You launched. There are listings, there are signups, there is traffic. There are almost no transactions. The instinct at this point is to market harder, and the instinct is almost always wrong.
Why Buying Traffic Is the Natural First Move
When a marketplace goes quiet, buying traffic is usually the first move, and the logic is sound on its face. We have users, they are not transacting, therefore we need more users. It is also the only lever most founders have direct control over, which makes reaching for it entirely rational.
I did exactly this with my own money, twice, before I understood what I was looking at.
Here is why it does not work. Traffic multiplies whatever is already happening. If a hundred visitors produce two transactions, a thousand visitors will produce twenty, and you will have paid ten times as much to learn the same thing you already knew. Worse, you will have burned through a chunk of your addressable market at the exact moment your product was least able to convert them, and those people do not come back a second time out of curiosity.
A stalled marketplace is a diagnostic problem. The good news is that the diagnosis is genuinely tractable, because there are only four places it can break.
The Four Loops
We run a framework called A-CAR: Attract, Convert, Adopt, Retain. It is not complicated and that is the point. Every marketplace that is alive is completing all four in sequence. Every marketplace that is stalled has stopped at one of them, and you can find out which in an afternoon with data you already have.
- Attract is whether the right people arrive at all, on both sides.
- Convert is whether an arriving user completes a first transaction.
- Adopt is whether they come back and do it a second time.
- Retain is whether they keep doing it, on your platform, rather than around it.
Attract is the loop everyone checks first, for the good reason that it is the only one an analytics dashboard shows you by default. The other three are where most stalls actually live.
We have since built the method into a tool. Marketplace Roadmap walks the same four loops, scores what you find against the eight criteria we use to decide what gets built next, and hands you a Kanban roadmap and editable PRDs at the end. It is free while it is in beta. The rest of this article is the manual version, and the manual version works perfectly well.
Loop One: They Cannot Find a Match
The failure looks like traffic with no transactions and a high bounce rate on search.
The number that matters here is your search-to-result rate: what percentage of searches on your platform return something the searcher would plausibly want. Not results returned. Results that fit. If someone searches for a photographer in Ottawa on a Saturday and you return four photographers in Toronto on a Tuesday, that is a zero-result search wearing a costume.
This number is very hard to judge from the inside, because when you test your own marketplace you naturally search for things you know are there.
The most common cause is geographic dilution. You had reasonable supply in one city, growth felt slow, so you opened five more. Two hundred providers concentrated in one market becomes forty per market across six, and forty is thin enough that most searches fail. Every one of those six markets now delivers a worse experience than the single market did, and the original market got worse too because your attention left.
The fix is unglamorous and it is contraction. Pick the strongest market, pull back to it, and get liquidity genuinely deep before expanding again. Founders resist this because it looks like going backwards to investors. It is the single most reliable intervention we run.
Loop Two: They Find a Match and Do Not Book
Traffic is fine, search works, and people still leave.
This is where trust lives. A marketplace asks a stranger to give money to another stranger on the strength of your say-so, and everything that makes that feel safe is doing real work: reviews, verification, clear cancellation terms, visible pricing, a refund policy that a nervous person can find in under ten seconds.
The specific killer at this stage is thin review counts. A supplier with two reviews converts dramatically worse than one with twenty, and early marketplaces are full of suppliers with two reviews. This is a chicken-and-egg problem nested inside the original chicken-and-egg problem, and the way through is to manufacture the first reviews honestly: broker transactions by hand, deliver them properly, ask for the review directly.
The other common cause is that your take rate is visible and feels unearned. If a buyer sees a service fee before they have experienced any service, they price your platform against going direct. Reviews, protection and dispute handling are what make the fee legible.
“Nobody tells you they left because they did not trust you. They just leave, and the analytics record it as a bounce.”Darren Cody, Marketplace Studio
Loop Three: One Transaction and Then Silence
This one is easy to miss, because the first-transaction chart looks healthy right up until it does not.
A marketplace is not a marketplace until people come back. One transaction proves the exchange is possible. The second proves it is preferable. If your cohorts transact once and then flatten, what you have is something people were willing to try. Turning that into something they use is a different piece of work, and it is the piece most of the value sits in.
The fix depends on why. Sometimes the category is genuinely low frequency, in which case the marketplace has to be built around that reality rather than pretending otherwise, and the economics need to work on a first transaction because there may not be a second. Sometimes the first experience was mediocre and nobody complained, they just did not come back. And sometimes there is simply no reason to return, because you never gave them one.
We wrote at length about why the second transaction matters more than the first. The short version is that everything valuable in a marketplace, from reviews to habit to unit economics, is built on repeat behaviour, and acquisition cost only amortises across a relationship.
Loop Four: Leakage
The quiet one, and the one almost nobody measures.
Two people meet on your platform. They complete one transaction. Then they exchange numbers and do the next twelve directly. Your dashboard records a successful transaction and a lapsed relationship. In reality your marketplace worked perfectly and then got cut out.
Leakage is easy to miss because the transactions you can see all look fine. The signal is in the shape of the relationships rather than the volume of transactions. If most buyer-supplier pairs have exactly one transaction and then go quiet, and those buyers do not reappear with anyone else, the platform is doing the introduction and not much after it. That is a solvable problem, but it is a different problem from the one the transaction count suggests.
Penalties do not fix this. Contractual anti-circumvention clauses are close to unenforceable at small transaction values, and attempting to enforce them poisons supplier sentiment at exactly the moment you need it stable. What works is making staying genuinely easier than leaving: handling payment and invoicing, holding the scheduling, keeping records both sides want, resolving disputes, offering protection that only exists on-platform.
What a Healthy Set of Numbers Looks Like
Benchmarks vary enormously by category, so treat these as orientation rather than targets. Conversion improves over the first year as trust and density build, and it improves faster in lower-trust categories.
| Marketplace type | Month 1 conversion | Month 6 | Month 12 |
|---|---|---|---|
| Service | 10 to 20% | 20 to 40% | 35 to 55% |
| Experience | 6 to 12% | 15 to 25% | 25 to 45% |
| High-trust or high-value | 3 to 8% | 10 to 20% | 18 to 30% |
If you are twelve months in and sitting at month-one numbers, that is not a slow start. That is a broken loop, and it will not fix itself with more traffic.
The Uncomfortable Possibility
Sometimes all four loops are working and the marketplace still does not grow, and the honest conclusion is that the exchange you built does not need to exist. The friction you set out to remove was tolerable. The two sides were already finding each other well enough.
That is a real outcome and it happens to good founders with good execution. It is also much cheaper to discover in month nine than in year three, and there is usually an adjacent version of the idea that does work, with a narrower wedge or a different constrained side.
The thing worth avoiding is another year of traffic spend against a loop that has not been diagnosed, because the diagnosis is usually a fortnight of work and the spend is not.
Diagnosing which loop broke is the first two weeks of every go-to-market engagement we run, and it is the reason we do not sell a growth retainer until we know what is actually wrong. If your marketplace is live and stuck, a thirty minute call is usually enough for both of us to form a view. We have also written in more depth on solving the chicken-and-egg problem.
