Lessons from Grubhub & Pinterest

Think of a typical Friday night. Maybe you’re heading out to a local concert, which you RSVP’d to through Eventbrite. But before you head out, you order a pre-show meal from DoorDash. Finished with your meal, you Uber to the venue — but not before tossing on that new hat you bought off Etsy.

In the span of a few hours, you used four marketplace apps — without thinking twice.

Although these marketplaces have now become stitched into the fabric of our routines, their paths to product-market fit were far from seamless. Just ask Casey Winters, the seasoned growth expert behind massive marketplace success stories like Grubhub and Pinterest. 

A little backstory: On the heels of Grubhub’s $1M Series A, Winters was hired in 2008 to drive demand for the platform. At the time, the company had no marketing team and was hovering around 40,000 users across three cities. By the time Winters left after a five-year stint, Grubhub served 1000+ cities and over 1M customers. He built the growth strategy from the ground up (although this was back when “Head of Growth” wasn’t even really a title). He helped launch the mobile app, piloted the loyalty program, and laid the groundwork for vociferous growth with a sophisticated SEO and SEM strategy.

Winters had officially caught marketplace fever. He then moved to work on growth at Pinterest, leading the charge of turning a middling SEO strategy into an acquisition juggernaut. Then he joined Eventbrite as the Chief Product Officer, just nine months before the pandemic completely upended the events industry (to put it mildly). In this trial by fire, Winters managed to launch new tools for event creators to pivot to virtual and reach larger customers (helping Eventbrite turn profitable in the process).

Across his storied growth career, Winters has cultivated a healthy appreciation for the black diamond-grade slopes these marketplaces must navigate to unlock sustainable growth. “When you’re building a marketplace, you’re tasked with generating demand and supply — so it can take longer to get the product to fully unlock,” he says.

Marketplaces tend to take longer to reach product-market fit because you’re effectively building for two customers at the same time. 

As a growth leader and now an angel investor and advisor for marketplace startups, Winters has a multi-dimensional view into the exact moves that separate ubiquitous marketplaces from the also-rans. In this exclusive interview, he demystifies the 0-1 process into a simple punch list of do’s and don’ts that can turn promising startups into massive, multibillion-dollar marketplaces. 

Winters shares battle-tested advice with case studies from the likes of Uber, Faire, Pinterest and DoorDash for each step of the marketplace lifecycle: choosing an idea to work on, acquiring your early customers on both the supply and demand side, and expanding beyond that initial core evangelist group into a wider market. Let’s dive in. 

Picking an idea

Do: Choose a market with many suppliers

Just about any product can be bought and sold through an online marketplace. But even though a product category has hints of demand, that doesn’t necessarily mean it’s going to grow into a mammoth offering. That’s why Winters emphasizes that founders must be thoughtful about choosing a market that actually has growth potential — right from the start. 

“Across the board, marketplace founders are focusing on finding scale from the get-go,” says Winters. “They’re asking themselves, ‘When I deliver value to the customer, how’s that going to create scalable growth loops that help me acquire more of them?’”

It’s critical to present the user with substantial choice. When a user opens Grubhub, they see pages and pages of delivery options. When they open Rover, they find tons of pet lovers to walk their pooch. And when they open Amazon, they find, well, everything.

Their value prop is clear: We know the thing you want and will connect you with the many businesses that are selling that thing. But marketplace founders can only unlock that value if they’re working in a market with a wide range of diverse suppliers. 

Marketplaces with narrow selection are also less valuable to the suppliers themselves. The marketplace isn’t helping them beat out competition because there isn’t much competition — so they’re less likely to pay that platform a sizable commission. 

“If there are only a few suppliers — like flights, for example — it’ll be hard to have a high enough take rate (the fee a platform charges for facilitating a transaction) to build a successful marketplace,” says Winters. “So for a travel marketplace like Expedia, they actually make most of their money off hotels, which is a category with a lot more variety and fragmentation.”

Don’t: Target customers who always want the same thing

The dream for any business is to create a product where customers come back again and again. But when building a marketplace, you actually don’t want to target markets where customers make the same exact purchase over and over. 

“On the demand side, you’re looking for customers with high frequency and high promiscuity, for lack of a better word,” says Winters. “You want your marketplace to serve customers who want to engage with different suppliers over time.”

