MARKETING ## DEFEND YOUR VISION COLUMN ## The key to effective RESEARCH ## STATEMENT ## Carter Roberts on mobile discounts In the afternoon, the Where did the strategy and climate moral slope gets graphical icons on change slipperier our tech gadgets come from? New Thinking, Research in Progress hbr.org STRATEGY Making “Freemium” Work Many start-ups fail to recognize the challenges of this popular business model. by Vineet Kumar ver the past decade “freemium”—a com- bination of “free” and “premium”—has Obecome the dominant business model among internet start-ups and smartphone app developers. Users get basic features at no cost and can access richer functionality for a subscription fee. If you’ve networked on LinkedIn, shared fles through Dropbox, watched TV shows through Hulu, or searched for a mate on Match, you’ve experienced the model firsthand. It works for B2B companies as well—examples include Box, Splunk, and Yammer. Several factors contribute to the appeal of a freemium strategy. Because free features are a potent marketing tool, the model allows a new venture to scale up and attract a user base ILLUSTRATION: SROOP SUNAR ILLUSTRATION: May 2014 Harvard Business Review 701 IDEA WATCH LATE ADOPTERS ARE MORE PRICE-SENSITIVE AND SEE LESS VALUE IN UPGRADING EARLY ADOPTERS ARE WILLING TO PAY FOR THE PREMIUM OFFERING without expending resources on costly ad campaigns or a traditional sales force. The monthly subscription fees typically NEW FEATURES ARE charged are proving to be a more sustain- INTRODUCED, DRIVING NEW CONVERSIONS able source of revenue than the advertising model prevalent among online frms in the early 2000s. Social networks are powerful SHARE OF word gets two gigabytes of cloud-based drivers: Many services ofer incentives for PREMIUM USERS storage free. If people run out of space, referring friends (which is more appealing they can pay $9.99 a month (or, alterna- when the product is free). And freemium THE LIFE CYCLE OF UPGRADES tively, $99 a year) for 100 GB of storage. The Freemium companies typically see the is more successful than 30-day free trials free version is adequate for basic docu- share of paid subscriptions (and therefore or other limited-term ofers, because cus- ments, but anyone who wants to back up cash flow) rise and fall and then rise again tomers have become wary of cumbersome photos or other media quickly hits the limit, in a predictable way over time. Those cancellation processes and fnd indefnite and the reasons to upgrade are obvious. that don’t account for this phenomenon free access more compelling. For many LinkedIn users, the advan- risk failure. But despite its popularity and clear tages of upgrading are murkier. I’ve used benefits, freemium is still poorly under- LinkedIn for several years to keep in touch stood. It has inherent challenges, as dem- with colleagues, and I routinely receive e- onstrated by the many start-ups that have mails urging me to upgrade—but the ongo- tried but failed to make it work. For several ing value of doing so is not apparent. (The years I have studied freemium models in company offers four premium subscrip- depth—by coauthoring (with my HBS col- tions, some aimed at specific customer leagues Bharat Anand, Sunil Gupta, and Fe- segments, such as recruiters or salespeople, lix Oberholzer-Gee) a case on the New York and most featuring deeper search function- Times’s paywall strategy (a variation of the rich, and it’s time to cut back. This kind of ality, better e-mail capability, and more vis- freemium model); conducting a deep dive tuning was evident at the New York Times ibility into who has viewed your profle.) into the user data of a storage and synchro- website. After years of unrestricted access, Although LinkedIn is successful—it was nization company; and coauthoring (with in 2011 the paper began limiting users to 20 one of the frst freemium companies to go my HBS colleagues Clarence Lee and Sunil free articles a month; people had to sub- public—it could probably monetize more Gupta) a working paper on how freemium scribe if they wanted to read more. Over users if the distinctions between its free companies can use referrals to spur usage subsequent months the company realized and paid oferings were clearer. and upgrades. Through this work I’ve come it was still giving away too much and was What is your target conversion up with six questions that start-ups consid- getting too few subscribers as a result, so in rate? Imagine that you’re the CEO of a free- ering a freemium model should ask. 2012 it cut the number of free monthly ar- mium start-up and you’re handed