We’ve seen a lot of companies struggle to achieve viral growth: Aardvark, Triangulate, Cake.

What’s going on? Does LTV have a skewed sample, or is it hard to harness viral growth?

by Jeanne Hwang

After three afternoons of work, we got an abysmal 10 sign-ups. The challenge was to create a viral campaign to get as many new members as possible for Gilt’s JetSetter website. Our campaign was to draw travelers into a contest to match photos with a destination, share their own photos and enter a raffle for becoming a member. We knew virality wouldn’t be high, but we thought, “boy, don’t you want others to see your awesome travel photos and share this great opportunity with your best friends?”

The answer was NO. Our viral coefficient was effectively zero (virality (K) = number of invitations sent (i) * conversion %), as new members had no incentive to increase the number of raffle participants, which decreased their chances of winning. It seems obvious now, but we hoped that we could attract most of the new members through our own outbound marketing (Twitter, Facebook, blogs) supplemented by friends who would share with friends because of course you’d want to help out your friends!

In Cake’s case, much of it was due to the inherent nature of the target market (“men in their 40’s just don’t want to share,” as Steve said at lunch). But Jetsetter’s target market and Triangulate’s target market are young, internet savvy consumers. In fact, in Triangulate’s case, dating is an inherently social business, yet I suspect that virality ended with the first cycle of outreach. Sunil said in class that non-single people didn’t want to be associated with a dating site – the wingmen had very little incentive to send out invites to their friends to join Triangulate, as they wanted minimal association with it.

It seems viral growth is only possible when it’s at the juncture point of three scenarios:

If a user shared a raffle link with a friend, their value to the friend would increase, but it would decrease their own chance and thus, the value to themselves.

Asking a friend to rate a hot-potential-date’s-photo on Triangulate would increase the value to the dater, but the friend risks being associated with the dating site by sharing that photo one more time.

This is another way to think of what David Skok states as the key to viral growth, which is to entice them and address their concerns. So the question is, in businesses that have inherent contradictions, is there a way to decrease the effort or the risk, in order to achieve virality?

Read More

Understanding When to Launch

by Katherine Nadler

For an entrepreneur, one of the most consuming decisions is determining when to launch. Arguably, a product is never complete; therefore, it is up to founders to determine what level of incomplete is acceptable. To do so they must allow their business model and market context to drive their decision making. Asking the right questions is very important.

Does the business model rely on virality? When a product anticipates customer acquisition through virality it makes sense to launch early, sacrificing some of the quality that extra time would afford. As much as entrepreneurs may convince themselves that their product is perfectly positioned to generate “word-of-mouth” or network effects, the best way to know for sure is to test that hypothesis in practice (often first through a beta launch). While the world’s become inherently social as a result of game changers like Facebook, Zynga, Twitter and Foursquare, it does not mean that products will effortlessly generate the same social response.

How damaging is a “buggy” product to my company’s value proposition? Within different businesses, the same mistakes can have very different impacts. Using Airbnb as an example, a glitch in their website where a message to an owner got lost would be annoying but would not impact the value proposition of the company. Users would still be able to find places to stay and base their opinions on that experience. On the other hand, lost files on Dropbox would be nowhere near as innocuous a mistake. This is because Dropbox draws attracts users with its claim of simplicity, reliability and safety. Therefore, when thinking about when to launch, it is important to consider the deal breakers for customers and prioritize those above all else.

What is the competitive landscape? Ventures that are entering an existing market have a completely different set of challenges from one looking to enter a new market. In an existing market, the product must be superior to those offered by established players; this can be very difficult for a resource-constrained young startup. However, in a relatively new market there might be some benefit to actually letting others enter the market first. While letting competitors grab user attention first seems counter intuitive, looking at how financial account aggregator Wesabe lost its market to Mint.com provides some interesting perspective. In a blog written by the former co-founder of Wesabe, he reflects on how “There's a lot to be said for not rushing to market, and learning from the mistakes the first entrants make. Shipping a ‘minimum viable product’ immediately and learning from the market directly makes good sense to me, but engaging with and supporting users is anything but free. Observation can be cheaper.” When entering a market where someone has gotten there first, rushing into the market prematurely with a subpar product can be very damaging while waiting can offer great rewards.