For example, no founder has been able to build a successful marketplace app for haircuts. After all, some people only get a haircut a few times a year — that’s low frequency. And in many cases, customers are loyal. They get their haircut at the same salon or barber shop for years and years — which means there’s low willingness to try different providers. 

“If users are only going to buy from one supplier forever, then the ability for the marketplace to add value and keep those transactions on your platform goes down dramatically,” says Winters. If you build a marketplace app for hair salons, folks might try a couple different salons before they find the one they like — and then will ditch the marketplace entirely to just book with their new favorite salon directly.

Meanwhile, a platform like Grubhub works because users want to try many different restaurants and cuisines. They may order delivery up to a few times a week — sushi one night, Thai the next, then Italian the week after. Unsurprisingly, the food delivery market has managed to support several successful food delivery apps. 

Winters clarifies that you don’t need to fulfill all of these market criteria perfectly. “If you have low promiscuity, then you have to compensate with some other quality,” says Winters. “And if you’re going to be low frequency, then you need to be high average order volume. And if you can’t be high AOV, then you need to have a high take rate.”

Winters points to Zillow as a good example of a successful marketplace that doesn’t neatly fit these criteria. Real estate is a very low-frequency business. Some people only buy one or two houses in their lifetime. But Winters says Zillow’s home value estimator tool, Zestimate, has encouraged users to keep coming back — even if they aren’t there to make a transaction.

“Zestimate means every homeowner feels like they have a relationship with Zillow all the time,” says Winters. “So it’s a way for them to stay top of mind in a way that low-frequency marketplaces have a hard time doing.”

Finding early customers

Do: Find a scalable acquisition loop 

“The point that gets talked about the least in marketplaces, but is extremely important, is figuring out if you have a sustainable advantage for acquiring either supply or demand,” says Winters. “Successful marketplaces tend to have a really scalable acquisition loop.”

In the early days of Grubhub, it was risk-free for suppliers to join the platform. The sales pitch was simple: “If you join our platform and we don’t drive any orders to you, you don’t have to pay us anything,” he says.  

At Grubhub, suppliers only paid us if we brought them business, so it didn’t cost them anything to try us out. This made it easy to quickly scale up supply.

“Because we had a pay-performance model that was easy to get restaurants to try, our cost of sales was really low compared to enterprise sales,” says Winters.

Once a marketplace team has found an effective way to scale up supply, that variety becomes a launchpad for demand acquisition. 

“A lot of times you’re using some sort of unique data or content advantage from supply to then attract the demand,” says Winters. “SEO becomes a common vector for that. At Grubhub, it was a matter of taking all the restaurant menu data from suppliers and surfacing it to people when they search for food delivery.” (For more on how marketplaces can use content loops to scale their customer acquisition, check out this deep dive from Winters on The Review.)

A quick word of caution — this will not be as simple as copy-pasting another marketplace’s playbook. Airbnb and Yelp both have legendary stories about turbocharging their nascent marketplaces through SEO. But Winters warns today’s founders that the game has changed since the early 2010s. Many of the acquisition strategies aren’t as easy to pull off a decade later.

“In today’s environment, paid search is pretty hard to get an edge on,” says Winters. “Any sort of edge you have will generally get splintered by competitors coming into the market.”

The challenge with paid search is it always looks better yesterday than it does today.

So when Winters evaluates marketplaces as an angel investor, he’s looking for teams that have thought about liquidity early on and unlocked some defensible turf.

“We saw companies during the zero-interest-rate environment spend a lot of money to build up supply or demand,” says Winters. “But generally it’s harder now to convince investors in 2024 to just give you a billion dollars every quarter. You need some unique data advantage or acquisition loop that’s going to be a cheaper way to scale up liquidity in the marketplace.”

