Posts Tagged Valuation

Facebook Seasson 3 – Face/Off

1000 Days & a bit … As of January 2015 a Facebook share was around Double the price of their initial public offering (IPO) back in May 2012.

That increase in share price & value correspond to an investor belief (conscious or otherwise) that Facebook’s can grab around 40% of the online spend in the (near) Future versus a mediocre 20+% back in 2012.

If you had bought Facebook shares on September 4th 2012 and sold them again December 22nd 2014 (yes just before Christmas) you would have earned 5 times your original investment!

Back in 2012, some month after Facebook’s (FB) IPO, I wrote a Blog Facebook Values … Has the little boy spoken? on the FB value and how to get a feel whether the share price was Naked (“The Ugly”), Half Dressed (“The Bad”) or Nicely Dressed (“The Good”).

I asked whether it was a good time to have confidence and invest in Facebook … My answer (at the time) was that the share price of about 20 US$ (August 2012 timeframe) appeared too low compared with the potential for capturing Online Advertisement Spending Market Share, that furthermore was poised to increase substantially going forward as funds were being redirected from traditional advertisement spend to digital / online media.

The primary source of revenue for Facebook (then & now) is from Online Advertisement Spend. Thus, I looked at what long-term share (chosen arbitrarily to be 5 years) of the Online Ad Market should Facebook have in order to justify its value and share price. Very simply I ramped up the revenue share from its current value to a target share that would correspond to Facebook’s Market Capitalization or Share Price.

In the following I will ignore all the goodies that Facebook have launched or acquired over their lifetime, such as

  • Instagram (Apr 2012).
  • Whatsapp (Feb 2013).
  • Oculus (Mar 2013).
  • Atlas Ad Server Platform (acquisition Feb 2013 & re-vamped Sep 2014).
  • Autoplay Video Ads (Mar 2014).
  • Regular Facebook Software Updates.

(note list is not intended to be exhaustive).

All of the above (and much more) serves to make sure that “People use Facebook to stay connected with friends and family, to discover what’s going on in the world, and to share and express what matters to them”. (i.e., Facebook’s vision statement) … and (I assume) to make maximum profit out of the Facebook “addiction” by providing a very efficient advertisement platform enabled by the gazillion of personal data/information & impressions we all continuously volunteer by using Facebook.

However, while the technologies (e.g., algorithms & communications software utilities) behind are very exiting it all serves one purpose … deliver the most efficient ad to the user and make as much money out of that customer touch point. I believe that the potential and value of Whatsapp is huge and in message volume already exceeds the number daily SMS transactions globally. This still largely remains un-explored by Facebook. The question will be whether FB will primarily use Whatsapp as another Ad delivery vehicle or also as a mean to generate communications revenues in both the messaging and the voice consumer segments.

The conclusion back in August 2012, was that the share price of Facebook, based on its equivalent long-term share of the the Online Ad Market Spend, appeared low and one should expect the price (and value) to increase.

FB share price value & share of online spend

  • Above figure: Analysis presented in my 2012 “Facebook – time to invest?”. Note that the the share price dynamics are illustrated relative to the IPO price of 38 US$ or more accurately the stock price at closing on 18th of May 2012.

So what has happened in the almost 1000 days since the Facebook IPO?

Well after the share price dropped to 17.73 (at closing) on September 4th (2012), which roughly halved the Market Cap of Facebook, the FB Journey has been one of growth.

The cynic might of course point out that so has the rest of the market. However, while for example Nasdaq100 is ca. 68% higher (as of January 20th 2015) compared to the 18-May 2012, Facebook is almost double its IPO value (over the same period). If you take the lowest point (4 Sep 2012) and the highest point (22 Dec 2014) you have a 4.6 times ratio between high & low points.

So if you did the right thing and bought the lowest and sold at the highest … well I told you so (joking!) … Congratulation would be called for!

Here is the FB Journey ( = A Walk on the Wilder Side?) seen from an investors perspective

fb share dynamics

  • Above figure: illustrates Facebook stock price development since the IPO until end of January 2015 with commentary to the peaks and the dips.The Red Dot on September 4th 2012 represents the lowest historical share price (i.e., 17.71 @ closing) and the Green Dot the highest historical share price (i.e., 81.45 @ closing).

Taking the stock price dynamics as shown above, how would the previous analysis come out looking at what Online Ad Revenue Share could justify (approximately) the share price development over the period.

Well … it could look something like this based on Online Advertisement Market Share;

share price & online ad share

  • Above Figure: illustrates the share price dynamics relative to the IPO price, i.e., 100% level (at closing 18 May 2012). Further a relative simple valuation model based on Facebook’s long-term (i.e., 5+ years) online advertisement market share is used to derive the Online Ad Spend share that a given FB share price corresponds to. The methodology has been described in detail in “A walk on the Wild Side”.

Current stock price range is fairly consistent with a 40+% long-term (i.e., 5 year linear ramp up from 2013 share and then keeping the share at the level going forward) share of the Online Advertisement Market. I would expect Facebook to hit at least 10% for 2014 (based on eMarketer data of the total online advertisement market).

Will Facebook be able to grow to 40+% share of the Online Ad Spending?

In my opinion that does not sound completely un-reasonable …

Though it would imply that (a) other social media players also relying on the Online Ad Market are going to lose their livelihood, some (b) business plans might look somewhat more sombre for others and (c)  Facebook needs to take Google Head-On.

It will be very exiting to follow Facebook’s Q4 2014 and Full Year 2014 Earning Call on 28th January 2015 at 2 PM Pacific Time (i.e., 11 PM CET & 1 AM AST 29 January).

I expect to see (28 January 2015); Below find the comparison between my predictions and the real thing (i.e., actual data) out of the Q4 Earnings details as presented at the Q4 & 2014 Full Year Earning Call (28th of January 2015), I also recommend to read the transcript of the earning call on SeekingAlpha Blog. (Strike out text is my predictions prior to Earning Call)

  • Facebook beating Q4 earnings expectations and came out at 3.851B US$ coming out at 3.84B US$ (Low) to 4.16B (US$) (high).
  • MAU 1,393 1,386 Million and Mobile MAU of 1,189 1,182 Million (both lower limit estimates).
  • Share price at closing was slightly higher than previous trading day ending at 76.24 US$ per share (+0.6%) will drop during the day of the earning call to around 71 – 72 US$ per share.
  • Given the already high value of the stock, I do not expect much gain over the days after the earning call. Likely to recover to the level 5 – 10 days before the Q4/Full 2014 Earning Call.

… and I might be completely mistaken (and will be crying all the way to the bank)… but at least we will know within the next hours & days to come!

