15% for all subsequent sessions with the same client). StyleSeat provides a good model to emulate here.


Founded in 2010, Homejoy was a home cleaning marketplace that connected house cleaners with households that needed home cleaning services. “Was” because it closed operations in 2015 after raising $40 million in funding. The company followed the Uber model, in which it set the price per hour for customers (typically $25 an hour in the U.S.) and kept a significant share for themselves (about $10 on average). Cleaners were hired as independent contractors. At the height of its popularity, Homejoy was operating in more than 30 cities, including Los Angeles, London, Berlin, and New York, and had raised money from Google Ventures and PayPal cofounder Max Levchin, among others.

Like StyleSeat and CoachUp, Homejoy was also inherently prone to leakage. Most households want to find a reasonably-priced, reliable, and high-quality cleaner they could use regularly. Once they find such a cleaner, they would be perfectly happy to deal with them directly if there are no great benefits of going through Homejoy, and especially if Homejoy is costing them more (cleaners could quote a lower hourly rate if going directly since they wouldn’t have to cover the $10 an hour that Homejoy takes).

Consistent with this, post-mortems on Homejoy’s failure point to the fundamental problem it faced: spending too much on acquiring new customers (e.g. by offering them discounted rates), but not retaining them, in part due to leakage. Only about a quarter of its customers continued to go through Homejoy after the first month, and less than 10 percent did so after six months. Part of the problem is that Homejoy did not even match customers with cleaners they had previously used and liked. So if a customer did find a cleaner they liked, they would have been forced to deal with them directly so as to ensure they could use them again. By the time Homejoy rectified this flaw, the damage had been done. And because cleaners were independent contractors, Homejoy had trouble ensuring the quality of their service. For instance, it couldn’t give them training or tell them how to perform the clean service, since that would risk workers being classified as employees, something they had to fight lawsuits on. The inconsistency of cleaner quality, combined with the inability to stick to the same cleaner once users found a good one and the high platform fees, was a recipe for an ugly leakage problem, with the best cleaners going off the platform, and Homejoy being left with a lower quality pool of cleaners.

Many of these issues would not have arisen had Homejoy adopted a different business model, based on employing workers full-time after careful screening, training them to provide a high quality of service, and firing those who did not perform. As noted in our first post on leakage, this was the model adopted by Hello Alfred. While such a model does not scale as fast, it can solve the leakage problem, and in this context make the business more sustainable.

Homejoy’s main competitor, Handy, did not suffer the same fate despite sticking to the same marketplace model of using independent contractors and charging significant transaction fees. Based on accounts from 2015, Handy did suffer from many of the same issues. One difference is that Handy’s transaction fees seem to have been significantly lower (a 20% fee, on average), which meant there was less economic incentive for parties to deal directly. And Handy did exercise more control over its cleaners (via training and incentive schemes to make sure they provided a high-quality service) even though this put the company at risk of having its workers classified as employees. Looking at Handy’s current job rates, it pays professionals more for repeat work via the platform, and for maintaining a high rating on the platform, thus further reducing their incentive to take their clients off the platform. Finally, it could be that over time Handy found ways to provide the benefits of keeping transactions on the platform, like a good scheduling system that allows customers to reschedule seamlessly (by contrast, Homejoy’s algorithm had a flaw that didn’t allow enough time for cleaners to get from one job to another), and the ability for clients to note preferred professionals so that they get priority in being allocated to the client.


We conclude with a few broader lessons based on these three examples:

  • As we also emphasized in our first post on leakage, providing clear value and incentives to keep transactions on the platform (for one or, better, both sides) is the most important thing. The key here is to provide benefits that are tied to keeping transactions on the platform. StyleSeat has plenty of such benefits, Homejoy had few if any, and CoachUp still does not have enough.

  • Fees should be commensurate with the benefits offered by the platform for repeat transactions. If it not possible to offer a lot of benefits, then consider a different business model, like charging a fixed monthly fee or hiring the workers as full-time employees.

  • Addressing leakage is important before embarking on aggressive growth. Otherwise, the business will burn through money to attract new clients and new professionals, but ultimately, they will not stay on the platform. Growing with a leaky bucket is not a recipe for success.