Here are a few recent marketplace examples that Winters says fit the bill:

• Faire, an online wholesale marketplace: “Part of their unique advantage is they were able to build a platform where brands can onboard their existing retailers for free and not pay any commission,” says Winters. “That allowed Faire to cross-sell those boutiques to other brands, which cheapened the cost of acquisition.”
• Power, a clinical trials marketplace: “A lot of people are searching for alternative methods of care, so they’re going to Google and searching for a detailed set of symptoms,” says Winters. “Then they try to find remedies even their doctors might not be aware of. To find these users, Power built the most informative pages you could possibly build on those topics.” This wealth of search terms helps Power become a matchmaker, helping patients with unique symptoms find trials for under-the-radar treatments.
• Fermat Commerce, a personalized retail landing page marketplace: “Part of how they got traction was by identifying founders in the DTC ecommerce community and convincing those people that Fermat was a great product,” says Winters. “So it almost became a B2B influencer model where these nodes in the network told everyone else that they should be trying out Fermat.”

Winters encourages founders to think about all the relevant networks that are potential traffic drivers — then figure out if there’s a way to get more than their fair share of the attention economy. 

“That’s generally the thing I try to counsel founders on,” says Winters. “Where are the people you’re trying to reach? And how can we think about some sort of product change that gives us the edge to reach them more scalably?” 

Each channel has their tradeoffs, says Winters, and as the company matures you’ll probably end up combining a few different acquisition channels together. “But there’s generally one channel that you can grow most quickly from in the early product-market fit stages. I like to see founders who have not only identified that channel, but spend a lot of time baking that into the core product,” says Winters.

The challenge for all companies is while there are a lot of hacky things you can do at the early stage to grow — what I call kindle strategies — these are just to validate your product hypothesis and get early users to test. They are not sustainable in the long-term and won’t help you become a billion-dollar company.”

Don’t: Shy away from some upfront non-scalable work

Uncovering a scalable acquisition loop is key if you want to launch a successful marketplace — but it doesn’t allow you to skip out on pounding the metaphorical (or even literal) pavement. 

“Tony Xu was delivering meals himself for the first year of DoorDash,” says Winters. “Because he didn’t know how to acquire drivers yet, and he didn’t want to work on that portion of the business until he validated that there was a real product there.”

Founders need to be willing to do a bunch of non-scalable stuff to unlock that initial liquidity.

By acting as the first (and only) dasher, Xu was attempting to answer the only two questions that mattered in the beginning: 

• Is there enough supply in this marketplace? 
• Is there enough high-frequency demand? 

Don’t: Obsess over software at first

Winters admits early Grubhub was “embarrassingly bad” on a software level. “When someone ordered food from Grubhub, we sent a fax (yes, a fax) to the restaurant with the order. On the fax was a confirmation code. Then, we sent an automated call that asked for the confirmation code. If you typed that in, we knew you were making the order,” he says. 

But they got away with it because (it bears repeating) the product you’re offering in a marketplace business is your supply — not your software. 

This is a key distinction for marketplaces. In SaaS, teams can spend years building, prototyping, designing and layering on slick features. While marketplace builders do this too, it should not be an early focus. Instead, top billing is bringing in more suppliers and establishing high standards of service.

The product in a marketplace is the supply, not your software. The software has to work to connect you to supply, but if supply doesn’t do a good job, they’re going to blame the marketplace.

Retaining customers

Do: Understand all three steps of customer acquisition

As the oft-repeated story goes, in the early days of Facebook, the team noticed that users who acquired at least 10 friends in the first 14 days were much more likely to stick around. Getting to that tenth friend was considered the “magic moment” in Facebook customer acquisition. 

But in marketplaces, Winters warns that customer acquisition is not nearly so simple. Instead of one magic moment, marketplace customer acquisition is a three-part process:  

• The setup: “This is the work a user has to do to have a chance of experiencing the value of the product for the first time,” says Winters. “For Grubhub that’s giving us your address so we can tell you which restaurants will deliver to you.” 
• The “aha!” moment: “This is the first time the customer experiences value from the marketplace,” says Winters. A Grubhub user’s aha moment might be the first time they realize there’s a second page of restaurants to choose from — including that burrito spot they didn’t know delivered.
• The habit: This is the sweet spot — when a customer comes back to the marketplace again and again. They turn to delivery from Grubhub when they don’t have time to cook. Or they reward themselves with a delivery meal each Friday night.

“Getting to that habit moment is all about repeating the aha moment several times,” says Winters. “Then we can reliably predict that the customer’s going to come again and again and again.”

