How to design a referral program at andrewchen

Above: Dropbox’s ingenious development effort — A referral program to give/get storage

Why a referral program?
Referral programs — the “give $5, get $5” uses you see in numerous apps — have actually ended up being popular over the last few years. They have huge benefits over paid marketing channels, because you offer your CAC to your users, who then invest it within your item, as opposed to handing it over to Google or Facebook. Because they are a type of viral marketing — using your network of users to generate more users — they tap in to your item’s network results, as I explain in The Cold Start Problem. This is especially helpful for items that target high acquisition expense specific niches, whether that’s crypto users or on-demand motorists, whose CAC are typically >$200, considering that the users typically understand each other.

An effective referral program can be 20-30% of your acquisition mix, as one of numerous acquisition loops. It’s not a silver bullet, however it’s worth including to enhance other marketing efforts.

The history of the referral program
How did a structured type of consumer recommendations enter into being? It’s stated that the very first recorded referral program was developed by Julius Caesar, who in 55 BC would paid his soldiers 300 sestertii (something like a 3rd of their yearly pay) to refer a buddy to sign up with the army. And countless years later on, we still utilize, plus or minus, the exact same concept. It appears as though every customer app has actually carried out some type of a referral program, though I argue it actually started in ~2008, which is when Dropbox’s ingenious referral program was presented.

Yes, the most well-known early application of referral programs originated from Dropbox, which influenced a generation of start-ups — especially YCombinator-backed start-ups — to try out comparable concepts. Why did this make good sense for them? CEO/cofounder Drew Houston’s made a really handy discussion explaining his journey towards referral programs, and the basic trajectory was the following:

• First, do all the important things you’re “supposed” to do
  • Big bang launch at a tech conference. Try some AdWords, hire a PR company / VP of Marketing
  • Paid marketing programs developed a CAC of $233-388 for a $99 item
  • Then attempting affiliate programs, show advertisements, and numerous other things — which all stopped working
• … however then stopping working! And understanding none of it works that well
• Then understanding the secret was to speed up word-of-mouth and viral development by offering a “give and get storage space” program
• Boom. 100,000 users to 4 million in simply 15 months, with 35% of day-to-day signups

The whole deck is fantastic — developed approximately a years back, however still really appropriate — and I extremely motivate you to inspect it out here.

Referral programs work effectively for specific type of items, especially ones that are currently spreading out by means of word of mouth. In Dropbox’s case, there is a natural usage case in between pals and associates — shared folders — which naturally match the referral channel. Referrals drive that forward, offering a financial reward to inform pals. As another example, at Uber where I ran the referral programs for motorists and riders at numerous points (and invested >$300M/year on them), the program for motorist recommendations was naturally effective. Drivers were typically from specific sub-communities, whether freshly gotten here immigrants or limousine motorists, and individuals were naturally currently discussing the making chance. Referrals, in some cases as high as $500/signup, sped up that in a huge method.

And yet referral programs have their limitations. Of course they don’t actually work that well for items that have low LTV — that’s why we don’t see complimentary social picture sharing apps reward their users for recommendations. There’s no LTV to arbitrage versus, and the referral amounts develop a type of consumer acquisition expense. They likewise tend to decrease in value in time. Years after the rollout of Dropbox’s referral program, I had the chance to sign up with Dropbox as a consultant, where I got a first-hand appearance at the information. By then, the natural virality of their core item — simply the procedure of individuals sharing their folders and files with others — had actually come to entirely control user acquisition. This had actually ended up being the main approach of spread, and the referral program ended up being much lesser. I’ll talk about why, later, however this appears to be the natural pattern of things — referral programs are really handy at the start of a market. Eventually it ends up being lesser, which’s okay.

But we get ahead of ourselves. Let’s start initially by looking at how a referral program is normally specified.

The structure of a referral program
We see the exact same rough patterns in referral programs that are carried out throughout the market. Airbnb, Uber, Instacart have them, therefore do Coinbase and Wealthfront. There are variations naturally, as some concentrate on providing and getting dollars. Some ask you to share a code, or a link, or link your addressbook to welcome pals.

One method to arrange all these variations is to divide them into the following — and you require to response a series of concerns for how you structure the program:

• Ask
  • When do you ask the user to refer?
  • Why do you refer? Is it connected to a vacation, or a specific promo?
  • What’s the message?
• Target
  • Which users do you target? All of them?
  • How do you set referral quantities?
• Incentive
  • What’s the reward, is it extrinsic ($) or intrinsic (points, storage, and so on)?
  • Do you offer the inviter or recipient the exact same benefit?
• Payback
  • What is the success requirements for the program?
  • How do you think of cannibalization?

Let’s usage an example to explain this.