Every situation will likely have many considerations to weigh when deciding when to launch. In my opinion, founders must be careful not to jump to conclusions. Just because a founder believes in the lean startup methodology does not mean he or she should introduce a flawed product. Additionally, founders must consider all different stages of launching and realize that they have the power to separate the launch of their product from the launch of their venture in the press. At the end of the day, the best that anyone can do is make an informed decision and hope for the best.

Read More

Thinking About Virality Before It's Too Late

by Jesse Garcia

While generating viral growth may be hard, it is not difficult to understand ways to increase the likeliness of a product going viral. Some of the companies discussed in Launching Technology Ventures, including Triangulate and Cake Financial, struggled with the challenge of generating viral growth. For these companies, virality became a priority after having developed a business model and begun product development. However, there are opportunities to increase the likeliness of viral growth when developing the company’s value proposition, and so entrepreneurs should think about designing for virality from day one.

The first consideration in determining the virality of a product is the potential benefit from sharing it. Very simply, sharing occurs between the sender and the recipient. These benefits can include social currency, enhancement of social status, entertainment, education, and potentially economic rewards, and must be experienced by both sides. Social currency, which we can define as the utility derived by a sender from benefitting the recipient, is the common denominator present in all cases where viral growth occurs. Triangulate struggled to achieve virality because in dating, an individual can experience both successes and failures, leading to limited opportunities for social currency. Similarly, Cake Financial failed to achieve virality because investing is also an activity where an individual can experience failures. Cake Financial’s struggles were exacerbated by the fact that there was no economic reward to sweeten the sharing opportunity, as there is ample evidence that the investment performance of active retail investors cannot consistently beat the market.

The other variable in the virality equation is the costs associated with sharing. These can include time and effort. Startups can minimize these costs in many ways, such as integrating with Facebook and other existing social networks. Triangulate and Cake Financial both took advantage of opportunities to minimize the cost of sharing, leaving limited room for improving virality by lowering the cost of sharing.

The result is a cost-benefit analysis of sharing, and it is abundantly clear that there must be a net benefit to sharing for a product to go viral. As explained above, Triangulate and Cake Financial failed to solve the benefit part of the equation. Importantly, the options available to either company to provide this benefit were most numerous before the business model was in place. To increase the likeliness of viral growth for a product, a startup should consider accounting for virality, and the necessary positive net sharing benefit, in its value proposition.

By thinking about it earlier, the virality dilemma becomes easier to solve. For example, trying to change users’ behaviors will not likely lead to a positive net sharing benefit. By simply observing existing human behaviors and ascertaining whether a particular activity is inherently social, a startup can begin solving the dilemma. Moreover, a study of human emotions and heuristics could also provide important insights into whether a NEW activity can truly become social. The over-confidence bias, which is based on numerous surveys where more than 50% of a population believes that he or she is better than average, suggests that users do not tend to see themselves as failures and may not want failures associated with them (as in dating or investing). This bias helps us understand why Triangulate and Cake Financial struggled to go viral.

Read More

Virality Challenges

by Kara Yu

We’ve seen a lot of companies struggle to achieve viral growth: Aardvark, Triangulate, Cake. What’s going on? Does LTV have a skewed sample, or is it hard to harness viral growth?

It’s rare at HBS for us to see failures, but in LTV, we were fortunate to see not only one but several cases where a little bit of founder exuberance and false positives resulted in a lot of work on a product was not inherently viral. In all these cases, the founders created a product that they thought they would

Aardvark had issues with its product/market fit. While I was always a fan of Aardvark and thought it was a great idea to connect users with experts in their social group, it seems like most people thought otherwise. Even though it was popular with initial users, it was never able to cross the chasm. Ultimately, Aardvark solved a problem that users did not have. Some people found it helpful, but no one found it crucial to their lives. Aardvard also made assumptions about people’s tendencies to give back to a community, and share information. By not adequately using lean startup methodology for its customer development process, Aardvark had created a great product that people did not need. The virality of this product was way overestimated because people just did not find it useful.