… So I ended up being fairly close to MAU (0.5% lower), Mobile MAU (0.6% lower) and earnings expectations (0.3% from my lower bound). However, predicting the stock movement … yeah … not so good. Still next couple of days will be interesting to follow. Dave Wehner, Facebook Chief Financial Officer, was where careful in managing expectations for Facebook Topline in 2015. Concerns about exchange rate effects on the Topline could result in a 5% lower revenue than it would have been with 2014 exchange rates. Basically the revenue growth in US$ would reduce somewhat due to exchange rate effects. This is likely to have some negative impact on Facebook profitability and their margin as their fundamental cost base is in US$.

Methodology

See also my Social Media Valuation Blog “A walk on the Wild Side”.

Following has been assumed in FB Valuation Assessment:

  1. WACC 9.11%
  2. 2013 FB capture 6% of total online ad spend.
  3. FB gains a sustainable share of online ad spend X%.
  4. 5 yr linear ramp-up from 2014 9.6% (assessed) to X%, and then maintained at that level.
  5. Other revenues 15% in 2014, linearly reduced to 10% after 5 yrs and then maintained.
  6. Assume FB can maintain a free cash flow yield of 25%.

It should be noted that the above analysis is in all likelihood oversimplifying. However it is not terrible difficult to add complexity. Though given the inherent uncertainties involved in predicting the future, the approach presented is good enough to get an idea about a given investments (or stock purchase) attractiveness.

For the financial history buffs, the Nasdaq100 is ca. 5% from the level of the dot.com foreshock (or pre-crash) of March-2000 and has surpassed the big crash of July-2000.

Acknowledgement

I greatly acknowledge my wife Eva Varadi for her support, patience and understanding during the creative process of creating this Blog.

Appendix

fb & nasdaq100

  • Figure above: shows the share price development from May 18 2012 (IPO date of FB) to 20 January 2015 of Facebook and Nasdaq100 composite. While Facebook largely under-performed in 2012 and well into 2013, its recovery from mid-2013 until January 2015 has been spectacular.

FB daily returns

 

FB vs Nasdaq100

  • Figure above: illustrates Laplace distribution representations of the daily returns of Facebook and Nasdaq100 over the period from 18 May 2012 to 20 January 2015. Note the slight right shift in centre point from the 0%. For a more detailed analysis of the Nasdaq100 and application of the Laplace distribution see the Business Forecasting Blog “The Nasdaq100 Daily Returns and Laplace Distributed Errors”.

online ad spending

  • Figure above: Online Advertisement Spending forecast from eMarketer (August 2013) representing the period 2013 to 2017. From 2018 and to 2022 Forecasts have been extrapolated based on 1st and 2nd derivative of the previous period growth. The resulting trend have been checked against other available projections.

facebook mau vs ad spend region etc

  • Figure above: illustrates for 2013 Facebook Monthly Active Users (MAU) in terms of share of population versus Region (source: Facebook Annual Report 2013), Region’s share of Ad Spend (source: eMarketer), Mobile Internet Penetration (i.e., CDMA2000, UMTS, HSPA, LTE, Mobile WiMax, source: Pyramid Research), and (fixed) internet penetration (i.e., the percentage of population having access to internet).

2017 facebook mau vs ad spend region etc

  • Figure above: illustrates for 2017 Facebook Monthly Active Users (MAU) in terms of share of population versus Region (Source: Authors Facebook Model), Region’s share of Ad Spend (Source: eMarketer), Mobile Internet Penetration (i.e., CDMA2000, UMTS, HSPA, LTE, Mobile WiMax, Source: Pyramid Research), and (fixed) internet penetration (i.e., the percentage of population having access to internet).

2013 mau vs reg lte wifi etc

  • Figure above: illustrates for 2013 Facebook Monthly Active Users (MAU) in terms of share of population versus Region (Source: Facebook Annual Report 2013), LTE penetration (Source: Pyramid Research), WiFi residential potential estimated from the broadband household penetration (Source: Pyramid Research), and  Mobile Internet Penetration (i.e., CDMA2000, UMTS, HSPA, LTE, Mobile WiMax, Source: Pyramid Research). For Facebook’s Autoplay Video feature it is important for the user to either have WiFi access or for a decent cellular performance LTE.

2017 mau vs reg lte wifi etc

  • Figure above: illustrates for 2017 Facebook Monthly Active Users (MAU) in terms of share of population versus Region (Source:Source: Authors Facebook Model), LTE penetration (Source: Pyramid Research), WiFi residential potential estimated from the broadband household penetration (Source: Pyramid Research), and  Mobile Internet Penetration (i.e., CDMA2000, UMTS, HSPA, LTE, Mobile WiMax, Source: Pyramid Research). For Facebook’s Autoplay Video feature it is important for the user to either have WiFi access or for a decent cellular performance LTE.

, , , ,

1 Comment

The Economics of the Thousand Times Challenge: Spectrum, Efficiency and Small Cells

By now the biggest challenge of the “1,000x challenge” is to read yet another story about the “1,000x challenge”.

This said, Qualcomm has made many beautiful presentations on The Challenge. It leaves the reader with an impression that it is much less of a real challenge, as there is a solution for everything and then some.

So bear with me while we take a look at the Economics and in particular the Economical Boundaries around the Thousand Times “Challenge” of providing (1) More spectrum, (2) Better efficiency and last but not least (3) Many more Small Cells.

THE MISSING LINK

While (almost) every technical challenge is solvable by clever engineering (i.e., something Qualcomm obviously have in abundance), it is not following naturally that such solutions are also feasible within the economical framework imposed by real world economics. At the very least, any technical solution should also be reasonable within the world of economics (and of course within a practical time-frame) or it becomes a clever solution but irrelevant to a real world business.

A  Business will (maybe should is more in line with reality) care about customer happiness. However a business needs to do that within healthy financial boundaries of margin, cash and shareholder value. Not only should the customer be happy, but the happiness should extend to investors and shareholders that have trusted the Business with their livelihood.

While technically, and almost mathematically, it follows that massive network densification would be required in the next 10 years IF WE KEEP FEEDING CUSTOMER DEMAND it might not be very economical to do so or at the very least such densification only make sense within a reasonable financial envelope.

Its obvious that massive network densification, by means of macro-cellular expansion, is unrealistic, impractically as well as uneconomically. Thus Small Cell concepts including WiFi has been brought to the Telecoms Scene as an alternative and credible solution. While Small Cells are much more practical, the question whether they addresses sufficiently the economical boundaries, the Telecommunications Industry is facing, remains pretty much unanswered.

PRE-AMP

The Thousand Times Challenge, as it has been PR’ed by Qualcomm, states that the cellular capacity required in 2020 will be at least 1,000 times that of “today”. Actually, the 1,000 times challenge is referenced to the cellular demand & supply in 2010, so doing the math

the 1,000x might “only” be a 100 times challenge between now and 2020 in the world of Qualcomm’s and alike. Not that it matters! … We still talk about the same demand, just referenced to a later (and maybe less “sexy” year).