Winters and his team would map out exactly where the setup, aha, and habit moments occurred in the customer lifecycle. Then they would run experiments to see how those moments influenced each other — and which knobs could be turned to build more habits and boost retention rate. For example, Winters found that Grubhub customers who made a second delivery order within the first 30 days of signing up were much more likely to build a habit. 

Don’t: Overdo the discounts

With the insight that a second order in the first 30 days often leads to habitual use, it’s tempting to offer a tantalizing discount to nudge customers towards that behavior. But Winters warns that discounting is a dangerous dance.

“In a marketplace, you’re selling people on the value of the selection and quality of suppliers on your platform. If you then give a discount, you’ve changed the value proposition to saving money,” says Winters.

Discounting to product-market fit doesn’t get you to product-market fit.

That’s not to say that you should throw out discounts entirely. The exception is when the data indicates that nudging users to try something will dramatically boost the LTV of that customer. For example, Grubhub found that users who downloaded the app had a much higher LTV than users who didn’t. So the Grubhub team decided to give $10 off to anyone who downloaded the mobile app. This boosted the volume of mobile users — a group that ordered at a higher frequency.

Don’t: Rely on supply to bring demand

Winters often catches marketplace teams leaning too heavily on a “Field of Dreams” mentality — building up their selection of suppliers, then relying on these suppliers to bring their customers to the platform. While this “if you build it, they will come” model isn’t destined to fail, it lowers the ceiling for overall marketplace success. The marketplace isn’t providing suppliers with much value, so they’re less likely to offer large commissions.

“You still work with buyers and sellers, but the difference between a marketplace and what I would call a SaaS network is that a marketplace’s primary value prop to the supply is that the platform brings the demand,” says Winters. 

If supply brings all the demand, then you’re just a fulfillment tool for their own customer base. 

For example, a company like Square works with buyers and sellers, but the suppliers are generating all of the demand. Customers are not flooding into suppliers because they use Square terminals. And the take rate reflects that — Square is not charging nearly as high of a fee for suppliers as a marketplace that is generating demand for the suppliers. So Square is not a marketplace, it’s a SaaS network — a remarkably successful one. 

When you’re considering building a marketplace business, think through what the core product value is and how the economics will work. Does the money you make allow you to more profitably acquire supply?

“It’s harder to build a massive scale business when supply brings the demand, so you see fewer venture-backed wins in that category,” says Winters. “Whereas if you own demand, which is generally the most important goal of every marketplace, then you’re going to see a lot more venture scale wins.” 

Winters is seeing a new generation of marketplaces that offer some additional products to help repeatably drive demand. For example, Faire offers retailers free branding tools, like creating custom email campaigns within Faire, which helps drive more demand toward Faire’s wholesaler marketplace.

“A lot of customer development is figuring out what hoops you need to jump through to scale a way to deliver demand,” says Winters. “And that might involve actually building more real software.”

Expanding the customer base

Do: Raise supply standards over time 

“Grubhub’s software itself was hacky for a long time, but the restaurants were getting more demand,” says Winters. “Meanwhile, folks ordering delivery were seeing way more options than they had in the past.”

For Grubhub, this wealth of supply was the product. “But if the suppliers don’t do a good job, customers are going to blame the marketplace because the marketplace recommended that supplier,” says Winters. “So you generally have to raise the standards of supply over time.”

Winters suggests marketplaces do this in two ways:

• Raise standards for inclusion in the marketplace: “What we did at Grubhub is if your restaurant’s ratings were below a certain threshold, we just kicked you off,” says Winters. “If you couldn’t reliably deliver food of good quality and we got enough complaints or enough low reviews, we kicked you off because we knew you weren’t going to deliver a good service.”
• Incentivize supplier behaviors that drive more demand: Winters advises marketplaces to tell suppliers, “If you do X, you’ll rank higher and get more orders.” Airbnb’s Instant Book feature is a powerful example of supplier incentive in action. Before it became standard on the platform, hosts could offer Instant Book as a way to stand out from other properties and attract demand. 

Winters nudges marketplaces to be generous with sharing their intel with suppliers. “You know what the customer wants and how to acquire customers — and using this knowledge can help both the marketplace and the suppliers.” 

Don’t: Be rigid with your value prop

Winters notes that these standards can evolve over time. For example, in the early days of ridesharing, proving the safety of the product was paramount — but that evolved as ridesharing became more commonplace. 