Cake also had a great idea in a market that was ripe for innovation. While a similar product achieved great popularity (mint.com), growth for Cake stagnated. Even though Steven been diligent about getting validation from industry experts and top Angels, he neglected to receive feedback from his most important audience about the actual product. While conceptually, the idea was great, implementation depended a lot on the user interface. Due to his focus on the backend and neglect for the front end, the product was just simply not useful.

In the end, while virality is very difficult to predict, there is a necessary but not sufficient condition that in order to achieve virality, the product has to be either “cool” or useful to the user. I would never recommend anything to my friends that they don’t find useful. While I am a huge fan of Buzz and Google Latitude, I never mentioned them to my friends because I knew that they wouldn’t have appreciated them. People tend to only make product recommendations that improve or at least maintain their social standing with their friends. Dropbox managed to keep fairly low CAC because it was able to be useful enough to users that they told all their friends. The benefit that both parties received (in terms of extra storage) was a great motivator but had it been a less helpful product, it would not have mattered.

In short, virality is hard to achieve but at least in the cases of Cake and Aardvark, the main problem was product market fit and not their word of mouth marketing strategy.

Read More

Referral Marketing is Sick - Use it to Go Viral

by Jonathan Krieger

In startup circles Viral Marketing is often discussed as a low cost way to acquire new customers and rapidly scale a business.

David Skok of Matrix Partners uses this blog post to break viral marketing into two components:

1. Viral coefficient – Number of new customers acquired for each existing customer
2. Viral cycle-time – Length of time required to acquire a new customer 

The higher the viral coefficient and lower the cycle time, the more “viral” the company and the faster it will grow.

One way that companies can impact their viral coefficient is through referral marketing. Referral marketing uses incentives to encourage specific customer behavior- in this case new customer acquisition. Referrals occur when an advocate recommends your product or service to a friend who becomes a new customer. A 2010 study found that 83% of satisfied customers reported a willingness to recommend services to others, however only 29% of customers actually do so. Referral marketing aims to close this gap, providing incentives that motivate the remaining 54% of satisfied customers to become advocates.

Referral incentives are the rewards offered to motivate advocates, and can come in a variety of forms:

Monetary incentives are easy-to-quantify rewards that can be directly linked to financial gain. These include cash, gift certificates, company credit, and future or retroactive rebates and discounts. 

Non-monetary incentives are more opaque rewards that are still desired by your advocates. These may include special access to products, experiences (i.e. consultations/training) and information or public recognition and credit (e.g. badges/certificates).

Companies should use incentives that are:

1. Tailored to their target advocates 
2. Encourage customer engagement and 
3. Allow for profitable customer acquisition cost (CAC) 

According to David Skok, best practice for recurring revenue businesses is to target CAC that is no more than 1/3rd of customer lifetime value (LTV)*

A University of Chicago study found that non-cash incentives were 24% more effective at boosting performance than cash incentives.* While cash may be a simple, and cost effective way to encourage referrals, companies should think critically about how they can move beyond cash to reward advocates in meaningful ways.

Cloud based storage service Dropbox provides a best-in-class example of how referrals can drive business growth. In April 2010 an impressive 35% of new Dropbox signups came through referrals.*** Their referral program offers an additional 250 MB of storage space to both the advocate and friend for each new signup. This reward meet’s all the aforementioned criteria. It is highly desired by Dropbox users who quickly fill-up the 2 GB included in their free trial. It also encourages deeper customer engagement; making Dropbox more “sticky” as customer use it to store and share more data, and it is extremely low-cost (approx. $0.014 per month – based on Amazon S3 pricing).

Dropbox also uses a very effective two-sided (win-win) reward system to give extra storage space to both advocates and friends. One of the main factors contributing to the success of referral programs is their ability to leverage pre-existing trust between advocates and friends. Introducing referral incentives to this relationship can create skepticism, distrust and misalignment. “Are you referring me to Dropbox because I’ll love it, or because they’re paying you to do so, and what does this say about our relationship?” Transparently rewarding both parties, resolves this problem, and according to a 2007 study in the Journal of Marketing this two-sided system is the preferred means of close-tie referrals.

