The Dangerous Seduction of the Lifetime Value (LTV) Formula

Posted on September 4, 2012. Filed under: advertising, Internet, Uncategorized, Venture Capital, Web/Tech | Tags: , , , , , , , |

Many consumer Internet business executives are loyalists of the Lifetime Value model, often referred to as the LTV model or formula. Lifetime value is the net present value of the profit stream of a customer. This concept, which appears on the surface to be quite benign, is typically used to compare the costs of acquiring a customer (often referred to as SAC, which stands for Subscriber Acquisition Costs) with the discounted positive cash flows that will come from that customer over time. As long as the sum of the discounted future cash flows are significantly higher than the SAC, then people will argue it is warranted to “push the accelerator,” which typically means burning capital by aggressively spending on marketing.

This is a simplified version of the formula:

Image

The key statistics are as follows:

  • ARPU (average revenue per user)
  • Avg. Cust. Lifetime, n (This is the inverse of the churn, n=1/[annual churn])
  • WACC (weighted average cost of capital)
  • Costs (annual costs to support the user in a given period)
  • SAC (subscriber acquisition costs, sometimes refereed to as CAC = customer acquisition costs)

The LTV formula, when used correctly, can be a good tactical tool for monitoring and comparing like-minded variable market programs, especially across channels. But like any model, its proper use is entirely dependent on the assumptions used in that model. Also, people who have a hidden agenda or who confuse a model with reality can misuse it. For many companies that subscribe to its wisdom, the formula slowly takes on more importance than it should. Seduced by the model, its practitioners often lose sight of the more important elements of corporate strategy, and become narrowly fixated on the dogmatic execution of the formula. In these cases, the formula can be confused, misused, and abused, much to the detriment of the business, and in many cases the customer as well.

Here are ten reasons to avoid worshiping at the LTV altar:

  1. It’s a Tool, Not a Strategy.  Heavy LTV companies forget that the LTV model does not create sustainable competitive advantage. You shouldn’t’ confuse output with input. The LTV formula is a measurement tool to be used by marketing to test the effectiveness of their marketing spend – nothing more and nothing less. If one asserts that buying customers below what they charge them is a corporate strategy, this is in essence an arbitrage game, and arbitrage games rarely last. Too many of the variables (specifically ARPU and SAC) are outside of your control, and nothing would prevent another player from executing the exact same strategy. It’s not rocket science; it’s a formula that any business school graduate can calculate. Do not fool yourself into believing it creates a proprietary advantage.
  2. The LTV Model Is Used To Rationalize Marketing Spending.  Marketing executives like big budgets, as big budgets make it easier to grow the top line.  The LTV formula “relaxes” the need for near term profitability and “justifies” the ability to play it forward – to spend today for benefits that are postponed into the future. It is no coincidence that companies that put a heavy emphasis on LTV are also the ones that have massive losses as they scale, frequently even through an IPO. Consider that most companies limit any “affiliate fee” they would be willing to spend to 5-10% of sales. Yet when they are marketing, they use different math. They use LTV math, and all the sudden it’s acceptable to spend 30-50% of revenue on customer acquisition. Find the most boisterous executive recommending excessive spending, and you will usually find a loyal servant of the LTV religion.
  3. The Model is Confused and Misused. Frequently the same group that is arguing for more spending is the same one that “owns” the LTV calculation. (This is a mistake – finance should monitor LTV).  As a result, it is not uncommon for one to see shortcuts taken that allow for greater freedom. As an example, marketers often divide spend by total customers to calculate SAC rather than just those customers that were “purchased.” If you have organic customers, they shouldn’t be included in the spend calculus. They would have arrived regardless of spend. Also, many people discount “revenues” rather than marginal cash contribution. It is critical to bundle all future variable costs of supporting the customer in order to fairly estimate the future contribution. As an example of the sloppiness that exists around the formula, consider this blog post (http://blog.kissmetrics.com/how-to-calculate-lifetime-value/) from KISS metrics, a company whose aim is to “help you make smarter business decisions.” Not only do they include a version of the model that specifically ignores future costs, but also they recommend taking an average of three different results, two of which are clearly flawed. This voodoo-math has no place as part of a multi-million dollar marketing exercise.
  4. Business Isn’t Physics – The Formula Is Not Absolute. LTV zealots often hold an overly confident view of the predictive nature of the formula. It’s not “hard science” like say predicting gravity. It’s at best a “good guess” about how the future will unfold. Businesses are complex adaptive systems that cannot be modeled with certainty. The future LTV results are simply predictions based on many assumptions that may or may not hold. Yet the LTV practitioner often moves forward with a brazen naiveté, evocative of the first time stock buyer who just found out about the price/earnings ratio, or the newcomer to Vegas who has just been taught the basics of twenty-one. LTV models win arguments because executives perceive them to be grounded in science. Just because its math, doesn’t mean its good math.
  5. The LTV Variables “Tug” at One Another. This may be the single most important issue and it lies at the heart of why the LTV model eventually breaks down and fails to scale ad infinitum. Tren Griffin, a close friend that has worked for both Craig McCaw and Bill Gates refers to the five variables of the LTV formula as the five horsemen. What he envisions is that a rope connects them all, and they are all facing different directions. When one horse pulls one way, it makes it more difficult for the other horse to go his direction. Tren’s view is that the variables of the LTV formula are interdependent not independent, and are an overly simplified abstraction of reality. If you try to raise ARPU (price) you will naturally increase churn. If you try to grow faster by spending more on marketing, your SAC will rise (assuming a finite amount of opportunities to buy customers, which is true). Churn may rise also, as a more aggressive program will likely capture customers of a lower quality. As another example, if you beef up customer service to improve churn, you directly impact future costs, and therefore deteriorate the potential cash flow contribution. Ironically, many company presentations show all metrics improving as you head into the future. This is unlikely to play out in reality.
  6. Growing Becomes a Grind. Let’s say you have a company that estimates it will do $100mm in revenue this year, $200mm the next, and $400mm the year after that. In order to accomplish those goals it is going to invest heavily in marketing – say 50% of revenues. So the budget for the next three years is $50mm, $100mm, and $200mm. How realistic is it to assume that your SAC will drop as you 4X your spend? Supply and demand analysis suggests the exact opposite outcome. As you try to buy more and more of a limited good, the price will inherently increase. The number one place on the planet for marketing spend is Google Adwords, and make no mistake about it, this is an increasingly finite resource. Click-outs are not growing at a meaningful pace, and key word purchases are highly contested. Assuming you will “get better” at buying while trying to buy more is a daunting assumption. The game will likely get tougher not easier.
  7. Purchased Customers Underperform Organic on Almost Every Metric. Organic users typically have a higher NPV, a higher conversion rate, a lower churn, and more satisfied than customers acquired through marketing spend. LTV heavy companies are in denial about this point. In fact, many of them will argue until they are blue in the face that the customer dynamics are the same while this is rarely the case. A customer that “chooses” your firm’s services will be much more staisfied than one that is persuaded to buy your product through spend. Find any high-marketing spend consumer subscription company, and I will show you a company with numerous complaints at the Better Business Bureau. These are companies that make it almost impossible to terminate your subscription. When you are scheming on how to trap the customer from finding the exit you are not building a long-term brand.
  8. The Money Could Go to the Customer.Think about this. If you are a company that spends millions and millions of dollars on marketing, wouldn’t you be better off handing that money to the customer versus handing it to a third-party who has nothing to do with the future life-time value of the customer? Providing a better value-proposition to the customer is much more likely to endure goodwill than spending on marketing. A heavy marketing spend necessitates a higher margin (to cover the spend), and therefore a higher end user price to the customer! So the customer is negatively impacted by the presence or “need” of the marketing program. Plus, a margin umbrella now exists for competition that chooses to undercut your margin model with a more efficient customer acquisition strategy (such as giving the customer the money).“More and more money will go into making a great customer experience, and less will go into shouting about the service. Word of mouth is becoming more powerful. If you offer a great service, people find out.” – Jeff Bezos
  9. LTV Obsession Creates Blinders. Many companies that obsess over LTV, become overwhelmed by LTV. In essence, the formula becomes a blinder that restricts creativity and open-mindedness. Some of the most efficient forms of marketing are viral, social, and effective PR (public relations). Most companies that obsess about LTV are less skilled at these more leveraged techniques. Ironically, it’s the scrappy and capital starved startup with absolutely no marketing budget that typically finds a clever way to scale growth organically.  I love this historic slide from Skype comparing their SAC with that of Vonage, an iconic disciple of LTV analysis.