“Initially, there was the fear of getting into a stranger’s car, so it was all about proving legitimacy,” says Winters. Even when faced with a similar problem, Lyft and Uber approached it quite differently. “Uber early on was optimizing for luxury as a form of trust, while Lyft was optimizing for friendliness as a form of trust. Uber had licensed drivers, with luxury and black cars. And Lyft had pink mustaches and it was like your fun, friendly neighbor driving you around.”

But then, Uber makes a heel turn. “They decided that what matters most is price. So that became Uber’s focus instead of making sure all the cars were black cars. That focus on price scaled much better compared to their rivals’ models, which were still focused on legitimacy and trust.”

Do: Grow with your customers 

“In startups, your early adopters are going to put up with the most flaws,” says Winters. “They’re the ones with the most egregious need for the product. But then inevitably, as you try to expand the volume of users and transactions on your platform, you’re going to need to expand to people who have a less obvious need for your platform — which by default means they’re going to be harder and more expensive to acquire.”

Winters says this is a hurdle a lot of marketplaces can’t quite clear. But Grubhub managed to avoid this trap by proactively improving the product. They scaled up supply to meet the evolving expectations of their growing customer base.

The product needs to get better faster than the users you acquire get worse.

“At first, we were acquiring people who would order food regularly, which wasn’t that large of a market,” says Winters. “Only in a few cities, like New York and Chicago, do people do that. But then we went from having 10 online ordering options to a thousand, which brought in a whole new crew of people who weren’t ordering food as regularly or even at all before.” 

As customers got pickier, supply got better — and Grubhub was able to buck the trend. “Our retention actually got better every cohort,’” says Winters. “This wasn’t because we iterated on software. It mainly improved because we got really good at acquiring high-quality restaurants and showing them to people.”

Don’t: Rush into market expansion

Imagine a thriving marketplace where people buy and sell Funko Pops. The team nailed the previous steps. Customers are purchasing Funko Pops again and again from a wide variety of suppliers. The platform serves not just Funko Pop obsessives, but the mother who’s buying a Funko Pop for her son’s birthday. And the metrics indicate customers are building a habit.

There’s only one problem: the product is Funko Pops. Even if the platform drives a ton of Funko Pop transactions, the company is unlikely to find venture-scale success in such a narrow category. 

That’s why Whatnot — the livestream marketplace that originally just sold Funko Pops — decided to expand into other collector categories like sports cards, handbags, and sneakers. Today it’s a multi-billion dollar company. But before expanding into other categories, the team had to develop conviction in its core business’ effectiveness.

“You first need to ask yourself, ‘How are we reaching supply and demand?’” says Winters “‘How costly is acquisition? Is the acquisition really hacky? Is the process not scalable? Or is it something that will seem to scale pretty well as we get to 1,000, 10,000, 100,000 customers on the demand side?’” 

Early on, you have to focus like there’s no expansion business to worry about yet. Don’t try to launch 10 markets at a time if you don’t have a refined market expansion playbook.

“I find it best to assume that if the product’s working really well in one category, it has the potential to get in multiple categories,” says Winters. “But it takes time. Look at DoorDash. It took them 10 years to win the food delivery market and only now are they starting to expand. I remember having conversations with their team much earlier about making these expansions and I was like, ‘Don’t — you haven’t won yet.’”

If you can’t very clearly articulate how you won the initial market, do not expand yet, says Winters. “It’s very important for you to understand what steps really mattered in getting your first market to work,” says Winters. “Then you can figure out which steps to replicate to get the second market to work or the second category to work.”

Evolving from startup to scaleup

Do: Level up your data sophistication

While building intricate software is not on an early marketplace’s to-do list, robust data analysis should be. When Winters partners with marketplaces after Series A or Series B, the first step is building up data capability. “We enrich the dashboards that your team should be paying attention to and we double-check that you’re looking at the appropriate slice of data,” says Winters.

“If you’ve expanded into more than one product category or more than one city or neighborhood, you’ll find the aggregate data doesn’t really tell you anything anymore,” says Winters. “You need to ramp up your data sophistication a lot more quickly — or you risk conflating different trends in different cities or categories.”