In my previous Blogs, I have accounted for the dubious affair (and non-nonsensical discussion) of over-emphasizing cellular data growth rates (see “The Thousand Times Challenge: The answer to everything about mobile data”) as well as the much more intelligent discussion about how the Mobile Industry provides for more cellular data capacity starting with the existing mobile networks (see “The Thousand Time Challenge: How to provide cellular data capacity?”).

As it turns out  Cellular Network Capacity C can be described by 3 major components; (1) available bandwidth B, (2) (effective) spectral efficiency E and (3) number of cells deployed N.

The SUPPLIED NETWORK CAPACITY in Mbps (i.e., C) is equal to  the AMOUNT OF SPECTRUM, i.e., available bandwidth, in MHz (i..e, B) multiplied with the SPECTRAL EFFICIENCY PER CELL in Mbps/MHz (i.e., E) multiplied by the NUMBER OF CELLS (i.e., N). For more details on how and when to apply the Cellular Network Capacity Equation read my previous Blog on “How to provide Cellular Data Capacity?”).

SK Telekom (SK Telekom’s presentation at the 3GPP workshop on “Future Radio in 3GPP” is worth a careful study) , Mallinson (@WiseHarbor) and Qualcomm (@Qualcomm_tech, and many others as of late) have used the above capacity equation to impose a Target amount of cellular network capacity a mobile network should be able to supply by 2020: Realistic or Not, this target comes to a 1,000 times the supplied capacity level in 2010 (i.e., I assume that 2010 – 2020 sounds nicer than 2012 – 2022 … although the later would have been a lot more logical to aim for if one really would like to look at 10 years … of course that might not give 1,000 times which might ruin the marketing message?).

So we have the following 2020 Cellular Network Capacity Challenge:

Thus a cellular network in 2020 should have 3 times more spectral bandwidth B available (that’s fairly easy!), 6 times higher spectral efficiency E (so so … but not impossible, particular compared with 2010) and 56 times higher cell site density N (this one might  be a “real killer challenge” in more than one way), compared to 2010!.

Personally I would not get too hanged up about whether its 3 x 6 x 56 or 6 x 3 x 56 or some other “multiplicators” resulting in a 1,000 times gain (though some combinations might be a lot more feasible than others!)

Obviously we do NOT need a lot of insights to see that the 1,000x challenge is a

Rally call for Small & then Smaller Cell Deployment!

Also we do not need to be particular visionary (or have visited a Dutch Coffee Shop) to predict that by 2020 (aka The Future) compared to today (i.e., October 2012)?

Data demand from mobile devices will be a lot higher in 2020!

Cellular Networks have to (and will!) supply a lot more data capacity in 2020!

Footnote: the observant reader will have seen that I am not making the claim that there will be hugely more data traffic on the cellular network in comparison to today. The WiFi path might (and most likely will) take a lot of the traffic growth away from the cellular network.

BUT

how economical will this journey be for the Mobile Network Operator?

THE ECONOMICS OF THE THOUSAND TIMES CHALLENGE

Mobile Network Operators (MNOs) will not have the luxury of getting the Cellular Data Supply and Demand Equation Wrong.

The MNO will need to balance network investments with pricing strategies, churn & customer experience management as well as overall profitability and corporate financial well being:

Growth, if not manage, will lead to capacity & cash crunch and destruction of share holder value!

So for the Thousand Times Challenge, we need to look at the Total Cost of Ownership (TCO) or Total Investment required to get to a cellular network with 1,000 times more network capacity than today. We need to look at:

Investment I(B) in additional bandwidth B, which would include (a) the price of spectral re-farming (i.e., re-purposing legacy spectrum to a new and more efficient technology), (b) technology migration (e.g., moving customers off 2G and onto 3G or LTE or both) and (c) possible acquisition of new spectrum (i..e, via auction, beauty contests, or M&As).

Improving a cellular networks spectral efficiency I(E) is also likely to result in additional investments. In order to get an improved effective spectral efficiency, an operator would be required to (a) modernize its infrastructure, (b) invest into better antenna technologies, and (c) ensure that customer migration from older spectral in-efficient technologies into more spectral efficient technologies occurs at an appropriate pace.

Last but NOT Least the investment in cell density I(N):

Needing 56 times additional cell density is most likely NOT going to be FREE,

even with clever small cell deployment strategies.

Though I am pretty sure that some will make a very positive business case, out there in the Operator space, (note: the difference between Pest & Cholera might come out in favor of Cholera … though we would rather avoid both of them) comparing a macro-cellular expansion to Small Cell deployment, avoiding massive churn in case of outrageous cell congestion, rather than focusing on managing growth before such an event would occur.

The Real “1,000x” Challenge will be Economical in nature and will relate to the following considerations:

tco 2020

In other words:

Mobile Networks required to supply a 1,000 times present day cellular capacity are also required to provide that capacity gain at substantially less ABSOLUTE Total Cost of Ownership.

I emphasize the ABSOLUTE aspects of the Total Cost of Ownership (TCO), as I have too many times seen our Mobile Industry providing financial benefits in relative terms (i.e., relative to a given quality improvement) and then fail to mention that in absolute cost the industry will incur increased Opex (compared to pre-improvement situation). Thus a margin decline (i.e., unless proportional revenue is gained … and how likely is that?) as well as negative cash impact due to increased investments to gain the improvements (i.e., again assuming that proportional revenue gain remains wishful thinking).

Never Trust relative financial improvements! Absolutes don’t Lie!

THE ECONOMICS OF SPECTRUM.

Spectrum economics can be captured by three major themes: (A) ACQUISITION, (B) RETENTION and (C) PERFECTION. These 3 major themes should be well considered in any credible business plan: Short, Medium and Long-term.

It is fairly clear that there will not be a lot new lower frequency (defined here as <2.5GHz) spectrum available in the next 10+ years (unless we get a real breakthrough in white-space). The biggest relative increase in cellular bandwidth dedicated to mobile data services will come from re-purposing (i.e., perfecting) existing legacy spectrum (i.e., by re-farming). Acquisition of some new bandwidth in the low frequency range (<800MHz), which per definition will not be a lot of bandwidth and will take time to become available. There are opportunities in the very high frequency range (>3GHz) which contains a lot of bandwidth. However this is only interesting for Small Cell and Femto Cell like deployments (feeding frenzy for small cells!).

As many European Countries re-auction existing legacy spectrum after the set expiration period (typical 10 -15 years), it is paramount for a mobile operator to retain as much as possible of its existing legacy spectrum. Not only is current traffic tied up in the legacy bands, but future growth of mobile data will critical depend on its availability. Retention of existing spectrum position should be a very important element of an Operators  business plan and strategy.

Most real-world mobile network operators that I have looked at can expect by acquisition & perfection to gain between 3 to 8 times spectral bandwidth for cellular data compared to today’s situation.

