Consumer marketplace companies are hot—just look at all the ink being spilled on the recent IPOs of Uber and Lyft and the expected, upcoming listing of Airbnb.

But we see plenty of other types of marketplace companies—online hubs where buyers and sellers meet—in industries beyond ride sharing and lodging, and continue to believe there’s plenty of room for new, innovative startups to create value.

Building a successful marketplace company in the early days, though, is not always intuitive. Based on our experience working with these types of companies, we’ve isolated six key areas for marketplace entrepreneurs to consider when building long-term, defensible businesses. These are all factors we evaluate when looking at new marketplace businesses.

Know your TAM

There are numerous ways to calculate your total addressable market (TAM)—and we’ve seen a lot of them! TAM is important because defining your market will help you define your business. In our view, many entrepreneurs oversimplify the way they calculate market size, or aren’t entirely realistic about which market segments they can actually reach. And specificity is good! Outlining assumptions and providing reasons for decisions will help investors better understand your logic and thinking process around your TAM calculation.

Solid approaches to calculating TAM include “tops-down”, “bottoms-up”, and “theoretical value-add”.  The “top-down” approach utilizes industry research and a “process of elimination” type calculation – taking a target market, and then applying filters to narrow down the specific market segment.  Pitfalls include the assumptions that industry research analysts are correct in their market sizing and there’s no TAM expansion from your marketplace.

The “bottoms up” approach requires a more granular look at the market – researching and identifying specific companies or customers that have demonstrated a desire for a similar product and estimating the amount sold to each group (quantity and dollar amount).  This approach turns the TAM discussion from a simple “market penetration of an industry” statement to a more interesting “product-market fit” question and conversation.  The “theoretical value-add” approach is a bit more of a dark art, as the calculation depends on an estimation of the value provided to a customer base and how much value creation can be captured through pricing.

The best TAM estimations are triangulated using a combination of the three methods. Whether you’re releasing a new product, fundraising, or establishing new sales goals, take the time to understand your TAM. A well-defined, market-size analysis will help focus your team operations; establish ideal customer profiles and use cases; and form a feasible product roadmap and vision, to name a few.  Having a well-thought out TAM shows logic, direction and purpose.

Unlock “shadow markets” to drive expansion

Before Uber, who knew there was a vast market for people to use their own cars as taxis? Or for people to rent out their spare bedrooms (later, entire apartments and houses) as hotel rooms? Finding this type of previously opaque “shadow market” can lead to a brand-new business, and often one that can be expanded into adjacent markets to pull in even more revenue.

To put it in perspective: Today’s global taxi market is worth about $108 billion and is projected to shrink to $54 billion by 2030, according to a 2017 Goldman Sachs report. The global ride-hailing market, by contrast, is now worth about $36 billion but should grow 7.9x to $285 billion by 2030, ultimately outsizing the taxi market by 5.3x, according to the same research.  By unlocking a new class of riders and drivers, these ridesharing platforms have not only been able to steal market share from the taxi industry, but also grow the overall transportation ecosystem. It’s often tough to spot these hidden, nascent markets, but doing so can unlock significant value creation.

Thoroughly study your supply/demand dynamics.

How fragmented is your supply and demand? Which side of your network is more valuable in the marketplace–or in other words, which side has more scarcity? How unique is your supply base? What are the switching costs for each side? Consider these questions as you construct your marketplace, as they will influence take rate and potential market penetration, among other variables.

For example, Airbnb offers heterogenous supply, meaning each listing is unique. This has created a competitive moat for Airbnb, or an inherent advantage for its platform vis-à-vis competitors. However, this heterogeneity increases work for the consumer when they’re searching for a place to stay on Airbnb’s site. In turn, search becomes one of the largest obstacles for customer conversion on Airbnb.

