If it were only that simple, all two-sided marketplaces would be making fistfuls of money.

It’s never been easier to bring a two-sided marketplace online. But by now we’ve learned that just opening doors and recruiting customers and vendors usually results in a poor experience on both sides.

In other words, your marketplace can’t just be a directory of providers and a guessing game for customers. Your marketplace needs to make fulfilling customer needs seem like magic.

As I’ve honed my approach to digital marketplaces, both on my own and for major companies, I’ve developed a four-stage process for matching customers with vendors, and making those matches more accurate, more efficient, and more profitable for the marketplace platform.

This process will not only monetise your platform more quickly, it’ll make both your customers and vendors more successful, which means they’ll come back and bring others with them.

The basic ingredients of matchmaking

To get to a functional starting point for the requirements for every successful two-sided marketplace (2SM), I’m going to recite the history of 2SMs in two short paragraphs.

Remember Yellow Pages? Yeah, these were basically the second half of every phone book, and phone books were a way to find contact information for providers of products and services in a particular location. The early days of the Internet brought the Yellow Pages online, then as businesses started to build their own web presences, search engines categorised those businesses.

Those search engines spun out recommendation engines, a la Netflix. Then as the Internet started invading our privacy, those recommendations got more accurate. As eCommerce went mainstream, digital marketplaces offered value by removing most of the friction from just about any transaction. Then once almost every person on the planet became a shut-in with a computer in their pocket, digital marketplaces became the preferred way for people to transact.

Here’s the problem. Most of the 2SMs that I come across are just digital Yellow Pages.

If you want customers and vendors to transact through your digital marketplace, you have make a match between them. And not just any match, but the right match, one that eliminates guesswork, adds value, and removes friction on both sides.

This is what each side of your marketplace is paying you for when they transact.

To make a great match, you need data. Algorithms help, but you don’t need to be a data scientist, especially at first. What you do need are ways to collect that data (feedback loops), use that data (scoring), present that data (exposing the results), and learning from that data (post transaction).

Stage 1: Determination — What makes a match a match?

As quickly as you can, you’ll want to determine why your customer needs a vendor, and why your vendor is good for your customer. There are reasons on each side, and they vary from one customer to the next, and from one vendor to the next.

So let’s establish a baseline.

Consider your target customer. Is cost the most important factor in your target customer’s decision matrix? Ease of transaction? Quality of result? Speed of execution? Start by defining the scope of those reasons for each side, even manually if you have to, and document the reasons. Each reason is an axis on a scoring matrix for each side, and from there you can start scoring your target customers and vendors along the axes of those matrices.

Despite conventional wisdom, you don’t get usable data by learning from a bad match, so the less guessing you do up front, the better. Build feedback loops into your marketplace. Learn what you can from your customer as they find you, perhaps by using different landing pages. Then identify them as quickly as you can. Don’t waste time getting all their information up front, otherwise you won’t get anything.

Then, as you onboard them, before you even put the first vendor option in front of them, start learning about who they are and what they need. Adjust that learning and their scoring based on their actions within the marketplace.

Same thing with your vendors. You’ll have a longer and more thorough onboarding process here. Furthermore, the vendor vetting process, which should be mandatory, is an excellent opportunity to not just learn if your vendor is qualified to join your marketplace, but where are they best qualified to serve your customers.

You won’t get ton of information to help you make that first determination, but do it anyway. Create an initial score for each axis in your matrix on both sides. You can refine this scoring for each customer and vendor with more data you get later on from matches.

Stage 2: Efficacy — How accurate is your match?

Once you start making matches, the real scoring begins. Because you’re not just scoring how many customers you can match with a vendor, you’re scoring how accurate each match was.

The primary fear of a marketplace operator is their customers not finding what they’re looking for. This is a holdover from brick and mortar, as one of the worst customer experiences a customer can have is to arrive at a store and leave empty-handed. It’s rare that we walk into a retail store and are turned off by having too many choices.

That’s not how digital two-sided marketplaces work. In the digital world, customers don’t want more choices, they want the right choice.

So while the marketplace needs to be able to serve every customer with at least one vendor, this can quickly become a problem, as the need to put any vendor in front of a customer quickly begins to outweigh the need to put the right vendor in front of a customer.

This in turn will lead right back to the vendor directory approach, where the guesswork and friction is put right back into the customer’s lap. So on the other end of the spectrum, you’ll need to consider how many “right” options the customer should receive.

One way to solve this is by tiering the options, offering your customers a good/better/best set of options and letting them choose based on your criteria (they select from your options) or their own (offer to show them more “good” vendors).

There’s more learning here: Which vendor got chosen and were they the first or only option the marketplace offered? If they weren’t the “best” option for that customer, why not? How quickly did the transaction originate?

Stage 3: Exposure: How is the scoring reflected in your platform?

Here’s something I say a lot. Don’t ever show a customer a vendor’s score. That doesn’t serve anyone. If you do that, you’re asking the customer to make their own judgment call and you’re asking the vendor to leave your platform.

Furthermore, vendors will score differently from customer to customer, based on the individual needs of each customer and the variability in those needs.

So don’t show the score, show the results of the scoring.

That said, another way to use the scoring is to provide learning for both sides, but specifically for the vendor, by suggesting changes in how they operate within your marketplace. If you can get enough data to make an assumption like a vendor is “too expensive” or “not responsive enough” or their “offering is too limited” — and you can show, backed by data, that that’s the reason they might be missing out on more business from your customers, it is an incredible value add for the vendor.

Stage 4: Fullfillment — How well did the marketplace do?

The true value of a two-sided marketplace is in how well the platform attacks the customer’s problem with the intent to solve it.

A digital marketplace needs to be better than the standard it’s replacing. In other words, if I have a leaky pipe, I might know a plumber, or my neighbor might know a plumber, but I go to a digital marketplace to find the best plumber for my needs.

But sometimes even a good match can result in a poor experience. Feedback loops on the transaction and execution side can help reduce poor experiences and help you improve the business dynamics of your marketplace and thus, your bottom line.

This creates questions you need to be thinking about after the match: How quickly did the sale close? How quickly was the product delivered or the service executed? Was the customer happy with the results? Did the customer receive the most value along the axis that was most important to them?

Every transaction is a chance to refine the scoring that defines the determination for the next match. Better scoring means happier customers, more active vendors, and more traffic through your marketplace, maybe even higher margins down the road.

Remember, you’re not just scoring vendors and customers, you’re scoring the platform itself. The more matches you can make and the better those matches are, the more growth you can expect.

This article was originally published on Medium by Joe Procopio

Joe Procopio is a multi-exit, multi-failure entrepreneur. In 2015, he sold Automated Insights to Vista Equity Partners. In 2013, he sold ExitEvent to Capitol Broadcasting. Before that, he built Intrepid Media, the first social network for writers. You can read more and sign up for his newsletter at www.joeprocopio.com