For example, a typical Western European MNO have

  1. Max. 2x10MHz @ 900MHz primarily used for GSM. Though some operators are having UMTS 900 in operation or plans to re-farm to UMTS pending regulatory approval.
  2. 2×20 MHz @ 1800MHz, though here the variation tend to be fairly large in the MNO spectrum landscape, i.e., between 2x30MHz down-to 2x5MHz. Today this is exclusively in use for GSM. This is going to be a key LTE band in Europe and already supported in iPhone 5 for LTE.
  3. 2×10 – 15 MHz @ 2100MHz is the main 3G-band (UMTS/HSPA+) in Europe and is expected to remain so for at least the next 10 years.
  4. 2×10 @ 800 MHz per operator and typically distributed across 3 operator and dedicated to LTE. In countries with more than 3 operators typically some MNOs will have no position in this band.
  5. 40 MHz @ 2.6 GHz per operator and dedicated to LTE (FDD and/or TDD). From a coverage perspective this spectrum would in general be earmarked for capacity enhancements rather than coverage.

Note that most European mobile operators did not have 800MHz and/or 2.6GHz in their spectrum portfolios prior to 2011. The above list has been visualized in the Figure below (though only for FDD and showing the single side of the frequency duplex).

spectrum_details

The 700MHz will eventually become available in Europe (already in use for LTE in USA via AT&T and VRZ) for LTE advanced. Though the time frame for 700MHz cellular deployment in Europe is still expected take maybe up to 8 years (or more) to get it fully cleared and perfected.

Today (as of 2012) a typical European MNO would have approximately (a) 60 MHz (i.e., DL+UL) for GSM, (b) 20 – 30 MHz for UMTS and (c) between 40MHz – 60MHz for LTE (note that in 2010 this would have been 0MHz for most operators!). By 2020 it would be fair to assume that same MNO could have (d) 40 – 50 MHz for UMTS/HSPA+ and (e) 80MHz – 100MHz for LTE. Of course it is likely that mobile operators still would have a thin GSM layer to support roaming traffic and extreme laggards (this is however likely to be a shared resource among several operators). If by 2020 10MHz to 20MHz would be required to support voice capacity, then the MNO would have at least 100MHz and up-to 130MHz for data.

Note if we Fast-Backward to 2010, assume that no 2.6GHz or 800MHz auction had happened and that only 2×10 – 15 MHz @ 2.1GHz provided for cellular data capacity, then we easily get a factor 3 to 5 boost in spectral capacity for data over the period. This just to illustrate the meaningless of relativizing the challenge of providing network capacity.

So what’s the economical aspects of spectrum? Well show me the money!

Spectrum:

  1. needs to be Acquired (including re-acquired = Retention) via (a) Auction, (b) Beauty contest or (c) Private transaction if allowed by the regulatory authorities (i.e., spectrum trading); Usually spectrum (in Europe at least) will be time-limited right-to-use! (e.g., 10 – 15 years) => Capital investments to (re)purchase spectrum.
  2. might need to be Perfected & Re-farmed to another more spectral efficient technology => new infrastructure investments & customer migration cost (incl. acquisition, retention & churn).
  3. new deployment with coverage & service obligations => new capital investments and associated operational cost.
  4. demand could result in joint ventures or mergers to acquire sufficient spectrum for growth.
  5. often has a re-occurring usage fee associate with its deployment => Operational expense burden.

First 3 bullet points can be attributed mainly to Capital expenditures and point 5. would typically be an Operational expense. As we have seen in US with the failed AT&T – T-Mobile US merger, bullet point 4. can result in very high cost of spectrum acquisition. Though usually a merger brings with it many beneficial synergies, other than spectrum, that justifies such a merger.

spectrum_cost

Above Figure provides a historical view on spectrum pricing in US$ per MHz-pop. As we can see, not all spectrum have been borne equal and depending on timing of acquisition, premium might have been paid for some spectrum (e.g., Western European UMTS hyper pricing of 2000 – 2001).

Some general spectrum acquisition heuristics can be derived by above historical overview (see my presentation “Techno-Economical Aspects of Mobile Broadband from 800MHz to 2.6GHz” on @slideshare for more in depth analysis).

spectrum_heuristics

Most of the operator cost associated with Spectrum Acquisition, Spectrum Retention and Spectrum Perfection should be more or less included in a Mobile Network Operators Business Plans. Though the demand for more spectrum can be accelerated (1) in highly competitive markets, (2) spectrum starved operations, and/or (3) if customer demand is being poorly managed within the spectral resources available to the MNO.

WiFi, or in general any open radio-access technology operating in ISM bands (i.e., freely available frequency bands such as 2.4GHz, 5.8GHz), can be a source of mitigating costly controlled-spectrum resources by stimulating higher usage of such open-technologies and open-bands.

The cash prevention or cash optimization from open-access technologies and frequency bands should not be under-estimated or forgotten. Even if such open-access deployment models does not make standalone economical sense, is likely to make good sense to use as an integral part for the Next Generation Mobile Data Network perfecting & optimizing open- & controlled radio-access technologies.

The Economics of Spectrum Acquisition, Spectrum Retention & Spectrum Perfection is of such tremendous benefits that it should be on any Operators business plans: short, medium and long-term.

THE ECONOMICS OF SPECTRAL EFFICIENCY

The relative gain in spectral efficiency (as well as other radio performance metrics) with new 3GPP releases has been amazing between R99 and recent HSDPA releases. Lots of progress have been booked on the account of increased receiver and antenna sophistication.

spectral_efficiency_gain_per_technology

If we compare HSDPA 3.6Mbps (see above Figure) with the first Release of LTE, the spectral efficiency has been improved with a factor 4. Combined with more available bandwidth for LTE, provides an even larger relative boost of supplied bandwidth for increased capacity and customer quality. Do note above relative representation of spectral efficiency gain largely takes away the usual (almost religious) discussions of what is the right spectral efficiency and at what load. The effective (what that may be in your network) spectral efficiency gain moving from one radio-access release or generation to the next would be represented by the above Figure.

Theoretically this is all great! However,

Having the radio-access infrastructure supporting the most spectral efficient technology is the easy part (i.e., thousands of radio nodes), getting your customer base migrated to the most spectral efficient technology is where the challenge starts (i.e., millions of devices).

In other words, to get maximum benefits of a given 3GPP Release gains, an operator needs to migrate his customer-base terminal equipment to that more Efficient Release. This will take time and might be costly, particular if accelerated. Irrespective, migrating a customer base from radio-access A (e.g., GSM) to radio-access B (e.g., LTE), will take time and adhere to normal market dynamics of churn, retention, replacement factors, and gross-adds. The migration to a better radio-access technology can be stimulated by above-market-average acquisition & retention investments and higher-than-market-average terminal equipment subsidies. In the end competitors market reactions to your market actions, will influence the migration time scale very substantially (this is typically under-estimate as competitive driving forces are ignored in most analysis of this problem).