Image

10. Tomorrow Never Arrives.  The Utopian destination imagined by the LTV formula is a mirage. It almost never works out as planned in the long run. Either growth begins to slow, or you run out of capital to continue to fund losses, or Wall Street cries uncle and asks to see profitability. When this happens the frailty of the model begins to appear. SAC is a little higher than expected. You met your growth target, but the projected loss was bigger than expected. Wall Street is hounding you for churn numbers, but you are reluctant to give them out. The lack of transparency then leads to cynicism, and everyone assumes the worse. It turns out that the excessive marketing spend was also propping up repeat purchase, and pulling back to achieve profitability is increasing churn. Moreover, a negative PR cycle has ensued as a result of your stock decline, and the press’ new doubts about your model. This also impacts results, and customer perception of your brand. The bottom line is that “one day we can stop spending and be remarkably profitable” rarely comes to fruition.

It is not impossible to create permanent equity value with the LTV approach, but it’s a dangerous game of timing – you don’t want to be the peak investor. Let’s say a new business starts with an early market capitalization of A (see graph below). Through aggressive marketing techniques, and aggressive fund raising, the company is able to achieve amazing revenue growth (and corresponding losses), but nonetheless creates a rather sizable organization. At this point, the company is value at point B. Eventually, however, gravity ensues and the constraints outlined herein raise their head, resulting in a collapse to point C.  For early founders and investors at point A, they may do OK (as long as C>A), but it will be accomplished on the backs of later stage investors that helped fund the unsustainable push to point B. This is the story of many a telecom and cable provider expansion history, as well as a few recent Internet companies.

Image

This should not be misconstrued as a eulogy for the LTV formula. It has a very important place in business as a way to contrast and compare alternative marketing programs and channels. It is a tactical marketing tool that requires candor and thoroughness in its implementation. The fundamental reason that it is so amazingly dangerous and seductive is its simplicity and certainty. Generic marketing is conceptual. LTV marketing is specific. Building a plan to grow to a million users organically is an order of magnitude more difficult than doing it with the aid of the LTV formula. There is comfort in its determinism, and it is simply easier to do.

Some people wield the LTV model as if they were Yoda with a light saber; “Look at this amazing weapon I know how to use!” Unfortunately, it is not that amazing, it’s not that unique to understand, and it is not a weapon, it’s a tool. Companies need a sustainable competitive advantage that is independent of their variable marketing campaigns. You can’t win a fight with a measuring tape.

Read Full Post | Make a Comment ( 8 so far )

How To Monetize a Social Network: MySpace and Facebook Should Follow TenCent

Posted on March 9, 2009. Filed under: advertising, casual games, Internet, social networking, Uncategorized, video, Virtual Goods | Tags: , , , , , , , , |

“Little Joe never once gave it away
Everybody had to pay and pay”

                  — Lou Reed, Walk on the Wild Side

The consensus seems to be that social networks have a monetization problem.  On this topic, both the leading technology industry blogs and the world’s top news organizations agree.  The problem is not that these sites have no revenue.   I “guesstimate” that MySpace and Facebook have annual revenue run-rates of approximately $650mm and $450mm respectively – highly reputable numbers.  The perceived problem relates directly to revenue per user or page view, as these are two of the most heavily trafficked sites on the Internet.  As a comparison, other companies with similar usage, like Yahoo, are doing $7.2B in annual revenues.  When reporting earnings from Q4 of 2007, Google also opined on the difficulty in monetizing social networking sites.  Sergey Brin noted, “I don’t think we have the killer best way to monetize social networks yet.”