By using incentives Dropbox increases the number of successful referrals and, thereby, it’s viral coefficient. Several other internet startups including Fab.com, Birchbox and Gilt Groupe also use effective referral marketing to drive business growth. While the success of referrals will vary across companies and industries, referral marketing can be a highly effective customer acquisition tool that deserves a place in any company’s marketing mix.

* Skok, D. “Startup Killer: the Cost of Customer Acquisition,” For Entrepreneurs, 2009.
** Scott, J., “The Benefits of Tangible Non-Monetary Incentives” University of Chicago, 2004. http://www.forentrepreneurs.com/startup-killer/
***Eisenmann, T. Pao, M, Barley, L. “Dropbox: “It Jut Works,” Harvard Business School, 2011.

Read More

Harnessing Viral Growth

by Ernesto Humpierres


Achieving viral growth is probably one of the hardest things to do by a new venture. In my view the problem lies in the fact that there is so much choice for every conceivable functionality or solution that users are looking for. The day has only 24 hours and there is literally no time for users to try out all the apps they want, and pick up yours out of a haystack. Given this context, I want to discuss three elements (or dimensions) related to what could be impacting viral growth. These elements are: the degree of social validation, the advantages of the “stickiness” of a particular solution, and finally the ease of transmission of the solution.

At first it seems counterintuitive that social validation is important to something that is “viral”. The first perception I get from the concept of virality is that is something brewed from the ground up, without the need of hierarchies or leadership; it’s about the “people’s” choice, right? Well no, in an era of choice and mass media I believe we need those social validators more than ever to actually go viral, be it from a large specific constituency (i.e. Harvard students in the case of Facebook), or public personality (like Charlie Sheen attracting 1 million followers to his Twitter account in 24 hours, or Conan O’Brien saying that the one-liner has a business model when referring to Twitter). But wait; do not confuse social validation with need of awareness, think of Google’s highly visible efforts to dominate certain tech solutions, and no marketing money will buy them success (do you recall Google video?). So the key in my mind is to pick a constituency or group that can act as social validators and target them intensely. Think about it, targeting 20k Harvard students is probably more focused and effective than targeting the whole world wide web, and once you get the social validation from a prestigious group like that you can go on a differentiate yourself from the other dozens of applications out there similar to yours.

This leads me to the second element, the “stickiness”. The more skin in the game I have in an application the more I will be engaged with it and the higher the network effects. Facebook is sticky, the amount of memories, pictures and emotional baggage stored in its servers is way to big to abandon once you are caught. In the case of IMVU, well… it reminds me of Talk City back in the late nineties: there was really nothing forcing me to keep up with it as soon as I left the last chat room. So my recommendation is to force the creation of that stickiness. Think of Dropbox, once you have 2GB of important information up there you are committed to it. Just thinking about transferring and organizing that data again gives me a headache, plus information stored there can only grow.

Finally I want to refer to ease of transmission and adoption. The easier it is to pass the application to someone else the higher the virality coefficient. A YouTube video link is extremely easy to pass on, but convincing someone to fill up the 2-hour eHarmony questionnaire will definitely impact virality. Actually eHarmony realized how important this is and they have been working throughout the years to reduce the time it takes to finish open a profile in their site. Also, the fact that different people use different platforms means that passing the information will be challenging. In regards to the first problem I think LinkedIn found the solution. Joining is very easy, and after that the process of uploading and filling out more data in your profile is modularized in sub-processes, so to speak. You can fill out specific data and complete a full sub process in a short time, and then do that multiple times over multiple sessions to complete the whole process. So by segmenting tasks you make it easier for the user to become engaged with the application. Maybe I can dedicate 15 min for a task in LinkedIn but not for a 2-hour questionnaire in eHarmony. The solution to the second problem requires multi-homing, and I really don’t think there is currently and easy solution for that. 

Read More

Virtual Virality

by Natasha Prasad

In the summer of 2009, Zynga launched Farmville, an unglamorous, copycat game that no one could have predicted would one day become the most popular application on Facebook.  Today, Farmville has 80 million monthly users and yet, for all its successes, Zynga games continue to be called everything from “ludicrous” and “mind-numbing” to “spammy” and “exploitative”.  Love them or hate them, Zynga has exploded and the lean startup philosophy, it seems, is plastered all over it: customer-centric design, split-testing, rapid iteration, data driven decision-making and, perhaps most impressively, viral engineering.