The typical radio-access network modernization cycle has so-far been around 5 years. Modernization is mainly driven by hardware obsolescence and need for more capacity per unit area than older (first & second) generation equipment could provide. The most recent and ongoing modernization cycle combines the need for LTE introduction with 2G and possibly 3G modernization. In some instances retiring relative modern 3G equipment on the expense of getting the latest multi-mode, so-called Single-RAN equipment, deployed, has been assessed to be worth the financial cost of write-off.  This new cycle of infrastructure improvements will in relative terms far exceed past upgrades. Software Definable Radios (SDR) with multi-mode (i.e., 2G, 3G, LTE) capabilities are being deployed in one integrated hardware platform, instead of the older generations that were separated with the associated floor space penalty and operational complexity. In theory only Software Maintenance & simple HW upgrades (i.e., CPU, memory, etc..) would be required to migrate from one radio-access technology to another. Have we seen the last HW modernization cycle? … I doubt it very much! (i.e., we still have Cloud and Virtualization concepts going out to the radio node blurring out the need for own core network).

Multi-mode SDRs should in principle provide a more graceful software-dominated radio-evolution to increasingly more efficient radio access; as cellular networks and customers migrate from HSPA to HSPA+ to LTE and to LTE-advanced. However, in order to enable those spectral-efficient superior radio-access technologies, a Mobile Network Operator will have to follow through with high investments (or incur high incremental operational cost) into vastly improved backhaul-solutions and new antenna capabilities than the past access technologies required.

Whilst the radio access network infrastructure has gotten a lot more efficient from a cash perspective, the peripheral supporting parts (i.e., antenna, backhaul, etc..) has gotten a lot more costly in absolute terms (irrespective of relative cost per Byte might be perfectly OKAY).

Thus most of the economics of spectral efficiency can and will be captured within the modernization cycles and new software releases without much ado. However, backhaul and antenna technology investments and increased operational cost is likely to burden cash in the peak of new equipment (including modernization) deployment. Margin pressure is therefor likely if the Opex of supporting the increased performance is not well managed.

To recapture the most important issues of Spectrum Efficiency Economics:

  • network infrastructure upgrades, from a hardware as well as software perspective, are required => capital investments, though typically result in better Operational cost.
  • optimal customer migration to better and more efficient radio-access technologies => market invest and terminal subsidies.

Boosting spectrum much beyond 6 times today’s mobile data dedicated spectrum position is unlikely to happen within a foreseeable time frame. It is also unlikely to happen in bands that would be very interesting for both providing both excellent depth of coverage and at the same time depth of capacity (i.e., lower frequency bands with lots of bandwidth available). Spectral efficiency will improve with both next generation HSPA+ as well as with LTE and its evolutionary path. However, depending on how we count the relative improvement, it is not going to be sufficient to substantially boost capacity and performance to the level a “1,000 times challenge” would require.

This brings us to the topic of vastly increased cell site density and of course Small Cell Economics.

THE ECONOMICS OF INCREASED CELL SITE DENSITY

It is fairly clear that there will not be a lot new spectrum available in the next 10+ years. The relative increase in cellular bandwidth will come from re-purposing & perfecting existing legacy spectrum (i.e., by re-farming) and acquiring some new bandwidth in the low frequency range (<800MHz) which per definition is not going to provide a lot of bandwidth.  The very high-frequency range (>3GHz) will contain a lot of bandwidth, but is only interesting for Small Cell and Femto-cell like deployments (feeding frenzy for Small Cells).

Financially Mobile Operators in mature markets, such as Western Europe, will be lucky to keep their earning and margins stable over the next 8 – 10 years. Mobile revenues are likely to stagnate and possible even decline. Opex pressure will continue to increase (e.g., just simply from inflationary pressures alone). MNOs are unlikely to increase cell site density, if it leads to incremental cost & cash pressure that cannot be recovered by proportional Topline increases. Therefor it should be clear that adding many more cell sites (being it Macro, Pico, Nano or Femto) to meet increasing (often un-managed & unprofitable) cellular demand is economically unwise and unlikely to happen unless followed by Topline benefits.

Increasing cell density dramatically (i.e., 56 times is dramatic!) to meet cellular data demand will only happen if it can be done with little incremental cost & cash pressure.

I have no doubt that distributing mobile data traffic over more and smaller nodes (i.e., decrease traffic per node) and utilize open-access technologies to manage data traffic loads are likely to mitigate some of the cash and margin pressure from supporting the higher performance radio-access technologies.

So let me emphasize that there will always be situations and geographical localized areas where cell site density will be increased disregarding the economics, in order to increase urgent capacity needs or to provide specialized-coverage needs. If an operator has substantially less spectral overhead (e.g., AT&T) than a competitor (e.g., T-Mobile US), the spectrum-starved operator might decide to densify with Small Cells and/or Distributed Antenna Systems (DAS) to be able to continue providing a competitive level of service (e.g., AT&T’s situation in many of its top markets). Such a spectrum starved operator might even have to rely on massive WiFi deployments to continue to provide a decent level of customer service in extreme hot traffic zones (e.g., Times Square in NYC) and remain competitive as well as having a credible future growth story to tell shareholders.

Spectrum-starved mobile operators will move faster and more aggressively to Small Cell Network solutions including advanced (and not-so-advanced) WiFi solutions. This fast learning-curve might in the longer term make up for a poorer spectrum position.

In the following I will consider Small Cells in the widest sense, including solutions based both on controlled frequency spectrum (e.g., HSPA+, LTE bands) as well in the ISM frequency bands (i.e., 2.4GHz and 5.8GHz). The differences between the various Small Cell options will in general translate into more or less cells due to radio-access link-budget differences.

As I have been involved in many projects over the last couple of years looking at WiFi & Small Cell substitution for macro-cellular coverage, I would like to make clear that in my opinion:

A Small Cells Network is not a good technical (or economical viable) solution for substituting macro-cellular coverage for a mobile network operator.

However, Small Cells however are Great for

  • Specialized coverage solutions difficult to reach & capture with standard macro-cellular means.
  • Localized capacity addition in hot traffic zones.
  • Coverage & capacity underlay when macro-cellular cell split options have been exhausted.

The last point in particular becomes important when mobile traffic exceeds the means for macro-cellular expansion possibilities, i.e., typically urban & dense-urban macro-cellular ranges below 200 meters and in some instances maybe below 500 meters pending on the radio-access choice of the Small Cell solution.

Interference concerns will limit the transmit power and coverage range. However our focus are small localized and tailor-made coverage-capacity solutions, not a substituting macro-cellular coverage, range limitation is of lesser concern.

For great accounts of Small Cell network designs please check out Iris Barcia (@IBTwi) & Simon Chapman (@simonchapman) both from Keima Wireless. I recommend the very insightful presentation from Iris “Radio Challenges and Opportunities for Large Scale Small Cell Deployments” which you can find at “3G & 4G Wireless Blog” by Zahid Ghadialy (@zahidtg, a solid telecom knowledge source for our Industry).

