Hacking Growth: How Today's Fastest-Growing Companies Drive Breakout Success

by Sean Ellis

In reality, their success was driven by the methodical, rapid-fire generation and testing of new ideas for product development and marketing, and the use of data on user behavior to find the winning ideas that drove growth.

A Harvard Business Review article about growth stalls reported that 87 percent of the companies in a large study had run into one or more periods in which growth slowed dramatically, and that “on average, companies lose 74 percent of their market capitalization…in the decade surrounding a growth stall.”

Among the causes of stalled growth they cite are problems “in managing the internal business processes for updating existing products and services and creating new ones,” and “premature core abandonment: the failure to fully exploit growth opportunities in the existing core business.”

Walmart’s growth team enlisted the engineers to build an app that could allow customers to upload their receipts from shopping at Walmart via their phone’s camera and automatically receive cash refunds from the company if another chain had advertised any of their purchases for less. In addition, the engineering team realized that it could marry the data Walmart was collecting as part of its price matching program with the ad campaigns being run by their paid search teams, leading to big savings in ad spend by only bidding aggressively on items where they were the clear price leader.

In short, growth teams should be involved in all stages and all levers of growth, from attaining product/market fit to customer/user acquisition, activation, retention, and monetization.

Often, ideas for changes that aren’t part of the preestablished roadmap are met with resistance. Sometimes it’s because timing is already tight for making the planned enhancements, and sometimes because the changes being asked for are poorly conceived, much more difficult, time-consuming, and therefore costly, to enact than the person making the requests is aware of.

All fast-growth companies share one thing in common. Regardless of who their customers are, their business model, and the type of product, industry, or region of the globe they’re operating in, they all make a product that a large group of people love. They’ve built products that, in the eyes of their customers, are simply must-have.

The opportunity costs of pushing for growth too soon are twofold. First, you’re spending precious money and time on the wrong efforts (i.e., on promoting a product that no one wants); and second, rather than turning early customers into fans, you’re making them disillusioned, even angry, critics. Remember that viral word of mouth can work two ways; it can supercharge growth or it can stop it in its tracks.

Additionally, growth teams need to recognize that sometimes establishing what the core value is, or should be, isn’t about the features of the product or service itself, but rather a matter of connecting with the right core market, which, again, as we’ll explore, might be quite different from the originally envisioned one.

Or in other words, why that product is a “must-have.” This experience is what turns early adopters into power users and evangelists. For Yelp, that experience was the ability to discover promising local restaurants and businesses through trusted community reviews. For eBay, the aha moment was finding and winning one-of-a-kind items at auction from people all over the world. For Facebook, it was instantly seeing photos and updates from friends and family and sharing what you were up to. For Dropbox, it was the concept of easy file sharing and unlimited file storage. Or take Uber’s aha moment, which Uber cofounder and CEO Travis Kalanick explained as, “You push a button and a black car comes up. Who’s the baller? It was a baller move to get a black car to arrive in 8 minutes.”

Thus the key to knowing when it’s time to start the high-tempo push for growth is simple: Can you identify an aha moment that users love?

For example, Twitter struggled to sustain growth in its early days until it learned (from doing extensive analysis of its user data) that users who quickly started following at least 30 other users were much more engaged and likely to continue using the service. Digging into why following 30 people seemed to be the tipping point, the Twitter growth team found that getting a steady stream of news and updates from people they were interested in was the aha moment for people. Following 30 people created a stream of updates that made the service “must-have.”

One caveat: the Must-Have Survey isn’t recommended much beyond the stage of determining whether you’ve achieved core product value. For one thing, once your growth has taken off, it’s not a good idea to even suggest to your customer base that the product might be discontinued by asking them how they’d feel if it were no longer available. Can you imagine the panic if Facebook sent its users a survey suggesting it might go away?

Josh Elman, for example, highlighted just four questions that the Twitter team asked users who had gone dormant and subsequently returned: (1) Can you tell us why you signed up in the first place?; (2) What didn’t work for you? Why’d you bail?; (3) What caused you to come back and try it again?; and (4) What worked this time?