As startups grapple with that fundamental challenge of generating long term value in excess of customer acquisition costs, Zynga, it appears, has found the Holy Grail.  Campaigns, SEO, partnerships, Facebook ads, CRM – none of these buzzwords come cheap and, with a non-zero churn rate, it is tough to extract sufficient value from a user before he/she jumps ship.  Virality, on the other hand, is free, rapid and self-sustaining.  Contrary to its cringe-worthy popular usage, virality does not just mean “explosive growth”; rather it refers to the number of additional users brought in by each new user on your site. David Skok deconstructs viral growth into a Viral Coefficient, K and Viral Cycle Time, T.  

·       K = X (number of invites) *Y (acceptance rate %)

·       T = time between a new user discovering the site, inviting another user and that second user coming to the site.

So how do you make your site viral?  Can’t you just bump up X and Y and dial down T?  And why have so many lean startups struggled with virality?

Triangulate’s Sunil Nagaraj was so confident in the viral capabilities of his wingman-based dating concept, that he christened his startup Wings.  He later abandoned this angle (“pivoted”) in favor of a better matching engine.  Aardvark’s Ventilla and Horowitz created product evangelists (“Aardvocates”) and encouraged users to broadcast their answers on social platforms.  Neither strategy worked.  Dropbox, on the other hand, saw 4 million users generate 2.8 million referrals; thirty-five percent of 2010 signups originated from referrals and 20% from shared folders.   Why?

To draw on learnings from another HBS class, we can think of it in terms of the difference between “meet” products and “friend” products.  Online dating is a meet product: you are there to connect with people you don’t already know; roping in your offline friends adds little value to your situation.  DropBox, on the other hand, is a friend product – you want to share folders with your friends and so you actually derive value from having them join the network.  Building products that are intrinsically viral requires, in my opinion, scratching beneath the technology stack and developing a deeper understanding of human psychology and sociology.  It also depends, somewhat tautologically, on the nature of the solution itself. 

That said, adding a layer of virality to a product or site appears relatively straightforward, especially when you have some kind of currency to reward successful conversions with.  Paperless Post (coins) and Gilt Groupe (free shipping) have harnessed a more traditional approach while Groupon (deal tipping) and Living Social (group discounts) have been slightly more creative.

This aspect of virality, I believe, is where lean startup practices can help.  Hypothesis testing and iterative development can help introduce and manage viral layers, while rich customer data can help inform social strategy.  Maybe it won’t lead to Zynga-esque gangbusters, but at the end of the day, there’s more to startup life than virtual virality. 

Read More

Achieving Virality Is Hard

by Krzysztof Jedrzejek

When I think about consumer internet businesses in the context of virality, I split them into 3 main categories:

-      Fundamentally viral – usually multi-sided platforms that require a large user base to deliver value to the customer (like Facebook, eBay or Skype)

-     Potentially viral – where the user has to be motivated/incented to create a viral loop, but there are little network effects (Gmail or Gilt)

-     Lost causes – products that seem ‘anti-viral’ by nature (from our class - Cake, that soon realized large part of their customers do not want to share their investment ideas).

I believe achieving virality even in the first 2 categories is a complex task.

Although social networks have the best shot at exponential growth, they face the ‘chicken and egg’ problem – if the product does not work ‘standalone’ (without my friends on it), and they are not excited enough to join quickly after I invite them, I might abandon the product altogether.

What can you do to boost viral growth?

Build virality into the product

Expanding on David Skok’s viral coefficient equation (K):

K = (% of users that send invites) * avg. # of invites per user * click-through rate * (% of new users that will register)

Each element of the above funnel is measurable, and the product manager’s role is to make data-driven decisions on where to invest most resources.

I can see optimization techniques with different level of complexity.

For example, average number of user can be improved through a plug-in importing the Outlook address book or Facebook friends list. At an average of 130 friends that’s a good start to fill the top of the funnel.

In my understanding click-through aggregates delivering an email, opening it and actually clicking on the link. First is driven by navigating around spam filters, second by a catchy subject, and third by great design of the email body. A/B testing could be used to measure which block requires tweaking. Same for the registration part, which could be optimized with a simple and clean U/I with as few steps as possible.