When considering small cell deployment it makes good sense to understand the traffic behavior of your customer base. The Figure below illustrates a typical daily data and voice traffic profile across a (mature) cellular network:

a_typical_traffic_day_in_europe

  • up-to 80% of cellular data traffic happens either at home or at work.+

Currently there is an important trend, indicating that the evening cellular-data peak is disappearing coinciding with the WiFi-peak usage taking over the previous cellular peak hour.

A great source of WiFi behavioral data, as it relates to Smartphone usage, you will find in Thomas Wehmeier’s (Principal Analyst, Informa: @Twehmeier) two pivotal white papers on  “Understanding Today’s Smatphone User” Part I and Part II.

The above daily cellular-traffic profile combined with the below Figure on cellular-data usage per customer distributed across network cells

traffic_over_network_distribution

shows us something important when it comes to small cells:

  • Most cellular data traffic (per user) is limited to very few cells.
  • 80% (50%) of the cellular data traffic (per user) is limited to 3 (1) main cells.
  • The higher the cellular data usage (per user) the fewer cells are being used.

It is not only important to understand how data traffic (on a per user) behaves across the cellular network. It is likewise very important to understand how the cellular-data traffic multiplex or aggregate across the cells in the mobile network.

We find in most Western European Mature 3G networks the following trend:

traffic_over_cell_distribution

  • 20% of the 3G Cells carries 60+% of the 3G data traffic.
  • 50% of the 3G Cells carriers 95% or more of the 3G data traffic.

Thus relative few cells carries the bulk of the cellular data traffic. Not surprising really as this trend was even more skewed for GSM voice.

The above trends are all good news for Small Cell deployment. It provides confidence that small cells can be effective means to taking traffic away from macro-cellular areas, where there is no longer an option for conventional capacity expansions (i.e., sectorization, additional carrier or conventional cell splits).

For the Mobile Network Operator, Small Cell Economics is a Total Cost of Ownership exercise comparing Small Cell Network Deployment  to other means of adding capacity to the existing mobile network.

The Small Cell Network needs (at least) to be compared to the following alternatives;

  1. Greenfield Macro-cellular solutions (assuming this is feasible).
  2. Overlay (co-locate) on existing network grid.
  3. Sectorization of an existing site solution (i.e., moving from 3 sectors to 3 + n on same site).

Obviously, in the “extreme” cellular-demand limit where non of the above conventional means of providing additional cellular capacity are feasible, Small Cell deployment is the only alternative (besides doing nothing and letting the customer suffer). Irrespective we still need to understand how the economics will work out, as there might be instances where the most reasonable strategy is to let your customer “suffer” best-effort services. This would in particular be the case if there is no real competitive and incremental Topline incentive by adding more capacity.

However,

Competitive circumstances could force some spectrum-starved operators to deploy small cells irrespective of it being financially unfavorable to do so.

Lets begin with the cost structure of a macro-cellular 3G Greenfield Rooftop Site Solution. We take the relevant cost structure of a configuration that we would be most likely to encounter in a Hot Traffic Zone / Metropolitan high-population density area which also is likely to be a candidate area for Small Cell deployment. The Figure below shows the Total Cost of Ownership, broken down in Annualized Capex and Annual Opex, for a Metropolitan 3G macro-cellular rooftop solution:

tco_greenfield_rooftop_site

Note 1: The annualized Capex has been estimated assuming 5 years for RAN Infra, Backaul & Core, and 10 years for Build. It is further assumed that the site is supported by leased-fiber backhaul. Opex is the annual operational expense for maintaining the site solution.

Note 2: Operations Opex category covers Maintenance, Field-Services, Staff cost for Ops, Planning & optimization. The RAN infra Capex category covers: electronics, aggregation, antenna, cabling, installation & commissioning, etc..

Note 3: The above illustrated cost structure reflects what one should expect from a typical European operation. North American or APAC operators will have different cost distributions. Though it is not expected to change conclusions substantially (just redo the math).

When we discuss Small Cell deployment, particular as it relates to WiFi-based small cell deployment, with Infrastructure Suppliers as well as Chip Manufacturers you will get the impression that Small Cell deployment is Almost Free of Capex and Opex; i.e., hardly any build cost, free backhaul and extremely cheap infrastructure supported by no site rental, little maintenance and ultra-low energy consumption.

Obviously if Small Cells cost almost nothing, increasing cell site density with 56 times or more becomes very interesting economics … Unfortunately such ideas are wishful thinking.

For Small Cells not to substantially pressure margins and cash, Small Cell Cost Scaling needs to be very aggressive. If we talk about a 56x increase in cell site density the incremental total cost of ownership should at least be 56 times better than to deploy a macro-cellular expansion. Though let’s not fool ourselves!

No mobile operator would densify their macro cellular network 56 times if absolute cost would proportionally increase!

No Mobile operator would upsize their cellular network in any way unless it is at least margin, cost & cash neutral.

(I have no doubt that out there some are making relative business cases for small cells comparing an equivalent macro-cellular expansion versus deploying Small Cells and coming up with great cases … This would be silly of course, not that this have ever prevented such cases to be made and presented to Boards and CxOs).

The most problematic cost areas from a scaling perspective (relative to a macro-cellular Greenfield Site) are (a) Site Rental (lamp posts, shopping malls,), (b) Backhaul Cost (if relying on Cable, xDSL or Fiber connectivity), (c) Operational Cost (complexity in numbers, safety & security) and (d) Site Build Cost (legal requirements, safety & security,..).

In most realistic cases (I have seen) we will find a 1:12 to 1:20 Total Cost of Ownership difference between a Small Cell unit cost and that of a Macro-Cellular Rooftop’s unit cost. While unit Capex can be reduced very substantially, the Operational Expense scaling is a lot harder to get down to the level required for very extensive Small Cell deployments.

EXAMPLE:

For a typical metropolitan rooftop (in Western Europe) we have the annualized capital expense (Capex) of ca. 15,000 Euro and operational expenses (Opex) in the order of 30,000 Euro per annum. The site-related Opex distribution would look something like this;

  • Macro-cellular Rooftop 3G Site Unit Annual Opex:
  • Site lease would be ca. 10,500EUR.
  • Backhaul would be ca. 9,000EUR.
  • Energy would be ca. 3,000EUR.
  • Operations would be ca. 7,500EUR.
  • i.e., total unit Opex of 30,000EUR (for average major metropolitan area)

Assuming that all cost categories could be scaled back with a factor 56 (note: very big assumption that all cost elements can be scaled back with same factor!)

