Posted in: Uncategorized

Marketplaces, Rating Systems, and Leakage

In the past three years I have had the pleasure of meeting many marketplace companies looking for seed funding. I have also had the privilege of investing in quite and continue to learn along the way. One of the things I spend a lot of time thinking about and exploring is what drives “leakage” and what marketplaces can do to keep transaction settlement on their platforms. One of the most common ways in which marketplaces try to keep transactions on-platform is by rolling out rating or review services for service providers (and in some cases for buyers as well). I’ve been thinking about when and why these rating systems work (they are not instant panacea) and more broadly about some observations about marketplace dynamics that tend to lead to “leaks”, or off-platform transactions where the platform provider does not capture the revenue associated with the transaction they helped facilitate.

What Good Marketplaces Do Well

I studied Economics in college, so I’ve always been attracted to the marketplace model. When applied and executed correctly, good marketplaces do many things to reduce transaction costs between buyers and sellers, including the following:

  • Reduce search costs for buyers and sellers – Good marketplaces make it easier for buyers and sellers to find and connect with each other. This liquidity is the lifeblood of a healthy marketplace.
  • Reduce transaction costs for both parties – When speaking about transaction costs, I don’t simply mean facilitating financial payments. Many marketplaces have standard fees and rules that make dealing with unknown strangers simpler and transactions more standard.
  • Reduce bargaining and enforcement costs – A good marketplace usually has both norms and rules that make dealing with unknown parties safer. These rules and services can include payment escrow, trust and safety teams, and dispute resolution for cases when things go wrong.

The nice thing about the marketplace business model is that marketplaces that provide the services above are often able to command a transaction fee or premium for doing so. However, doing so often means that the marketplace needs to be in the middle of the transaction. The real risk to most marketplaces is that they provide all of these valuable services and, at the moment of the transaction, the transaction “leaks” and the buyer and seller settle the transaction off platform.

Experience Goods Tend to Have Marketplace Leakage Issues

Experience goods are products and services that require actual usage and experience in order to evaluate the quality of said service. For example, housecleaning services and ridesharing services are both experience goods. On the other hand, any unboxed iPhone is effectively the same as any unboxed iPhone – all unboxed iPhones are effectively commmodities. For experience goods, the main risk for a buyer is often that the quality of service, once delivered and experienced, will prove to be of low quality. For commodity goods, the main risks are that the seller will not actually fulfill his or her end of the transaction by providing the product or that the product will prove to be counterfeit or otherwise defective.

The reason that experience goods tend to have leakage is that once you know that a given supplier has proven to be good, there is a tremendous incentive to use that person again. And given that a lot of the value that many marketplaces provide is associated with the search and identification of a trustworthy service provider, there can be a strong incentive for both parties to complete future transactions out-of-band as it allows both the buyer and supplier to avoid the marketplace transaction fee. If you want to kno more about marketplace, rakes, you should definitely read this post by Bill Gurley as I reference it often.

One way that many marketplaces try to combat leakage is by rolling out rating and review systems. More on the logic as to why below.

The Role of Rating Systems in Transaction Costs and Leakage

Ratings systems are a pretty simple and tried-and-true system to grasp. In most cases, buyers have the ability to review the experience in dealing with a seller. In some marketplaces, namely eBay and Uber, both sellers and buyers can review each other. In order for a review system to work in terms of enforcing norms of good behavior, two things have to be true:

1. Suppliers have to value their rating, either as a buyer-facing signal of quality or because supplier rating determines how leads are distributed and routed within the marketplace.

2. Buyers have to trust in the validity and quality of the rating as a reliable indicator of supplier quality.

When both of these conditions hold, there is at least a chance that reviews will be of some value to those in the market. If suppliers do not value their rating, they will not care about taking their transactions off platform because having a high rating does not tie to the things they care about – making more money and potentially getting more customers. And if buyers believe that ratings are gamed or otherwise influenced, they lose their value as a signifier of quality. Getting review systems right is not easy, but well-designed rating systems can provide a strong nudge to settle transaction on-platform.

There are three situations where I continue to see strong incentives to go off-platform even in the face of the existence of a rating system. I’ve outlined them below:

  1. Situations where there is real value to repeat usage with a chosen service provider. One area where I tend to see opportunities for leakage is the world of services marketplaces. In the case of tutoring, housecleaning, childcare, or any other service that is truly an experience good, there is almost always buyer risk on the initial transaction. But once you, as a buyer, find a service provider you like, the search is over. And, often the value of completing the transaction on the marketplace and paying the associated fees goes down. It might be more economical and simple for both the supplier and buyer to complete the transaction off-platform and avoid paying a fee.
  2. Marketplaces where the marketplace itself delivers a small percentage of the service provider’s total income.The other situation in which I tend to see marketplace leakage is where the marketplace delivers a relatively small portion of the service provider’s total revenue. In those cases, suppliers tend not to care about their ratings because the amount of money to be made or lost on that marketplace is small relative to their total income. Any money or leads lost due to a low rating can easily be made up elsewhere.
  3. Marketplaces where the buyer and supplier are geographically proximate and can easily complete the transaction offline. Last but certainly not least is what I describe as the Craigslist use case. In situations where buyers and sellers meet in person, there is a huge opportunity to simply settle in cash, Square, or some other form of payment rather than complete the transaction on the platform.

