Why Marketplaces Fail: The Role of Engagement

Small and infrequent transactions are risk factors, but their impact can only be evaluated in combination with other, more important traits

Sameer Singh
May 4 · 9 min read
Image credit: Unsplash

So far, I have discussed a handful of characteristics that create structural advantages or risks for marketplaces. This includes the geographic range of network effects, supply differentiation, SaaS integration and market fragmentation. Fragmentation is most useful as a first level filter to assess the viability of any marketplace, while the rest are second and third-level screening frameworks to evaluate defensibility and scalability. Another factor that influences the potential of marketplaces is the nature of engagement, i.e. the size and frequency of transactions. While this is less influential than other structural characteristics, it can become a major risk factor for some types of marketplaces.

The folks at a16z have previously attempted to categorize marketplaces based on the size and frequency of transactions. Theoretically, marketplaces with lower transaction size face the risk of unsustainable unit economics because repeated transactions are required to break even on customer acquisition spend. And marketplaces with lower frequency of use should face higher risks from competition and disintermediation (buyers and sellers going “off-platform” to transact directly). Infrequent interaction does not create many opportunities to build loyalty, i.e. if a user only interacts with a supplier once every few months, then they may consider going directly to that specific supplier or to another marketplace for subsequent needs. The visual below is a broad summary of this categorization.

In reality, the risks highlighted above only affect a subset of marketplaces with low transaction size or frequency. For example, Numbrs is a financial account aggregator and marketplace with low frequency of use, but disintermediation is not a concern because transacting directly with financial service providers is a high friction experience. Similarly, Airbnb is only used a couple of times a year, but faces little direct competition because it has a unique and largely exclusive supply base. Categorization based on engagement is not very meaningful by itself as it does not consider the role of other, more important factors. To get a better understanding of these risk factors, we need to determine the interplay between the nature of engagement and other marketplace characteristics. One way to examine this interplay is to take a deeper look at failed marketplaces and study the causes of their failure.

Impact of Engagement on Failed Marketplaces

Marketplaces that have strong network effects and are first movers, even those with low transaction size or frequency, usually have winner-take-all (or winner-take-most) outcomes. This is because adoption makes their product significantly more useful than competitors. While this results in a long list of failed late-movers or “copycats”, that is by design, i.e. it is a result of network effects working the way they are supposed to. Airbnb (pre-Covid), again, is a perfect illustration of this. It is a dominant service used only a couple of times a year and has left many failed copycats in its wake, including Housetrip and Wimdu. In other words, copycats fail because first movers have structural advantages, not because of their own standalone flaws. Therefore, it would be more useful to study failed first movers if we want to understand the causes of marketplace failure beyond just “copycat syndrome” or poor execution. Failed marketplaces that meet this description include the following:

  1. Shyp: Connected consumers to local couriers, who picked up, packaged and shipped items via carriers like UPS, FedEx, etc. All customers had to do was enter the pick-up and destination address, in addition to a picture of the item. Raised >$60M in venture capital funding before failing.

The visual below plots these marketplaces on the marketplace matrix to summarize their relative scalability and defensibility.

Unsurprisingly, these are all tier-3 or “Uber for X” marketplaces, i.e. hyperlocal, with commoditized (or interchangeable) supply, and without SaaS integration. As I have explained previously, hyperlocal marketplaces have a more difficult time scaling because they are effectively a collection of discrete marketplaces in each hyperlocal market. And marketplaces with commoditized supply exhibit higher multi-tenanting (customers and suppliers using multiple marketplaces), face more competition and are less defensible. However, Uber and Deliveroo are also tier-3 marketplaces without SaaS integration. While they have faced their fair share of challenges, they did manage to scale and create viable businesses. What separates the Ubers from the Homejoys of the world? It is only at this point that engagement becomes relevant.

Role of Engagement

In every failed marketplace highlighted here, the value proposition was clearly valid and funding was abundant, but unit economics were a problem, i.e. customer lifetime value (LTV) relative to customer acquisition cost (CAC). Each of the failed marketplaces listed above had low ticket sizes (and gross profit) per transaction. And like Uber and Deliveroo, they were hyperlocal marketplaces that could not leverage pre-existing supply and demand to attract users in new markets. For example, when Homejoy expanded to a new city, they could not leverage cleaners in other cities to attract customers (or vice versa). Instead, they had to invest in acquiring both cleaners and customers in the new city all over again. In other words, expanding to new hyperlocal markets did not bring down customer acquisition costs over time. In this scenario, low transaction size (and gross profit per transaction) was a significant disadvantage as it made it even more difficult to recover these costs.

The combination of low transaction size and high customer acquisition costs meant that these marketplaces required a higher volume of transactions per user to break even. Unfortunately, this was very challenging because they were occasional needs at best. In addition, commoditized supply meant that multi-tenanting was rampant which intensified competition (from competitors like Handy for Homejoy, Rinse for Wash.io, etc.). Low frequency of use gave these competitors a bigger window of opportunity to poach customers via pricing promotions (while offering a comparable experience). This put downward pressure on customer retention (in addition to gross margins) and pushed up customer acquisition costs. Each of these startups eventually collapsed as unit economics proved unsustainable. The nature of transactions certainly played a part in their demise, but they merely exacerbated the structural challenges prevalent in their model.