There is ample historical data that proves web sites like these are inherently difficult to monetize.  Most other online communication products have had similar struggles.  Two great examples of this: the many leading players in the Instant Messaging (IM) space (AIM, ICQ, Yahoo Messenger) and the leading free email sites (Hotmail, Yahoo Mail).  These products/sites have always had some of the lowest eCPMs on the Internet.  Many speculate that this is because the user is so heavily engaged in using the product (i.e. communicating) that they are unlikely to be distracted by or engaged in an advertising message.  Another corollary to this point is that other Internet properties offer more direct purchasing intent based on the way they aggregate users.  Example here include TheKnot for brides, TripAdvisor for travelers, and even Google, where the search query highly delineates the direct intent of the user, allowing the advertiser to find users already in the purchasing funnel.  All of these properties have incredibly high eCPMs.

0700hk

Despite this conundrum, there is a solution.  Luckily for these U.S. based companies, a Chinese company named TenCent has already paved the way by identifying the optimal way to monetize this type of product.  For those that don’t know, TenCent is the owner of the leading IM franchise in China – a product known affectionately as “QQ”.  TenCent was founded in 1998, has 355 million users,  US$1.2B in annual revenues, and a US$11.2B market capitalization.  The stock chart for the past 5 years is included in the adjacent graphic.  The two primary drivers of revenue for TenCent are digital items and casual game packages and upgrades.  Advertising, which doesn’t work well on U.S. products like IM, doesn’t work well in China either.  Advertising revenues for TenCent represent only 12% of total revenues.  Recently, I asked a leading Internet analyst which company in China is best positioned above all others?  He quickly replied “TenCent”.

The spreadsheet below tries to highlight the monetization differences between TenCent, Facebook, and MySpace.  For each we have taken our best guess at monthly unique users, monthly page views, monthly revenues, and advertising as a percentage of revenue.  For TenCent, these numbers are published.  For MySpace and Facebook we used the best information we could find and/or infer.  We then calculated effective CPM (eCPM), revenue/user, and advertising revenue per user.  Lastly, we show these same numbers for TenCent with a cost of living adjustment.  In China the cell-phone ARPU (average revenue per user) is about 1/5th of that here in the U.S., so adjusting these numbers up by 5X gives you a much better number for comparing directly with the U.S. properties. 

social-spreadsheet2

The takeaways are quite straightforward.  The amount of advertising revenue on an adjusted basis at TenCent ($2.08) is quite similar to Facebook ($2.44) and MySpace ($5.85) (some may wonder why MySpace ad revenue per user is higher than Facebook – many believe they are more aggressive with ad placement and insertion).  The key difference in this comparison is obviously the revenue TenCent generates with business models that are largely absent on both Facebook and MySpace — digital items and casual game revenue.  For every $2 of adjusted advertising revenue TenCent has per user per year, they generate $17 in other revenue streams.  Benchmark Capital has invested in two private companies in the social/virtual world space – SecondLife and Gaia Online.  In both cases, the company revenues are significant, and in both cases advertising is not the leading business model.

More supportive data comes from the three leading social network players in Japan.  Believe or not, all three companies are already public and trade on different segments of the Tokyo Stock Exchange.  You will see in the same spreadsheet that Mixi, DENA (Mobage-town), and GREE have market capitalizations of US$511MM, US$1.5B, and US$1.1B respectively, and are all very profitable.   DENA and GREE, which interestingly are more popular on mobile than on the PC, have invested heavily in these two magic business models (casual games and digital items) and have revenues per user that dwarf that of Facebook or Myspace (DENA is 10X Facebook on this metric!).  As a result, these companies sport market capitalizations per user of over $100.  Here is the big punch line: Mixi is the actual leader in the market in terms of users, is the clear leader on the PC, is the company that most resembles U.S.-based social networks, and has remained focused on advertising as its core revenue steam.  Not surprisingly, their revenues per user are a fraction of DENA and GREE, as is their market capitalization.  