This can get complicated. While the activation metric for a company like Grubhub was fairly simple (two orders in the first 30 days signals a habit), a marketplace like Eventbrite requires more nuance. 

For example, a local improv troupe might host monthly shows through Eventbrite. Meanwhile, a music festival organizer might host only one or two festivals a year through the site. Both are habitual Eventbrite users — but their habits look quite different.

“If you put all of these habitual users in one cohort, it kind of looks like gobbledygook,” says Winters. “So we had to build custom cohorts for different subsegments of our supplier base. This allowed us to understand the many different types of creators and avoid labeling some as churning out just because they were hosting events less frequently.” 

Don’t: Ignore supply churn

One particular red flag marketplaces tend to mistake for yellow is supplier churn. “A lot of times, people are acquiring a lot of supply that doesn’t really get any transactions and then that supply will churn,” says Winters. 

So just like you identify the activation metrics for the demand side, you’ve got to pinpoint this tipping point on the supply side, too. “Getting even just two orders a day via Grubhub would convince a restaurant to stay on the platform,” says Winters. 

And of course, don’t neglect qualitative data. Lean on surveys to figure out if suppliers are satisfied with the service. Track if suppliers are churning out or if they’re coming back to the platform. What is the value prop that the demand side is most excited about? How are you measuring NPS?

Do: Be a fast follower

If competitors begin to emerge in your market in quick succession, typical startup advice is to run your own race. But with marketplaces, the ostrich strategy will not work, warns Winters. “One thing marketplace founders need to be careful about is if you’re aggregating supply and demand in the market, and someone else is aggregating the same supply and offering it at a lower cost or with a higher frequency type of transaction, you need to copy them immediately or you are going to get disrupted,” he says.

Take Rover’s quick pivot to sniff out the competition. “So Rover was initially a marketplace for dog boarding — which is low frequency. You’re only using dog boarding when you’re traveling,” says Winters. “And then Wag comes along as a dog walking marketplace. So all the people who love dogs on the supply side who would do dog boarding are also going to walk, but the demand for walks is a lot more frequent. Wag just shot out like a rocket compared to Rover.”

So Rover quickly introduced dog walking onto the platform — and took pole position. Had they ignored the competition, we probably would have forgotten they even existed, says Winters. 

Don’t: Fail the marshmallow test

As you expand, you may need to make some tradeoffs that sacrifice short-term gains in favor of more long-term growth investments. This was the conundrum Pinterest’s team faced as they tried to expand into international markets. 

While Pinterest’s U.S. customer base was growing quickly, international growth was nearly nonexistent. “It was always easy for the team to run an experiment that moved U.S. growth a couple of percentage points,” says Winters. “So naturally, that’s what everyone spent their time focusing on — instead of worrying about getting Brazil from zero to one.”

Finally, former CEO Ben Silbermann put his foot down. He told the company that metrics needed to be goaled in five non-US countries. In the meantime, they would move their focus from US growth — and if those metrics dropped a little, so be it.

“So it was the marshmallow test: Are you going to eat that marshmallow in front of you or wait and get the second marshmallow later on?” says Winters. “And even if the founders pass the marshmallow test, it can be hard for everyone else in the organization to pass the test too.”

It’s incredibly hard to get a team working on a product that has product-market fit to shift to something else that’s not going to show immediate return.

Shifting priorities requires forceful culture-setting from the top — and probably some added muscle. 

For instance, when Winters was at Eventbrite, the business was succeeding as a fulfillment SaaS network for event creators — but it wasn’t really a destination for consumers seeking out cool events. They decided they would need to shift gears and invest in jumpstarting consumer initiative to ensure long-term growth.  

So Eventbrite created an entirely separate business unit, which became the consumer team. “We wanted this consumer group to be completely removed from the SaaS side so they wouldn’t get sucked into the day-to-day trials and tribulations of Eventbrite’s SaaS business,” says Winters. 

The transition wasn’t easy — and Winters admits that even today Eventbrite is still “slowly hacking away” at establishing a consumer-oriented culture. But gradually, those initial investments are beginning to pay off. While Eventbrite was once driving practically none of the demand on the platform, it now drives about 30% of it. 

“So be patient,” says Winters. “If you’re a little bit more patient and invest in the core product experience, that will always pay back 10 times more as you scale.”