Homejoy (and likely Kitchensurfing) also saw extensive disintermediation which depressed transaction frequency even further. Disintermediation becomes a problem when a marketplace charges the demand and/or supply side more than the value it creates for them. This is a key challenge for marketplaces addressing occasional needs, where counterparty trust is important, but there are very few “filters” required to match demand and supply, i.e. on low frequency marketplaces with commoditized supply. For example, plumbing services were an infrequent need for Homejoy customers and there was very little information required to match a customer with a plumber. It was limited to location, timing and, to a lesser extent, the nature of the problem. So when a customer found a plumber they trusted, they could contact them directly for future needs and exclude Homejoy from the transaction. This is not as big of a problem for Airbnb, even though it has low frequency of use, because its supply is so diverse, i.e. it requires far more demand and supply-side considerations (which change over time) to produce a match. These considerations include duration of stay, location, type of property, pricing, etc., and are unlikely to remain static. This makes disintermediation difficult for both users and suppliers.

To summarize, low transaction size is a key problem for marketplaces that do not have the ability to lower customer acquisition costs over time (hyperlocal marketplaces). Low transaction frequency exacerbates this problem and can create disintermediation risks. This is especially problematic for marketplaces that struggle with multi-tenanting and competition (marketplaces with commoditized supply). So while low transaction sizes pressured unit economics on marketplaces like Uber and Deliveroo, high transaction frequency gave them much more “margin for error” than Shyp, Homejoy, Kitchensurfing and Wash.io.

Engagement & The Marketplace Matrix

Now we can apply these learnings to the marketplace matrix to get a more generalized understanding of when the nature of transactions is relevant. The visual below redistributes the same marketplaces we previously categorized by engagement at the beginning of this post.

Note: Marketplaces outlined in orange are SaaS-enabled

Tier-1 marketplaces: First movers within this category (differentiated supply and cross-border structures), like Airbnb, Numbrs and Preply, are not very sensitive to transaction type. For example, low transaction size is not much of a concern for Preply’s language tutor marketplace because their customer acquisition costs decline as they expand their demand and supply bases into new markets. Similarly, Airbnb’s low transaction frequency does not intensify competition or disintermediation because of diverse supply and their role in curating it.

Tier-2 marketplaces: Tier-2 marketplaces like Shpock (differentiated supply and hyperlocal structures) are very sensitive to low transaction size (and low ad rates) as they are not capable of reducing customer acquisition costs with regional expansion. In order to counter this, Shpock has attempted to blur boundaries between their hyperlocal markets and expand into higher value verticals like used car sales. On the other hand, tier-2 telehealth marketplaces like Teladoc (commoditized supply and cross-border structures) are very sensitive to low transaction frequency. This intensifies competition from similar marketplaces like Doctor on Demand and AmWell.

Tier-3 marketplaces: As we saw with failed marketplaces like Shyp and Homejoy, a combination of both commoditized supply and hyperlocal structures results in sensitivity to both transaction size and frequency. Since they already have significant structural weaknesses, they need either high transaction sizes (e.g. RigUp) or high transaction frequencies (e.g. Uber, Deliveroo) to be viable. This also explains the struggles of car sharing marketplaces like Maven and Getaround (even before the Covid crisis).

Impact of SaaS-integration: I have previously explained how a Come for the tool, stay for the network approach can improve both defensibility and scalability irrespective of marketplace tier. This is also a way to solve the challenges created by low transaction size and frequency. For example, Treatwell’s appointment management software for salons is a “single player” tool that it uses to acquire supply before attracting demand in a new market. So despite being a tier-3 marketplace, Treatwell could leverage SaaS to lower customer acquisition costs (and generate subscription revenue) which reduced their sensitivity to low transaction sizes. Similarly, the appointment scheduling system locked down one side of the marketplace (minimizing multi-tenanting and disintermediation), thereby increasing retention on the other side. This makes them less sensitive to lower transaction frequency as well. SaaS integration is an especially valuable tool for tier-3 marketplaces because of their structural sensitivity to small and infrequent transactions.

It is important to understand that the nature of transactions is not entirely under the startup’s control, but rather a derivative of the value proposition and market. The impact of transaction type is also complex as it cannot be evaluated in isolation. Rather, it needs to be evaluated in combination with structural characteristics, i.e. geographic range of network effects and supply differentiation. While the nature of transactions plays some role in all marketplaces, it can become a matter of startup life or death for marketplaces with pre-existing structural flaws. SaaS integration is one of the most effective ways for these startups to shift the odds in their favour.

Want to receive new posts by email? Sign up here.


I study, advise and invest in network effect based startups

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store