a report What should be free? Let’s say you’ve ticles to 10. Start-ups should expect to need showing your conversion rate (the percent- created a digital product that has 20 fea- similar tweaking to find the optimal bal- age of free users who have upgraded to a tures and you’ve chosen five that will be ance between trafc and paying customers. premium plan) for the most recent quarter. free to anyone who registers on your site. The balancing act can be tricky: Users may What fgure do you hope to see? Users who want the other 15 will have to revolt when asked to pay for things they are A rate of 1% is probably too low, espe- pay. How do you know whether you’ve accustomed to getting free. cially if you rely on subscription revenue made the right choices? And if you suspect Do customers fully understand the alone. (Some players, including the New that you haven’t, what should you do? premium offer? Communicating two sets York Times and LinkedIn, also collect on- Recall that one of the chief purposes of of benefts complicates your marketing ef- line ad revenue.) It signals either that too freemium is to attract new users. If you’re forts. If customers don’t clearly grasp what much of what you’re providing is free—giv- not succeeding with that goal, it probably they would gain by upgrading, you will ing users little reason to upgrade—or that means that your free oferings are not com- monetize fewer of them than you other- consumers don’t understand or value your pelling enough and you need to provide wise might. premium features. more or better features free. If you’re gen- Dropbox and LinkedIn are a study in But, less obviously, a very high conver- erating lots of traffic but few people are contrasts. The former has attracted 200 sion rate isn’t necessarily good. Remem- paying to upgrade, you may have the op- million users with a simple proposition: Ev- ber that one of the benefts of a freemium posite problem: Your free oferings are too eryone who enters a username and a pass- model is the ability to generate traffic. 702 Harvard Business Review May 2014 HBR.ORG Suppose that 50% of the users of your free early adopters of LinkedIn, because their of your free users, which takes two forms: product upgrade to premium. You might business depends on their ability to iden- Some of them become subscribers, and think that your model is working well; but tify and connect with professionals.) So some draw in new members who become perhaps your free product is not very com- over time, conversion rates typically dip subscribers. In our HBS working paper and pelling, which will limit your potential ac- as the user base expands to include people in ongoing research, we have found that a quisitions. All other things being equal, you who are more price-sensitive or who see free user is typically worth 15% to 25% as would do better to convert 5% of 2 million less value in the service. Although free- much as a premium subscriber, with sig- monthly visitors, for example, than to con- mium companies universally have a very nifcant value stemming from referrals. We vert 50% of 100,000 visitors. The best long- low marginal cost for each new user—oth- have also found that frms can increase the term strategy is generally to aim for a mod- erwise the model wouldn’t make sense— value of referrals by carefully managing erate conversion rate (in my research, I’ve those costs aren’t zero. At a minimum, free referral incentives and communications. If found that most companies’ range from 2% users put demands on server space and you’re considering a freemium model, pay to 5%) coupled with a high volume of traf- customer service. close attention to why and how satisfied fc. If you’re targeting a small market, you Companies that fail to understand these users might help your product go viral. should aim for a higher rate. realities may feel a cash crunch as the num- Are you committed to ongoing in- Are you prepared for the conver- ber of free users grows and the cost of ser- novation? It’s a mistake to see freemium sion life cycle? Let’s assume you’re at- vicing them therefore rises. This is often merely as a customer acquisition tool and tracting plenty of traffic and new users, the reason that companies launched with to drop the free version when new custom- and your conversion rate is 5%. You want to freemium models pivot away, converting ers stop coming in or when the upgrade forecast growth and revenue. Can you sim- to free time-limited trials or eliminating rate dives. Users who join late are typically ply draw a couple of straight lines, on the free offerings altogether.