  • Target Unit Annual Opex cost for a Small Cell:
  • Site lease should be less than 200EUR (lamp post leases substantially higher)
  • Backhaul should be  less than 150EUR (doable though not for carrier grade QoS).
  • Energy should be less than 50EUR (very challenging for todays electronics)
  • Operations should be less than 150EUR (ca. 1 hour FTE per year … challenging).
  • Annual unit Opex should be less than 550EUR (not very likely to be realizable).

Similar for the Small Cell unit Capital expense (Capex) would need to be done for less than 270EUR to be fully scalable with a macro-cellular rooftop (i.e., based on 56 times scaling).

  • Target Unit Annualized Capex cost for a Small Cell:
  • RAN Infra should be less than 100EUR (Simple WiFi maybe doable, Cellular challenging)
  • Backhaul would be less than 50EUR (simple router/switch/microwave maybe doable).
  • Build would be less than 100EUR (very challenging even to cover labor).
  • Core would be less than 20EUR (doable at scale).
  • Annualized Capex should be less than 270EUR (very challenging to meet this target)
  • Note: annualization factor: 5 years for all including Build.

So we have a Total Cost of Ownership TARGET for a Small Cell of ca. 800EUR

Inspecting the various capital as well as operational expense categories illustrates the huge challenge to be TCO comparable to a macro-cellular urban/dense-urban 3G-site configuration.

Massive Small Cell Deployment needs to be almost without incremental cost to the Mobile Network Operator to be a reasonable scenario for the 1,000 times challenge.

Most the analysis I have seen, as well as carried out myself, on real cost structure and aggressive pricing & solution designs shows that the if the Small Cell Network can be kept between 12 to 20 Cells (or Nodes) the TCO compares favorably to (i.e., beating) an equivalent macro-cellular solution. If the Mobile Operator is also a Fixed Broadband Operator (or have favorable partnership with one) there are in general better cost scaling possible than above would assume (e.g., another AT&T advantage in their DAS / Small Cell strategy).

In realistic costing scenarios so far, Small Cell economical boundaries are given by the Figure below:

Let me emphasize that above obviously assumes that an operator have a choice between deploying a Small Cell Network and conventional Cell Split, Nodal Overlay (or co-location on existing cellular site) or Sectorization (if spectral capacity allows). In the Future and in Hot Traffic Zones this might not be the case. Leaving Small Cell Network deployment or letting the customers “suffer” poorer QoS be the only options left to the mobile network operator.

So how can we (i.e., the Mobile Operator) improve the Economics of Small Cell deployment?

Having access fixed broadband such as fiber or high-quality cable infrastructure would make the backhaul scaling a lot better. Being a mobile and fixed broadband provider does become very advantageous for Small Cell Network Economics. However, the site lease (and maintenance) scaling remains a problem as lampposts or other interesting Small Cell locations might not scale very aggressively (e.g., there are examples of lamppost leases being as expensive as regular rooftop locations). From a capital investment point of view, I have my doubts whether the price will scale downwards as favorable as they would need to be. Much of the capacity gain comes from very sophisticated antenna configurations that is difficult to see being extremely cheap:

Small Cell Equipment Suppliers would need to provide a Carrier-grade solution priced at  maximum 1,000EUR all included! to have a fighting chance of making massive small cell network deployment really economical.

We could assume that most of the “Small Cells” are in fact customers existing private access points (or our customers employers access points) and simply push (almost) all cellular data traffic onto those whenever a customer is in vicinity of such. All those existing and future private access points are (at least in Western Europe) connected to at least fairly good quality fixed backhaul in the form of VDSL, Cable (DOCSIS3), and eventually Fiber. This would obviously improve the TCO of “Small Cells” tremendously … Right?

Well it would reduce the MNOs TCO (as it shift the cost burden to the operator’s customer or employers of those customers) …Well … This picture also would  not really be Small Cells in the sense of proper designed and integrated cells in the Cellular sense of the word, providing the operator end-2-end control of his customers service experience. In fact taking the above scenario to the extreme we might not need Small Cells at all, in the Cellular sense, or at least dramatically less than using the standard cellular capacity formula above.

In Qualcomm (as well as many infrastructure suppliers) ultimate vision the 1,000x challenge is solved by moving towards a super-heterogeneous network that consist of everything from Cellular Small Cells, Public & Private WiFi access points as well as Femto cells thrown into the equation as well.

Such an ultimate picture might indeed make the Small Cell challenge economically feasible. However, it does very fundamentally change the current operational MNO business model and it is not clear that transition comes without cost and only benefits.

Last but not least it is pretty clear than instead of 3 – 5 MNOs all going out plastering walls and lampposts with Small Cell Nodes & Antennas sharing might be an incredible clever idea. In fact I would not be altogether surprised if we will see new independent business models providing Shared Small Cell solutions for incumbent Mobile Network Operators.

Before closing the Blog, I do find it instructive to pause and reflect on lessons from Japan’s massive WiFi deployment. It might serves as a lesson to massive Small Cell Network deployment as well and an indication that collaboration might be a lot smarter than competition when it comes to such deployment:

softband_wifi_deployment

, , , , , , , , , , , , , , , , , ,

1 Comment

Facebook Values … Has the little boy spoken?

Facebook has lost ca. 450+ Million US$ per day since its IPO … or about 40 Billion US$ … in a little under 90 days (i.e., reference date 17-08-2012).

This is like loosing an Economy such as the Seychelles every second day. Or a Bulgaria  in less than 90 days. (Note: this is not to say that you could buy Bulgaria for $40B … well who knows? 😉 … the comparison just serves at making the loss of Facebook value more tangible. Further one should not take the suggestion of a relationship between market value of a corporation such as Facebook with GDP of country too serious as also pointed out by Dean Bubley @disruptivedean).

That’s a lot of value lost in a very short time. I am sure Bulgarians,”Seychellians” and FB investors can agree to that.

40 Billion US Dollar?  … Its a little less than 20 Mars Missions … or

40 Billion US Dollar could keep 35 thousand Americans in work for 50 years each!

So has the little boy spoken? Is the Emperor of Social Media Naked?

Illustration: THORARINN LEIFSSON http://www.totil.com

Let’s have a more detailed look at Facebook’s share price development since May 18th 2012.

The Chart below shows the Facebook’s share price journey, the associated book value, the corresponding sustainable share of Online Ad Spend (with an assumed 5yr linear ramp-up from today’s share) and the projected share of Online Ad Spend in 2012.

In the wisdom of looking backwards …  is Facebook, the Super-Mario of Social Media, really such a bad investment? or is this just a bump in a long an prosperous road ahead?

I guess it all rise and fall with what ever belief an investor have of Facebook’s ability to capture sufficient Online Advertisement Spend. Online Ad spend obviously includes the Holy Grail of Mobile Ad Revenues as well.

FB’s revenue share of Online Ad Spend has raised steady from 1.3% in 2009 to ca. 5% in2011 and projected to be at least 6% in 2012.