At a recent public investor conference in San Francisco, Alexander Tamas, an executive associated with the leading free email service in Russia (Mail.ru), noted that his company felt that the U.S. companies have little understanding on how to monetize a product like Mail.ru, and that they were taking their clues from TenCent in China.  Most of the public market investors in the audience, who have witnessed TenCent, DENA, and GREE’s remarkable success, nodded in agreement.  Gaia also presented at this conference and the crowd was standing-room only.  The questions from the audience made it even more apparent that the buy-side investors have a strong appreciation for the digital item business model. 

It is peculiar to have a situation where the NY-centric public market investors are more open minded to a new business model prior to the entrepreneurial executives on the west coast, but that is clearly the case here.  It is not hard to see why investors like this model.  When Pony Ma, the founder of TenCent, first described the digital item model to me five years ago I was blown away.  He was selling virtual clothes and accessories for digital avatars that represented his users online.  Think about it; this is a beautifully high gross margin business with very low marginal costs.  He even told me he thought digital shirts should deteriorate over time like real ones.  Pure genius. 

sunglassesIt is my perception that most U.S. executives have trouble conceiving and believing in the digital item model.  For starters, they simply think it’s strange.  “Why would someone buy clothes for their virtual avatar?  That’s weird.”  What they fail to realize is that U.S. consumers pay for “virtual” things all the time.  In the attached picture you see a pair of expensive Chanel sunglasses that retail for $329.  If you removed the Chanel logo from them, and offered them for $50 cheaper, you could not sell a pair.  Not one.  Why?  People are buying an image that they want to project about themselves.  Without the logo, they fail to make that statement.  The same is true for watches, clothes, cars, sodas, beers, cell phones, and many more items.  People care greatly about how they are perceived, and are willing to part with big bucks to achieve it.  Digital items are merely the same phenomenon online.

Another reason that digital items are a great monetization model for a social network is congruence of fit with the core activity of the site.  We already discussed how for TheKnot, the decision to come to the site is very consistent with identifying exactly “who” the advertiser is trying to reach and at “what time”.  For social networking sites, one of the key “experiences” of users is self-expression.  Think about it: is the Facebook news feed more about the reader or the poster?  Isn’t someone’s MySpace page all about self-expression?  If people are there to represent and express themselves, shouldn’t you build a business model that charges for the ability to better differentiate oneself?  Shouldn’t you also charge for ego-gratification on a sliding scale (the bigger the ego, the more the charge)?

These same executives like to believe that digital items are distinctly an Asian phenomenon – a convenient theory will prove to be a dangerous rationalization over time.  Here are some numbers from a U.S. corporation.  As I mentioned we are investors in LindenLab and their leading vitual world, SecondLife.  In SecondLife, the users are the ones that get to sell digital goods (rather than the company as in TenCent’s case).  Linden makes its money providing the platform services underneath this powerful economy.  At this moment in time, the economy inside of SecondLife – the amount of digital goods and services – sold each year between SecondLife users, is over a US$450mm annual run rate.  Of this, developers are realizing over $100MM in real profits extracted from Linden’s in-world to real-world currency exchange.  And keep in mind that SecondLife has much fewer users than either Facebook or MySpace.

Another interesting data point exists in the Facebook and MySpace application developer programs.  Best I can tell, the startups that are generating the most revenue on top of either platform are either selling digital items/avatars, or providing casual game packages — the exact two business models that are the drivers at TenCent, DENA, and GREE.  This is hardly a coincidence.

Despite these arguments and the fact that others have also been arguing this same point, it would be surprising if either MySpace or Facebook move in this direction.  First, they would need to have executive buy-in, which is not obvious at this point.  Second, they would need to hire people with experience in executing against this model.  Like any other endeavor in life, there are right ways and wrong ways to exploit these models, and there are already many experts in the field of digital items and casual games.  Lastly, they would need to prioritize this direction over other programs.  Currently, MySpace seems extremely focused on music, and Facebook on user-based communications.

The good news is that if they ever get around to deploying these models, they will not have trouble convincing Wall Street it’s a good idea – Wall Street is already there.

