#267 1999 · PayPal · Financial technology
PayPal couldn't get eBay to officially support it, so it built a robot that bought things on eBay and demanded to pay with PayPal
the problem
A payment method only spreads if buyers already use it, but buyers only adopt it if sellers already accept it
background
In 1999, eBay was the obvious place for a peer-to-peer payment startup to find users, because eBay auctions were exactly the kind of stranger-to-stranger transaction that needed a payment method neither party had to trust the other for. But eBay had no official relationship with PayPal and no reason to promote a third-party payment tool, and PayPal faced the standard two-sided cold-start problem in its sharpest form: eBay sellers wouldn't bother offering PayPal as a payment option until enough buyers were asking for it, and buyers wouldn't sign up for PayPal until sellers were already accepting it.
The standard fix for a marketplace cold start — advertise to build awareness on both sides simultaneously — is slow and expensive, and PayPal had no leverage to make eBay itself endorse or integrate the service. PayPal needed sellers to feel demand for PayPal that hadn't organically arrived yet.
what everyone would do
The standard fix for a marketplace cold start was to advertise to build awareness on both sides simultaneously, hoping enough sellers and buyers would each independently sign up in response to marketing — slow, expensive, and offering no leverage over the host platform whose users PayPal actually needed.
what they saw
PayPal saw that sellers wouldn't offer PayPal as an option until they felt actual, concrete buyer demand for it, and that demand didn't need to arrive organically at scale, it just needed to be real and directed at specific sellers. Instead of trying to convince sellers abstractly through advertising that buyers wanted PayPal, PayPal could manufacture that demand directly by becoming a real buyer itself, placing genuine bids conditioned on accepting it.
the move
PayPal's team built an automated script that crawled live eBay auctions, emailed sellers directly asking them to accept PayPal, and then placed real bids on their listings insisting on paying with PayPal — winning some of those auctions and actually paying for the items, which were later given to charity. Sellers who received a real bid conditioned on accepting PayPal experienced concrete, immediate demand for the service, not a hypothetical pitch, and started offering PayPal as an option so as not to lose out on buyers.
why it works
An automated script crawling live eBay auctions, emailing sellers, and placing real bids insisting on PayPal meant sellers experienced concrete, immediate demand rather than a hypothetical pitch they could ignore, since some bids were actually won and paid for with real money. A seller wanting to win that specific sale had an immediate, individual incentive to accept PayPal for that transaction rather than some abstract, someday incentive, and once a seller had gone through the setup to capture one bid, they kept the option enabled for future buyers too, since the setup cost was now sunk and leaving it active cost nothing. This converted a slow, expensive two-sided awareness problem into a series of individually resolved, concrete transactions, each permanently seeding one more real PayPal-accepting seller without needing eBay's cooperation at any point.
the payoff
PayPal ended 1999 with fewer than 10,000 users, passed 100,000 by the end of January 2000, crossed 1 million within a few months after that, and reached roughly 5 million users by that summer — a trajectory PayPal's own early leadership and later industry retrospectives describe as driven primarily by an embeddable payment button for sellers and a $10-per-referral cash incentive, with the eBay auction-bidding bot serving as an early, narrower tactic that seeded real seller adoption before those larger mechanisms scaled.
where it breaks
The mechanism requires the seeding party to actually be willing and able to spend real money as a genuine participant, not merely simulate demand — a non-binding signal like a survey wouldn't create the same concrete incentive, since sellers only respond to a bid they can actually win and get paid for. It also depends on the target platform not actively blocking or banning this kind of automated bidding activity — stricter bot detection or terms-of-service enforcement could shut the tactic down entirely. And the tactic only seeds one side of the two-sided market directly; the case's own outcome notes PayPal's actual mass-adoption trajectory was driven primarily by later, broader mechanisms like an embeddable payment button and referral incentives, meaning the bot was a narrow, effective unsticking tactic for the cold-start problem specifically, not a scalable growth engine on its own.
what came after
The eBay bot is retold in Eric Jackson's book 'The PayPal Wars' and cited across growth-marketing literature as an early example of a two-sided platform manufacturing real demand on one side of a marketplace to unstick a cold start, a tactic since echoed by other platforms (Reddit's founders posting under invented usernames is a close cousin) that needed to look and function as though critical mass already existed before it actually did.
references
- [1]PayPal's Robot ArmyStrategy Breakdowns, 2023strategybreakdowns.com
- [2]9 Iconic Growth Hacks Tech Companies Used to Boost User BaseThe Next Web, 2014thenextweb.com