For example, take Airbnb’s host referral program:

You might break this down into the following classifications:

• Ask: Invite somebody who can host their whole location or personal space
• Target: All Airbnb users
• Incentive: Earn $200
• Payback: CAC is better/comparable to other marketing channels (simply hypothesizing!)

This is the standard structure, and now that we have this in location, it’s time to discuss a variety of design factors to consider required when producing a referral program.

The Ask
Product folks typically begin by painful over the ask. They question if it’s too minor to develop a “Get $5, Give $5” referral program, or if that’s too standard. But I believe that’s the incorrect location to focus — after all, you can constantly word smith and test numerous variations later on when you have the program up and running.

The genuine concern is, WHERE do you make the ask? And my response is easy: Ask lot of times, in numerous locations, with various messages, and in-context with whatever action you’re asking the user to take. What you discover, after instrumenting all your referral UI, is that there’s simply a specific conversion rate on this screen. And that a lot of users, if you put the referral performance on a banner someplace random in the item, just don’t connect with the referral functions. Rather than attempting to raise conversions, rather, reveal the screen more frequently — get more impressions!

Thus, make the referral ask part of the primary circulations. After the user is purchasing something within your app, ask if they desire $X money back now, by welcoming somebody. Or if they connect with a buddy within the app — presuming the item enables invites of some sort — follow up by asking if they desire to welcome others. And include it to the onboarding circulation, and at completion of crucial deals when the user is otherwise done, and you may too catch engagement. And for god’s sake, don’t make it appear like “an ad” with huge splash text and graphics — make it prepare, like something that’s part of the typical UI where the user can connect.

One of my preferred concepts from Uber is the idea of “holidizing” a referral project. For motorists, as the vacations approached, you may inform them to make money towards presents and celebrations, by taking part in a referral program. Or for the added to a significant show in the area, you may run a unique tiered project where referring 1 buddy gets you X, however 5 gets you 5*X and a big reward on top. There’s something excellent about refurbishing the messaging monthly to align to significant vacations, with brand-new quantities, brand-new images, and otherwise.

The Target
The heading finest practice is that your referral program ought to target brand-new users to refer their pals — this indicates triggering users throughout their preliminary onboarding streams, and including e-mails as part of the onboarding, to name a few area. This remains in direct contradiction to folks who typically argue to let users experience the item initially, have a great experience, prior to they’re struck up to welcome. Why concentrate on brand-new users? First, mathematically, it’s simplest to make a huge effect when you are striking a friend of 1000 brand-new users when it’s as close to 1000 as possible, not in day 30 when the friend will have churned and come down to 150. And in the mathematics of the viral element, you have a much better possibility to struck >1 when you have 1000 users welcome 1000 users than to ask 150 to welcome 1000. Second, brand-new users typically have more pals who haven’t yet utilized the item, due to the fact that they are brand-new themselves. Once they have actually gone through the referral program a couple of times, then they will have naturally tapped out their networks.

And naturally, the most basic thing to do is a “give $5, get $5” and consider that deal to everybody, in an untargeted style. But a item leader quickly recognizes that this mishandles — maybe it’s finest to offer some users $15 and others $5, depending upon their worth. This is precisely what numerous market business have actually done, when it’s simple to sector their network into high-value cities like New York and SF versus, state, Memphis — you can set customized referral amounts in each location. But why stop at cities? Perhaps you do an analysis and find out specific leading attributes of high-value users as their account balance, or the kinds of other apps they utilize, or otherwise — when you think about this as customizing an ephemeral deal to users, then you can run whatever promos you desire.

The Incentive
You’ll keep in mind in the initial Dropbox deal, the reward itself was storage area not dollars — this is the issue of intrinsic versus extrinsic benefits for users that take part in your program. Many referral programs for mobile video games tend towards intrinsic benefits too, making you points if you welcome pals. The benefit of intrinsic benefits is that it’s especially cost reliable when the reward is something you can manage, like points. The issue with intrinsic benefits, naturally, is that external users — individuals who have actually never ever become aware of your item — are the least responsive to points or otherwise. Dropbox’s storage deal is possibly someplace in the middle, considering that it’s at least a concrete type of worth. As a result, a lot of referral programs have actually tended towards dollars in time, though I believe the essential concept is to focus on brand-new, outdoors users, and think of how to make the reward as concrete as possible.