(follow me on twitter)

More Information:

Wikipedia on TenCent
TenCent IR About Page
TenCent IR Investor Intro
Stock Information and Company Financials on Mixi
Stock Information and Company Financials on DENA
Stock Information and Company Financials on GREE

In addition to these, most of the large US investment banks are covering TenCent, DENA, and GREE with English based research.  If you have a relationship with one of these banks, you can likely ask for their reports.

 

 

Read Full Post | Make a Comment ( 104 so far )

Perfect Online Video Advertising Model: Choose Your Advertiser

Posted on February 27, 2009. Filed under: advertising, Internet, online video, Uncategorized, Venture Capital, video, Web/Tech | Tags: , , , |

idea_bulb

Many companies and visionaries have pontificated about the future of video ads and different techniques for monetizing online videos.  A big part of this is driven by the fact that while YouTube is a huge user success, its a less proven monetization success.  On a recent trip to NYC, an idea came up which I can’t get out of my head, and the more I think about it, the more I believe that it is the Holy Grail for the future of online advertising.  But before I disclose the big “ah-ha,” a few caveats:

1) No one will ever monetize commodity content well.  If the same video is on YouTube, Veoh, and Metacafe, you won’t have the “right” to ask the consumer to wait for an advertisement.  Only by”controlling” unique/premium content can you ask the user to participate with advertisement.

2) Some networks, like ABC, are already doing a great job with online video advertisement.  I have heard numbers as high as 80% of the revenue per viewer hour compared with regular network television.  That said, I think the model below will dramatically enhance this number.

3) Having multiple sales forces selling the same premium video advertisement is counter productive. It drives down pricing.

4) “Distribution” is a confused word online.  Everything can be one-click away.  As such, there is no real reason to have someone “distribute” your video in the old classic sense – especially if it is important for you to control the advertising.

Enough with the caveats — here is the idea.  For online premium and unique VOD content, content owners should let the user pick one of a group (say 4-9) of sponsors for the show they are about to watch.  With online video this would be easy.  As you launch the show, it gives you an array of thumbnail choices for sponsorship.  The great thing here is that all parties win.  The consumer is happier with the advertising because its relevant to something on their mind, and they are more likely to pay attention when it “interrupts” the programming.  The advertisers gets a user that has qualified themselves as being interested in the product or category which is huge.  Known intent is massive.  And lastly, the content owner will likely end up with ad rates higher than they have ever seen previously.

In many ways, this is similar to how a user “self-declares” their interest when typing a Google search.  And this is way, way, way better than behavioral targeting.  When you think about it, behavioral targeting is a euphemism for “guessing”.   Just because I am a male between 18-24 and watching “Lost” doesn’t mean I want an XBOX.  You are more likely to guess that i might want it, but you would be 10X better off if I chose XBOX as my sponsor at the start of the show.  Then you would KNOW I have interest — no more guessing.

Making predictions is always a dangerous game, but I am fairly certain that this will be the video ad model of the future.  It makes way too much sense not to work.

3/4/09 Follow Up Note: 

I received some great direct feedback on this note from many leaders in the field including the CEOs of Hulu and Brightcove.  The main message was “we already thought of that and we are already doing it!”  I guess there are no new ideas only good ones thought of again.  But the good news is that they all agreed this is a powerful model.  On fair criticism is that it requires a great deal of liquidity in terms of numbers of advertisers to make this happen.  I suspect that is right, and as a result this may not play out for some time.  But when it does, it will be powerful.

Read Full Post | Make a Comment ( 28 so far )

    About

    …focusing on the evolution and economics of high technology business and strategy. By day, I am a venture capitalist at Benchmark Capital.

    RSS

    Subscribe Via RSS

    • Subscribe with Bloglines
    • Add your feed to Newsburst from CNET News.com
    • Subscribe in Google Reader
    • Add to My Yahoo!
    • Subscribe in NewsGator Online
    • The latest comments to all posts in RSS

    Meta

Liked it here?
Why not try sites on the blogroll...