Provide incentives

I immediately think of Gilt with their 25$ credit for every active user that you recommend. That’s a substantial amount directly adding to their CAC, but has worked great to dynamically grow user base. However, with significant profit margins Gilt can afford it, many of the consumer internet startups that haven’t figured out monetization yet don’t have that luxury.

Use platforms/ word of mouth

For the 2^nd category of products (no network effects), virality can be supported by leveraging ‘streaming’ platforms, such as Facebook or Twitter. If the application posts something interesting on my Wall, it immediately gets view by hundreds of my friends.

Word of mouth can be a powerful and cost-effective technique to grow the user base, sometimes even in a counterintuitive way. Here, again Gilt comes to mind – by starting out as an exclusive/’invitation only’ site they managed to create a lot of buzz and motivate early adopters to ‘seek’ invitations and provide initial traction to the website.

Finally…

Can those factors lift the K coefficient above 1 (ensuring exponential growth)? Looking at compounded percentages in the equation, it still seems very difficult to achieve that sustainably. Even then, it could lead to saturation and further churn issues.

The key strategy could be to focus on controlled high growth phases (K>>1) interrupted by refining the product along the way.

Read More

Viral Growth

by Andrew Perlmutter

Lean startup theory makes viral growth seem like the simplest piece of building a large-scale business from scratch. At the very beginning of the process, the entrepreneur avoids wasting money while rapidly and iteratively testing hypotheses about a new venture based on customer feedback. By consistently refining the product and pivoting the business model, the entrepreneur eventually achieves product-market fit. As Marc Andreessen describes it, PMF is defined as “being in a good market with a product that can satisfy that market.” The telltale signs of PMF are always obvious to the entrepreneur: “customers are buying the product as fast as the entrepreneur can make it and the entrepreneur is hiring sales and customer support staff as fast as he can.” It is in this moment, with the business model validated, that the entrepreneur is supposed to step on the accelerator, open up the engine and scale the business as quickly as humanly possible. The result? Viral growth. To summarize, the entrepreneur reaches PMF, hits the “scale now” button (i.e. pumps in a lot of cash), and viral growth happens.

But the first five classes of LTV have complicated this version of viral growth.  In particular, each of the five entrepreneurs has followed lean startup principles, yet only Dropbox has achieved viral growth. Furthermore, many popular platform businesses such as Facebook, Twitter, and YouTube achieved remarkable viral growth despite scaling up far before the founders validated the business models.

So what does this mean? One dangerous response is to maintain an unwavering belief in lean startup theory by rationalizing the evidence from our early classes. Such an evangelist could argue that Dropbox is the only company from those close that achieved viral growth because it is the only company that reached product-market fit.  Similarly, this evangelist could also claim that Facebook / Twitter / YouTube attained PMF and validated their business models by winning the time and attention of users. Thus, Facebook/Twitter/YouTube did scale at the appropriate time pursuant to lean startup methodology. What the evangelist is really doing here is looking at whether a company has successfully scaled (through viral growth) to determine if it previously achieved PMF, and not the other way around.

This sort of logic is extremely damaging to lean startup theory because PMF is what gives the theory its predictive power. Simply put, the predictive principle is as follows; if a business hits PMF, then it can be scaled successfully. However, if we need to see viral growth before we know whether a company has hit PMF, then we have no ability to predict anything.

What’s really going on is that it is very difficult to know when to scale a business. Some businesses can/should scale before the business model is validated through monetization, but most other should not. The entrepreneur is faced with the dilemma of trying to figure out how to categorize his business. In the face of this uncertainty, lean startup theory offers the entrepreneur a set of tools that, when used correctly, improve the probability of his venture’s success. It does not present a comprehensive set of circumstances under which a business can scale successfully. It also does not ensure that the entrepreneur will achieve PMF by using the tools.  In fact, it cannot even define what PMF really is. But it does lengthen the entrepreneur’s runway and give him a scientific approach for developing a successful business.

So in my opinion, lean startup methodology is not actually a predictive theory. It is better categorized an approach to developing a startup that improves the chances of success.

Read More