Take the case of Highrise, a customer relationship management product that the firm Basecamp launched to complement its popular project management software, for whom an A/B test of the copy on the sign-up page revealed that simply changing the language from “Sign Up for Free Trial” to “See Plans and Pricing” netted 200 percent more sign-ups.26

There is more data available to growth teams than ever before, but all this data is essentially useless without the ability to parse it for useful insights.

The key mission at this stage is to look for the behaviors that differentiate those customers who find your product must-have—that is, those who use or buy repeatedly—from those who don’t. Specifically, analysts should be looking for features that are most used by the most avid users and any other distinctive aspects of their behavior in interacting with the product.

For example, at Netflix, by examining the movies and shows that customers were watching, the company found that Kevin Spacey films and political drama series were both hugely popular with their customers. That insight gave the company confidence to green-light the development of House of Cards, which became not only a huge hit, but also a must-have experience for many subscribers.

Similarly, at RJMetrics, a business intelligence company, the team found that users who edited a chart in the software during their free trial period were twice as likely to convert to paying customers as those who didn’t and that that number went up even higher when a trial user edited two charts. So what did RJMetrics do? They made the editing of a chart a key step in their new user orientation.

These distinctive behaviors and preferences can be hard to uncover, in part because sometimes they are so unexpected; paradoxically, you often don’t know what you’re looking for until you find it. Take how Yelp discovered that its most avid users were drawn to the site because it allowed them to write reviews: they didn’t know they were looking to tie review activity to repeat use; it was an insight that emerged by sifting through reams of website data.

Pinterest, which in its original incarnation was Tote, a mobile commerce app, pivoted to relaunch as a discovery and sharing site when Ben Silbermann saw that Tote users weren’t making purchases as intended, but instead were stockpiling massive collections of things they coveted on the app. With this knowledge, Silbermann changed course to design a product that made it easy to display these valued collections on the Web.

Such was the case for HubSpot, which sells enterprise level customer relationship management and marketing software. By rigorously analyzing user data, it discovered that customers who went through up-front product training were retained much longer than those who didn’t. So the company changed its sales policy to make paid product training a mandatory part of the new customer experience.

One of the most important changes they made was updating the new user experience (NUX) to focus heavily on helping users find friends; whereas in the original version of the NUX, the find-friends step was just one part of the overall orientation to how to use Facebook, now it was made the primary one. The growth team ran numerous experiments that stripped away more and more extraneous information from the new user starting pages and focused their attention on ways to help that new member quickly build his or her network, such as importing one’s email contacts to find friends already using the service and by stripping away other information that only distracted new users from finding friends.

They had considered employing several growth hacks to drive more adoption. For example, they thought about requiring people to whom users sent photos to also sign up for the app in order to download the photos. But they decided against that because they were afraid it would annoy people. But remember that growth hacking involves more than picking from a menu of hacks; it is, rather, a process of continuous experimentation to ensure that those hacks are achieving the desired results. If they were truly practicing growth hacking, they would have run a test to determine whether or not their assumption was true. Instead they kept focusing on improving the product; for example, by offering a feature that sent users an email with photos they’d taken on the same day the prior year. That greatly increased the number of users who started to visit the app daily. But with their goal to generate more income, increasing daily active use wasn’t the metric they needed to be focusing on. Their urgent requirement was to increase the number of subscribing users, not to make the users they already had more active.

You must be rigorously scientific in identifying the kind of growth you need and the levers that will drive it. Especially in the early phase of growth, you must set a highly disciplined course for experimentation that focuses intensely on the most important levers to achieve your goals. Speed of testing alone isn’t the goal; scattershot experimentation is a sure way to waste time and effort, and that’s true even if you’re testing at high tempo.

Instead, small teams must focus on those tests that promise to have the highest potential for impact first.