Of the three points mentioned above, I think the first tends to be the most pernicious. And the more marketplaces I meet, the more I am beginning to believe that the most pernicious situation of all is the combination of #1 and #2. I might even argue that marketplaces that are struggling mightily with #1 and #2 might want to rethink whether they are in the marketplace business or really in more of a lead-gen context. And when I meet new marketplaces, I try to get comfortable with how they will address the points above should they occur.

I think it’s really hard to build a marketplace that effectively balances the three concerns above. I think the ridesharing services, Uber in particular, are doing a great job of balancing all of these pressures. I suspect that Uber-delivered rides represent a significant chunk of income for their drivers. And drivers and riders are matched with each other in part based on their respective ratings or scores. That creates a pretty strong incentive to not “cheat” the platform and to care about your score as both a service provider and a buyer. That, to me, is evidence of a well-designed, well thought out system.

If you’re working on a marketplace and have insights to share, I’d love to speak with you and hear your experiences. As always, comments are open below. You can also chat with me on Twitter @chudson.

Comments (16) on "Marketplaces, Rating Systems, and Leakage"

  1. Thanks for sharing Charles! About 3 months ago I asked a task rabbit if I could pay him extra cash to complete an additional task (aka leakage). He politely asked that I submit the task through the app, and he can claim it. At first, I was surprised the allure of cold, hard cash didn’t waver him, but then I realized how much he stands to benefit from keeping it on the platform.

    That’s what great marketplaces do. Reward you for full transactions.

  2. Great post, Charles. One of the other things that reduces leakage is the transactional utility of the marketplace. A good example is Open Table. Once a restaurant acquires a diner through Open Table, there’s a large incentive for them to try to transact with that diner offline (to avoid paying Open Table a commission). But often that doesn’t happen because the diner likes the utility of Open Table’s scheduling system — they don’t have to call, they can view available times, etc.

  3. Awesome Post, Charles! Usually, I have found that you can build User / Customer loyalty by initially fronting the service / product yourself. This way, the marketplace gains traction as a conduit of trust, honesty and performance. In most cases, successful marketplaces carry a powerful brand and ethos, a culture they stand for.

    If the founder of a marketplace can display this character, usually both vendors and customers will choose not to rock the boat by settling offline. e.g. Airbnb.

  4. very insightful post, Charles – your conclusions are spot-on. Case in point: I recently used a tutoring marketplace to find a Biology tutor for my son. I spent several frustrating weeks (and many hours) sorting through and interviewing lots of not-quite-right candidates. Then I finally found a candidate I liked enough to book for a trial session – and SHE suggested that we move our transaction offline because the tutoring marketplace would take 40% of her fees. So she’s effectively using the service as a discovery platform for new clients.

  5. I’ve seen that happen quite a bit, especially when you have marketplaces (tutoring is especially guilty) where the transaction fee starts and stays high and there is a lot of repeat interaction. Paying 40% for lead gen might work, paying 40% time after time just feels like an aggressive take.

  6. We are seeing interesting behavior at the Panjo marketplace for auto, sport, and hobby enthusiasts. Within Panjo’s partner network one archetype of buyer and seller seeks to complete a transaction on platform as a way to support the partner’s Web-based community to which he/she is loyal. This anti-leakage behavior reflects a desire on the part of the buyer and seller to support the affinity-based community in the right place, at the right time, at the right price.

  7. Nice post Charles. I might add immediacy and location as factors that can inhibit leakage. Using Uber as an example where experience matters, even if I’ve found a preferred driver (point #1 above), and settlement can certainly take place in person (#3), users still rely on transacting through the platform for new “inventory” because search criteria (location) is constantly changing and time-to-service is key. These tend to be factors for most mobile-first services.

  8. David,

    I remember reading your post and really liking it. I have been thinking a lot about how the labor marketplaces (as opposed to marketplaces for items) work and where they can break down. And I’ve also been looking at places where a lead gen model is superior to a true marketplace. I’m still trying to sort out all of my thoughts.

  9. Great post Charles. I would say that in Uber’s case, another reason they avoid leakage is by making it easier to use Uber than to not use Uber. Most of the time, Uber is more expensive than just walking out the door and hailing a cab. Often, the reason I use it is because it’s reliably easier. Rather than waiting outside trying to wave down a cab, one comes to my door. Also, have you ever tried using a credit card to pay for a taxi normally? You have to ask if they accept credit, then watch the driver fish the portable POS system out of his glove box and wait for a few minutes while he turns it on. Uber cuts at least 5-10 mins out of the experience, which is sometimes half the time. To me, that’s often worth the $2-3 premium.

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to Top