There’s the standard concern of how to set the reward quantity. Typically this is based upon a standard estimation of CAC/LTV, which has significant weak points as it doesn’t consider cannibalization (which we’ll talk about later on). Instead, the focus is typically to choice a easy number — if you understand that the typical user who registers invests $20, then you can develop a referral program that rewards a $5 give/get with some margin of security. But the huge lever on the reward, naturally, is to increase the quantity — and the biggest quantity typically originates from tiered deals that have some type of damage. An example of this is to state, “$100 when you sign up and buy 5 things” instead of “$5 when you sign up.” Given that the distinction in between a signup and a repeat conversion rate may be 100x, you may be able to securely raise the quantity 20x. At Uber, this presumed as to integrate 2 unique numbers: A heading number that integrated both the preliminary signup conversion in addition to the very first month’s revenues (once again, as long as you drove X journeys in the very first couple of weeks). This led to a $3000+ number, a big upgrade from the preliminary $200 numbers we began with. These bigger heading numbers constantly checked better on A/B tests, whether in e-mail marketing or banner type, and while it may seem like the benefit ends up being unattainable, it’s possible to develop a 2nd or 3rd or 4th tier to support the huge heading number. You might state, make $X when you satisfy all the requirements, however then a smaller sized number, $Y, when you just satisfy a couple of. That method you get the marketing effect of the huge number however still have a alternative for users who don’t strike all the turning points.

The last element of the reward structure I’ll talk about is a symmetric versus uneven deal — that is, should it be a “give $20, get $5” or “give $5, get $20.” Which one sounds much better to you? This is anecdotal, however in screening I’ve seen, the inviter-centric quantity typically works much better — that is, catering to their self interest. However, I’ve likewise seen B2B contexts where in a expert setting, individuals tend towards welcoming more if they are viewed as selfless, providing a big $ discount to others. In completion, most likely simply worth A/B screening to see what works best.

The Payback
You’ll require some type of ROI metric to drive the method of the referral program. Are you investing the correct amounts, or should you increase the numbers? How much item effort should be taken into carrying out brand-new area? Etc. Is it working? These essential concerns are typically responded to with a timeless CAC/LTV analysis, and there’s a factor to doing that.

After all, if the life time worth of these users surpasses the expense of obtaining them, shouldn’t you simply go complete steam ahead? Well, possibly. What if you can get more affordable acquisition by means of another channel, like TikTok advertisements. Then any dollars that go to this may be much better invested in advertisements. Or what if enhancing a referral program takes engineering group far from crucial functions? So yes, naturally appearance at the CAC/LTV of your referral effort, however think of how you may compare the tradeoffs versus whatever else.

The harder ROI concern is cannibalization: How much of the users that you generate by means of recommendations would have can be found in through word of mouth anyhow? If you surge the referral quantities, going from $5 to $20, are you simply producing a “pull forward” result where users that would have gotten here totally free a couple of months from now are being available in unexpectedly, however at a expense, and to the hinderance of a later on month?

Cannibalization is a difficult result to tease out, however typically the objective is to step something like “Cost Per Incremental Customer” by A/B screening deals to a control group that gets the basic number, and a test group that gets the raised number. Because you’re attempting to capture natural users, you typically have to do this as a “twin cities” experiment, where we run one set of deals in Phoenix and another in Dallas — this is the Uber technique, and in B2B you may do it by means of one set of business versus another. And then you determine what the uptick in fact appears like. If there is a great deal of cannibalization, then the CPIC number will be big — this is the real CAC, cannibalization aside.

The other, easier type, is just to do an “On/Off test.” If you switch off all your recommendations for a couple of days, do you observe a huge drop in brand-new users? If yes, then your referral program is working. If not, then you are possibly paying a great deal of consumer acquisition expense for something that would be taking place anyhow.

The weak points of a referral program
As I pointed out at the introduction of this essay, Dropbox’s ultimately ended up being less based on their referral program. There’s a natural trajectory here, due to the fact that as the marketplace grows and more users have actually currently embraced the item, the less pals there are to welcome. You just require a couple of “I already have that” reactions to stop taking part in recommendations entirely. For items that have a real network result, as Dropbox does, the acquisition will become taken control of by intrinsic usage cases like folder sharing instead of something as extrinsic as a referral benefit.

And in a method, I discover myself primarily hesitant when groups approach me to develop a referral program. The very first thing I ask is, are you sure you wouldn’t rather develop a viral development engine? Building viral functions and a referral program are comparable issues — attempting to get users to welcome pals — however really viral functions around sharing and interacting are evergreen and develop long lasting worth. They aid users engage and maintain, and as a secondary result, produce brand-new users too. And it’s a big advantage to get these brand-new users onto your platform totally free. For Dropbox, that indicates investing in item functions like welcoming colleagues into tasks, or file sharing, or otherwise, instead of producing ever more intricate referral structures. At Uber, this may suggest structure virality into functions like “Share ETA” or costs splitting or group food buying.

In that method, I discover myself a unwilling fan of referral programs — they can work, and can end up being a 10%+ acquisition channel for items — however they will constantly take a rear seats for me, compared to structure excellent viral performance.

PS. Get brand-new updates/analysis on tech and start-ups

I compose a premium, weekly newsletter covering what’s taking place in Silicon Valley, concentrated on start-ups, marketing, and mobile.