For Airbnb, the North Star was nights booked. No matter what the team did, from getting more email subscribers or registering more users, if it didn’t improve the number of nights actually booked, it wasn’t increasing the number of aha moments users experienced, which for guests was staying in a place they were happy with,

“If you can’t be extremely clinical and extremely unemotionally detached from the thing that you’re building, you will make these massive mistakes and things won’t grow because you don’t understand what’s happened.”

Alex Schultz notes that the team at Facebook’s early charge to improve one key metric, which Mark Zuckerberg decided was monthly active users, helped the growth team break out of “analysis paralysis.” Doing more and more data analysis can be an especially alluring trap; because delving into data is scientific, we can convince ourselves that we’re just being rigorous and don’t want to experiment without sufficient evidence of likely success.

Learning more by learning faster is also the goal—and the great benefit—of the high-tempo growth hacking process.

The companies that grow the fastest are the ones that learn the fastest. The more experiments you run, the more you learn. It’s really that simple. The high volume is ideal, because most experiments fail to produce the results you’re hoping for.

Recall that the stages of the process are: data analysis and insight gathering, idea generation, experiment prioritization, running the experiments, and then returning to the analyze step to review results and decide next steps, in a continuous loop.

To start probing for areas of growth opportunity, they formulate a set of questions to guide their analysis, as follows: WHAT ARE MY BEST CUSTOMERS’ BEHAVIORS? • What features do they use? • What screens in the app do they visit? • How often do they open the app? • What items do they buy? • What is their average order size? • What time of day do they shop and on which days?

Over the four days following the team meeting, all members should submit as many ideas as possible for hacks to try to improve the revenue from the app users. Self-censorship is discouraged, and nothing should be considered too crazy to put out there.

Ideas should be submitted to an idea pipeline, following a templated format by which they should be submitted. It’s important to standardize the format so that ideas can be quickly evaluated, without the team needing to ask lots of questions. Instead of vague suggestions like “Our sign-up form is too hard; we should make it easier to sign up,” submissions must articulate exactly what change is to be tested, the reasoning behind why that change might improve results, and an explanation of how results should be measured.

When submitting ideas, the submitter should rate each idea on a ten-point scale, across each of the following three criteria: the idea’s potential impact, the submitter’s level of confidence in how effective it will be, and how easy it will be to implement. Then those ratings are averaged to provide an aggregate score for each idea. The entire bank of ideas are then ranked by their scores, and the team begins experimentation with the highest-scored ideas in the area of focus chosen by the growth team. For example, a highly rated customer acquisition idea will be passed on in favor of a lower-scored idea around retaining users if the growth team is currently focused on improving customer retention.

At GrowthHackers, we once ran a simple experiment that involved moving the location of a sign-up form to receive our weekly “Top Posts” email newsletter. We had originally put the form at the bottom of our home page because we thought that users would want to evaluate the content on the site—i.e., scroll through the feed of trending posts that we feature on the home page—before they could decide whether they wanted to sign up to get the newsletter. Then Morgan had the humble idea to move the invite from the bottom to the top, giving it more visibility. Truth be told, he wasn’t sold on the impact the change would have, assigning it a score of 4 out of 10. But we decided to test it nonetheless because he gave the experiment a 9 for ease based on feedback from the engineering team, who said it was relatively easy to implement, and on his confidence that it would improve email sign-ups to some degree by being more visible, giving it a confidence rating of 8 that it would do so. The results were staggering. Much to our surprise, we saw a 700 percent jump in sign-ups, far exceeding our humble assumptions about the experiment’s potential impact on our growth.

It’s vital that this meeting not be used for brainstorming ideas; ideas should be brainstormed well before using the process described above.

GROWTH FOCUS AREA: What part of the user experience or growth lever is the team focused on? Are there any short-term objectives that the team needs to work toward? If the focus is staying the same, this is a simple confirmation. If it’s changing: say, a shift from user acquisition to retention or monetization, a discussion of the new focus area and the reasoning behind the change should take place. Short-term objectives in service of these goals should also be noted, such as a short-term goal to get a certain percentage of app users saving items to their shopping lists.

simply asking them how they describe your product and its value to their friends or colleagues will inevitably elicit some potentially powerful language or phrasings.