Lyft, on the other hand, has a homogenous supply base—all of Lyft’s cars may be slightly different, but they all get you from point A to point B. In this type of marketplace, search is no longer an issue, but rather, the purchase experience becomes a matching problem the marketplace solves through a series of algorithms (eg, matching you with a Lyft driver who is near your location). Additionally, without the competitive moat of a unique supply base, homogenous marketplaces compete on other aspects of product like pricing or even brand loyalty.

Create consumer habits

One reason marketplaces have created so much value over time is because they have fundamentally transformed people’s daily lives – how we eat, how we move, how we learn, how we have fun. By creating new and better consumer experiences, these marketplaces have embedded themselves in our lives and increased the frequency by which people spend money on certain items—ordering more Ubers instead of driving a car or taking the bus, for example. Many failed marketplaces, by contrast, approach services or industries with purchasing cycles that are way too infrequent, which impedes the ability to create strong brand awareness, customer engagement and retention, and network effects. You should think about this from the get-go as you’re building out your product.

In marketplaces with sporadic transactions, our advice is to create product features that encourage interim use cases. For example, although most people buy or sell a home only once every few years (if that), real-estate marketplace Zillow has been able to boost customer engagement thanks to its Zestimate house valuation tool, which people use much more frequently.  By creating non-transactional experiences, like checking the value of your neighbor’s house—or homes in your dream neighborhood–consumers find continued value in Zillow and thus an additional reason to return to the platform.

Track and optimize growth efficiency—not just your growth

Growth in revenue and users is great—but it should be accompanied by positive and attractive unit economics. Oftentimes this may mean automating certain workstreams in which employees are manually processing data or answering customer questions—the automation decreases labor costs–or optimizing acquisition channels to try to acquire higher-quality customers and increase retention/LTV. In the early days of a startup, it’s understandable that some of these things will be done manually, in order to move fast and quickly gain information to improve a product or service. But this can create operational inefficiencies, and you should be mindful of how outlier use cases and manual work can compound over time—potentially making your business less attractive. Our advice is to maintain analytical rigor in collecting operational data (i.e. LTV, CAC, transaction size and volume, support costs, gross and contribution margins, cohort dollar retention, and burn rate, to name a few) and proactively optimize your business based on these metrics.

Build defensive “moats” and continue to reinforce them

To build a strong and profitable business, it’s important to have defensible “moats” around your marketplace. These moats can take many forms. Traditionally, moats can include network effects (a business that becomes more valuable the more people use it, i.e. Facebook, WeChat); IP/trade secrets (protected proprietary software or methods, i.e. patents, new techniques/processes); economies of scale (leverage due to size and reach); brand loyalty (a tight-knit community base and following); and/or high switching costs (embedded workflow creating dependencies in your consumers, in which dollar or time costs of switching become prohibitively high).

And, once you’ve established a moat for your business, don’t stop there.  Continue to add layers of defensibility when opportunities arise. These reinforcements tend to have compounding effects—so don’t give up too early as you work to buttress your marketplace.

Take Amazon as a case study.  In the early days, Amazon grew aggressively and built operational scale (moat #1 – scale) through its network of fulfillment centers, enabling fast delivery speeds, low prices and a vast selection of inventory. Community ratings and product reviews further reinforced trust on the platform and mitigated purchase concerns (moat #2 – brand).  By leveraging this infrastructure, Amazon was able to aggregate a unique supply base through its Amazon Third-Party Seller Marketplace (moat #3 – network effects, unique supply), which now successfully generates over 50% of its transactions.  On top of that, Amazon has innovated around voice (moat #4 – platform/IP) via Alexa, enabling a new ecosystem of apps and add-on products. All these layers of defensibility have worked together to strengthen Amazon’s competitive advantage.

The bottom line: Marketplaces are complex. But if constructed correctly, marketplaces tend to have a winner-take-all or winner-take-most outcome. It’s definitely not easy to build the next Uber or Airbnb, but there are some fundamental lessons entrepreneurs can follow to get going.