Before Zapier had a public launch, a marketing budget, or an investor, it had 800 paying customers. Each had handed over $100 for access to a beta. According to co-founder Wade Foster, speaking on the SaaS Club podcast, that took about nine months, and the channel that produced them was free and entirely public: threads in other companies' product forums where users were asking for integrations that did not exist.
Nine years later, Forbes reported Zapier at $140 million in annual recurring revenue, having raised about $1.3 million in total. The through line from one number to the other is a habit the founders started before the product worked properly.
The channel: someone else's support forum
Zapier's product connects software tools to each other. That means its buyers announce themselves in a very specific way: they go to the forum of a tool they already pay for and write some version of "does this connect to X?"
Foster's tactic was to find those threads on the forums of products like Evernote and Wufoo, and reply with a working answer that pointed at Zapier. He describes the yield honestly on the podcast: roughly 3 to 10 visitors per day from a post, with a conversion rate above 50%.
Ten visitors a day is a rounding error by growth-marketing standards. But look at the conversion. Half of the people who arrived became customers, because a person writing "I wish Evernote talked to my form tool" has already done the qualifying. They have named the problem, named both products, and publicly declared they want a fix. Foster's word for these threads is that they are 100% qualified audiences.
The compounding property matters too. A forum post does not disappear at the end of the day like an ad. It sits in search results, and the next person with the same question finds it. Twenty posts producing five visitors a day each is a hundred visitors a day, permanently, for the price of writing twenty helpful replies.
We have written about how to find these conversations systematically in mining public complaints. Zapier is the cleanest worked example we have found of doing it as a sales channel rather than as research.
The filter: charge $100 before the product is finished
The second decision is the one most founders get wrong in the opposite direction. Zapier charged $100, one time, for beta access to software that was incomplete.
Foster's reasoning, in his own words: "I want someone who is really invested in this problem that they're willing to spend a little bit of money to solve it."
Free beta users are cheap to acquire and worth very little as signal. They will use a broken product patiently, give vague feedback, and disappear when asked to pay. A person who paid $100 will complain in detail, because they now have standing. Every bug report is worth more, every feature request comes from someone with revenue attached, and the founders always know exactly how many people want the thing enough to buy it.
The price also solved a business problem. Nine months of nights and weekends is expensive in a way that has nothing to do with money, and $80,000 of beta revenue makes that stretch survivable. Founders systematically price too low at the start; charging for an unfinished product is the most aggressive possible correction.
The rejection that forced the discipline
There is a sequencing detail worth noticing. The founders built their first prototype at a Startup Weekend and applied to Y Combinator immediately. They were rejected. At that point they had no customers.
They spent the following months collecting the 800, then applied again and were accepted into the Summer 2012 batch, launching publicly that June with 30 integrations.
It would be neat to say the rejection was the making of them, and Foster does not claim that. What is observable is the order of events: a no from an investor, followed by nine months of doing the unscalable thing, followed by a yes. The customer work was not a substitute for funding. It was what made funding available, and it also made funding less necessary. Zapier took a single $1.3 million seed round in 2012 and, per Y Combinator's interview with Foster, never raised another, reaching profitability in 2014.
The mistake they admit to
During the beta, Zapier collected around 10,000 email addresses from people who were interested but not paying. Those addresses sat untouched until launch. Foster calls it a major mistake on the podcast.
The cost is easy to compute in hindsight. Ten thousand people had raised their hands over nine months. Any of them could have been sent a working example, asked what they wanted connected, or offered the $100 beta a second time when the relevant integration shipped. Instead they got silence, and by launch day most had forgotten why they signed up.
If you are running a waiting list right now, that is the lesson to take. A list you do not talk to decays. The interest that made someone sign up in March is gone by September unless you spend it.
What the early numbers turned into
Zapier's own figures show the shape of what followed: more than 300,000 registered users within about two years of launch, integrations passing 350 with roughly 250 of them built by outside developers on the platform.
That last split is the strategic payoff of the forum work. Every integration request Zapier answered in public was evidence of demand for a connection; eventually the company let other companies build those connections themselves. Forbes reported in March 2021 that Zapier had over 300 app partners and connected to more than 3,000 apps, on pricing that ran from free to $599 a month, with most customers on $19.99 or $49 plans. In January 2021, Sequoia and Steadfast Financial bought shares from Zapier's original investors in a secondary sale valuing the company at $5 billion, a transaction in which, per Forbes, the founders did not sell.
What transfers and what does not
Transferable: find the forums where your buyers describe your product in their own words; answer with something that works; charge early enough that the answer to "will anyone pay" arrives before you have spent a year building; and talk to your waiting list from day one.
Less transferable: Zapier's product had an unusually clean fit with this channel. Integration requests are specific, searchable, and constantly reposted. If you sell something people do not know how to ask for, the same tactic will find far fewer threads. Community moderators are also less tolerant of vendor replies in 2026 than they were in 2011, so the version that works now is genuinely useful answers where your product is one line at the end, not a link drop.
The general principle survives the differences. Zapier's founders did not create demand; they went to the places where demand was already being typed out in public, and then charged for the fix. Our piece on where first customers actually come from reaches the same conclusion from a dozen other founder accounts.
Figures are from Wade Foster's SaaS Club interview, Y Combinator's published interview with him, and Alex Konrad's March 2021 Forbes report, checked July 25, 2026.
Discussion
Sign in with Google or just a name. No email link, no password to remember.