Sometimes the changes in wording you arrive at will lead you to additional changes to make, not only in your copy, but in your overall branding and maybe even in the nature of your product itself—one of the reasons why growth hacking teams should consist of product developers and engineers as well as marketers, sharing data freely among them.

In stock market investing, experts agree that it’s best to spread your money across a wide swath of diverse types of businesses and sectors. But this is not the right strategy when it comes to finding the channels for marketing and distributing your product (which in Web business are often one and the same). Marketers commonly make the mistake of believing that diversifying efforts across a wide variety of channels is best for growth. As a result, they spread resources too thin and don’t focus enough on optimizing one or a couple of the channels likely to be most effective.

There are two phases in which to home in on your best channels: discovery and optimization. In the discovery phase, the growth team should experiment with a range of options, and this does not mean trying all sorts of things haphazardly to see what sticks. Channels must be researched thoroughly, then prioritized down to a few to target for experimentation, and we’ll introduce a simple but hugely helpful method for doing that in just a moment. Once you have found those one or two with the right fit, you can move to the second phase, optimization, in which you should be working to maximize both the cost-effectiveness and the reach of your channels as you keep scaling up.

That was the case with Dropbox. Services to help people easily share and store files online were brand-new when the company launched, so people weren’t searching Google for the solution Dropbox was offering, which was a key reason that the effectiveness of paid search ads was so limited. The referral program solved this problem.

Cost—how much you expect to have to spend to run the experiment in question. • Targeting—how easy it is to reach your intended audience and how specific you can be in whom your experiment reaches. • Control—how much control you have over the experiment. Can you make changes to the experiment once it’s live? Can you stop it easily or adjust it if it’s not going well? • Input time—how much time it will take the team to launch the experiment. Filming a television ad, for example, has a much longer input time than setting up a Facebook ad. • Output time—how long it will take to get results out of the experiment once it’s live. For example, search engine optimization experiments or social media may have longer output times than a radio ad does. • Scale—how large an audience can you reach with the experiment? Television has a much larger scale than advertising on topical blogs.

First, they do another analysis of their user data. They’ve been monitoring the data continuously, of course, keeping a close eye on the metrics that matter most, but whenever a team shifts focus to a new growth lever, it’s important to dive into the data with fresh eyes looking for insights specific to their new mission. Recall that they had discovered earlier that a large number of their best customers were coming from the grocer’s main website, and that’s still true. So they decide that they will focus on organic ways of leveraging the website more powerfully as one key channel, and will also experiment with new channels to help them cast their net wider and bring in more users who aren’t regular visitors to the website, as was done at the initial launch.

ORGANIC • Improve app merchandising on main website • Email regular shoppers who have loyalty cards but haven’t downloaded the app with messaging about the benefits of shopping via the app • Add a full page promoting the mobile app that pops up for website visitors when they access the site on their phones, also known as an app-install-roadblock page PAID • Run Facebook app install ads • Run a set of radio ads based on the success from their initial launch campaign • Retarget website visitors with ads to download the app, meaning: run a Web ad campaign shown only to previous website visitors VIRAL • Create a friend referral program for existing app shoppers that leverages their desire for additional discounts in exchange for inviting friends

Asking so little of users is not always possible, though; you’ll often need to offer users an incentive. The best way to do this is to create a double-sided incentive, that is, one that offers something to both the sender and the recipient. If you have a high payload, you may not need as compelling an incentive in order to get good results because even a fairly small percentage of responses will add up nicely. But if your payload is low, you’re likely going to need a more compelling incentive, for both parties, to drive up your conversion rate and frequency.

That’s why doing the legwork to learn about how your customers use your product and where potential loops can be created and optimized is essential for tapping into viral growth driven through network effects.

the team saw a massive spike in friend invites sent out when they integrated the referral program into the new user experience, whereas before it had only been accessible from a small link tucked into the corner at the top of the website home page.