Dropbox Referral Program: Why Free Space Beat Paid Search Ads

The room

The psychology: reciprocity

In 1960 the sociologist Alvin Gouldner proposed that almost every society shares a norm of reciprocity: people should help those who have helped them and should not harm them. Robert Cialdini and Kelton Rhoads later described how this rule creates webs of indebtedness that businesses and charities use every day. In one example they cite, a veterans’ charity roughly doubled its response rate, from 18 to 35 percent, by enclosing a few cents’ worth of address labels as a gift. In another, a $5 check sent with a survey request drew more responses than a promise of $50 paid after completion.

The Dropbox program touched reciprocity in three places. First, the company gave before it asked: 2 GB of storage that was promised to stay free, not a trial with a clock on it. Second, the invitation itself carried a gift. A friend who joined through a link received extra space, so the person sending the invite was doing the friend a favor rather than asking for one. Third, Dropbox repaid the inviter with space of their own. Houston credited PayPal’s signup bonus, not psychology research, for the design, so the reciprocity reading is an interpretation of why it worked, not something Dropbox said it measured.

The two-sided part matters because a reward changes how an invitation looks. If only the sender is paid, a recommendation can look self-interested. If the friend benefits too, the same message reads as generosity. Research on referral incentives by Rachel Gershon, Cynthia Cryder and Leslie John supports this: rewards aimed at the new customer recruited more people than rewards for the sender alone, partly because senders gain reputation by passing on something valuable.

Why smart people fall for it

Three forces helped a storage bonus do what paid search could not.

1. 1

   The gift comes before the ask

   Users had already received a useful free product before they were asked to share it, so the request followed a favor rather than preceding one.

2. 2

   Inviting looks generous, not salesy

   Because the friend also got extra space, the invitation read as a gift, which protected the sender’s reputation and made sending it comfortable.

3. 3

   Trust travels with the invitation

   A friend’s recommendation answers the question an ad cannot: whether an unfamiliar product is worth trying. Referred customers also tend to stay longer.

Why did the Dropbox referral program work when search ads failed?

Houston’s own answer was about demand. Search advertising is excellent at catching people who already know what they want. In 2009, few people were typing in a search for a product that quietly syncs files between computers, because most of them did not know such a thing existed. The obvious keywords were contested by other funded startups, and the long tail had almost no volume. In Steve Blank’s terms, Dropbox was creating a new market, and in a new market there is not much existing demand to harvest.

A referral does what search cannot: it creates demand by putting the product in front of someone who was not looking for it, through a person they already trust. Research on referral programs suggests the channel also brings better customers. Philipp Schmitt, Bernd Skiera and Christophe Van den Bulte studied about 10,000 customers of a German bank that paid a 25 euro reward per referral. Referred customers had about 18 percent lower churn and an estimated 16 to 25 percent higher long-term value, although their early margin advantage faded after about 29 months. The authors pointed to better matching: existing customers know both the product and their friends.

The economics also changed shape. A paid click cost money whether or not the visitor ever paid. Dropbox’s referral reward was paid in its own product, and only when a friend actually signed up. Houston’s slides do not say what the extra space cost the company, so the true cost per referred user is not public. But the structure tied spending to results in a way the search test never did.

Finally, the referral program amplified something that already existed. Houston’s slides say that word of mouth drove most of Dropbox’s growth, and shared folders brought in another fifth of daily signups simply through use. The bonus gave people a reason and a tool to do more of what satisfied users were doing anyway. That is a different job from persuading strangers to try something they have never heard of, and it is the job referral programs are best at.

Was it reciprocity, or simply a good deal?

The skeptical reading is that people wanted free storage and Dropbox paid them in it. That is partly true. A bonus that arrives only after a friend signs up is a conditional incentive, closer to a commission than a gift, and 2.8 million invitations in a month suggest that plenty of users were chasing space. Yet the examples Rhoads and Cialdini cite cut the other way: a small gift given upfront can outperform a larger reward promised later, which suggests that the conditional bonus alone does not explain the program.

The reciprocity reading rests on what came before the bonus and what the bonus said. Dropbox’s generous free plan came first and had no strings, which fits Gouldner’s norm: users who had been helped were inclined to help in return. And because the friend also received space, the invitation was framed as a favor from one person to another. Regan’s experiment showed how strong an unrequested favor can be, even when it comes from someone the recipient barely knows. An invite from a friend that arrives with a gift attached carries both the favor and the relationship.

Research on two-sided rewards suggests the framing matters. In a review of 351 online referral programs, Gershon, Cryder and John found that about 55 percent shared the reward between both parties, 40.5 percent rewarded only the sender and 2.6 percent rewarded only the friend. In their field experiments, rewards for the friend outperformed rewards for the sender, though in one field test the difference was marginal and conversion rates were low overall, which the authors described as typical.

What we cannot say is how much of Dropbox’s 60 percent lift came from reciprocity and how much from plain self-interest. Houston did not publish a test comparing one-sided and two-sided rewards. The fairest conclusion is that the program combined three things that are hard to separate: a real gift given first, an invitation that was socially comfortable to send, and a tangible reward for doing it.

How would a modern team design a referral program?

It would start by checking whether people already recommend the product. Dropbox had a community of users who were spreading it before any bonus existed, and it worked with Sean Ellis on surveys and tests to understand them. A referral program cannot manufacture enthusiasm. It can only make existing enthusiasm easier and more rewarding to share.

It would then design the invitation around the friend. The message that worked for Dropbox was about what both people get, which made sending it feel like a favor. Where possible, the reward would be paid in the product itself, as credit, capacity or features, so that incentives flow to people who actually use what the company sells. It would also set limits. By 2012 Dropbox capped referral space for free accounts at 16 GB.

A modern team would measure quality, not just volume. The German bank study found that referred customers were more valuable, but managers in that research also worried about moral hazard, such as customers referring poor prospects just to collect the fee. Van den Bulte warned that in business-to-business settings, cash referral fees paid to a client’s employees could look like a bribe. Tracking retention and value of referred customers over many months is the way to see whether the program is attracting the right people.

Finally, it would be honest about the trade-off. Every gigabyte given away is a gigabyte a user does not need to buy. Dropbox’s 2018 filing showed about 11 million paying users among 500 million registered accounts. That model works only if a small share of a very large base pays, and only if the product keeps earning the recommendations that feed it.

The manager's playbook

1. Give real value before askingLet people use a genuinely useful free version before you ask them to recommend it, so the request follows a favor rather than preceding one.
2. Reward both sides of the inviteGive the new user something on joining as well as the referrer, so the invitation reads as a gift to a friend.
3. Pay rewards in your own productWhere you can, reward with capacity, credit or features rather than cash, so incentives reach people who actually use what you sell.
4. Cap and verify every rewardSet a ceiling per account and credit rewards only after the new user is genuinely active, to limit gaming and spam.
5. Track referred customers’ retentionCompare churn and value of referred and non-referred customers over months, not just signup counts, to see whether the program brings the right people.

Questions people ask