Take a look at FB’s valuation (or book value) which at the time of the IPO (i.e., May 18th 2012) was ca. 80+ Billion US Dollars. Equivalent to a share price of $38.32 per share (at closing).

In terms of sustainable business such a valuation could be justifiable if FB could capture and sustain at least 23% of the Online Ad Spend in the longer run. Compare this with ca. 5% in 2011. Compare this with Googles 40+% om 2011. AOL, which is Top 5 of best companies at conquering Online Advertisement Spend, share of Online Ad Spend was a factor 15 less than Google. Furthermore, Top-5 accounts for more than 70% of the Online Ad Spend in 2011. The remaining 30% of Online Ad Spend arises mainly from Asia Pacific logo-graphic, politically complicated, and Cyrillic dominated countries of which Latin-based Social Media & Search in general perform poorly in (i.e., when it comes to capturing Online Ad Spend).

Don’t worry! Facebook is in the Top 5 list of companies getting a piece of the Online Advertisement pie.

It would appear likely that Facebook should be able to continue to increase its share of Online Ad Spend from today’s fairly low level. The above chart shows FB’s current share price level (closing 17-August-2012) corresponds to a book value of ca. $40 Billion and a sustainable share of the Online Ad Spend of a bit more than 10+%.

It would be sad if Facebook should not be able to ever get more than 10% of the Online Ad Spend.

From this perspective:

A Facebook share price below $20 does seem awfully cheap!

Is it time to invest in Facebook? … at the moment it looks like The New Black is bashing Social Media!

So the share price of Facebook might drop further … as current investors try too off-load their shares (at least the ones that did not buy at and immediately after the IPO).

Facebook has 900+ Million (and approaching a Billion) users. More than 500+ Million of those 900+ Million Facebook users are active daily and massively using their Smartphones to keep updated with Friends and Fiends. In 2011 there where more than 215 Billion FB events.

Facebook should be a power house for Earned and Owned Social Media Ads (sorry this is really still Online Advertisement despite the Social Media tag) … we consumers are much more susceptible to friend’s endorsements or our favorite brands (for that matter) than the mass fabricated plain old online  advertisement that most of us are blind to anyway (or get annoyed by which from awareness is not necessarily un-intended ).

All in all

Maybe the Little Boy will not speak up as the Emperor is far from naked!

METHODOLOGY

See my Social Media Valuation Blog “A walk on the Wild Side”.

Following has been assumed in FB Valuation Assessment:

  1. WACC 9.4%
  2. 2012 FB capture 6% of total online ad spend.
  3. FB gains a sustainable share of online ad spend X%.
  4. 5 yr linear ramp-up from 2012 6% to X%, and then maintained.
  5. Other revenues 15% in 2012, linearly reduced to 10% after 5 yrs and then maintained.
  6. Assume FB can maintain a free cash flow yield of 25%.

, , , ,

6 Comments

Social Media Valuation …. a walk on the wild side.

Lately I have wondered about Social Media Companies and their Financial Valuations. Is it hot air in a balloon that can blow up any day? Or are the hundred of millions and billions of US Dollars tied to Social Media Valuations reasonable and sustainable in the longer run? Last question is particular important as more than 70% of the value in Social Media are 5 or many more years out in the Future.  Social Media startup companies, without any turnover, are regularly being  bought for, or able to raise money at a value, in the hundreds of millions US dollar range. Lately, Instagram was bought by Facebook for 1 Billion US Dollar. Facebook itself valued at a $100B at its IPO. Now several month after their initial public offering, Facebook may have lost as much as 50% of the originally claimed IPO value.

The Value of Facebook, since its IPO,  has lost ca. 500 Million US Dollar per day (as off 30-July-2012).

What is the valuation make-up of Social Media? And more interestingly what are the conditions that need to be met to justify $100B or $50B for Facebook, $8B for Twitter, $3B (as of 30-July-2012, $5B prior to Q2 Financials) or $1B for Instagram, a 2 year old company with a cool mobile phone Photo App? Is the Social Media Business Models Real? or based on an almost religious belief that someday in the future it will Return On Investment. Justifying the amount of money pumped into it?

My curiosity and analytical “hackaton” got sparked by the following Tweet:

Indeed! what could possible justify paying 1 Billion US Dollar for Instagram, which agreeably has a very cool FREE Smartphone Photo App (far better than Facebook’s own), BUT without any income?

  • Instagram, initially an iOS App, claims 50 Million Mobile Users (ca. 5 Million unique visitors and 31 Million page-views as of July 2012). 5+M photos are uploaded daily with a total of 1+ Billion photos uploaded. No reported revenues to date. Prior to being bought by Facebook for $1 Billion, was supposed to have been prepared for a new founding round valued at 500 Million US$.
  • Facebook has 900M users, 526M (58%) active daily and 500M mobile users (May 2012). 250M photos are uploaded daily with a total of 150 Billion photos. Facebook generated ca. $5B in revenue in 2011 and current market cap is ca. $61B (24 July 2012). 85% of FB revenue in 2011 came from advertisement.

The transaction gives a whole new meaning to “A picture is worth a Billion words”  … and Instagram is ALL about PICTURES & SOCIAL interactions!

Instagram is a (really cool & simple) mobile & smartphone optimized App. Something that would be difficult to say about FB’s mobile environment (in particular when it comes to photo experience).

One thing is of course clear. If FB is willing to lay down $1B for Instagram, their valuation should be a good deal higher than $1B (i.e., ca. $4+B?). It will be very interesting to see how FB plans to monetize Instagram. Though the acquisition might be seen as longer-outlook protective move to secure Facebook’s share of the Mobile Market, which for Social Media will become much more important than the traditional desktop access.

So how can we get a reality check on a given valuation?

Lets first look at the main Business Models of today (i.e., how the money will be or are made);

  1. Capture advertising spend – typically online advertisement spend (total of $94B in 2012 out of an expected total Media Ad spend of $530B). With uptake of tablets traditional “printed media” advertising spend might be up for grabs as well (i.e., getting a higher share of the total Media Ad spend).
  2. Virtual Goods & credits (e.g., Zynga’s games and FB’s revenue share model) – The Virtual Economy has been projected to be ca. $3B in 2012 (cumulative annual growth rate of 35% from 2010).
  3. Payed subscriptions (e.g., LinkedIn’s Premium Accounts: Business Plus, Job Seeker, etc or like Spotify Premium, etc..).
  4. B2B Services (e.g.,. LinkedIn’s Hiring Solutions).

The Online Advertisement Spend is currently the single biggest source of revenue for the Social Media Business Model. For example Google (which is more internet search than Social Media) takes almost 50% of the total available online advertisement spend and it accounts for more than 95% of Google’s revenues. In contrast, Facebook in 2011 only captured ca. 4+% of Online Ad Spend which accounted for ca. 85% of FB’s total revenue. By 2015 eMarketeer.com (see http://www.emarketer.com/PressRelease.aspx?R=1008479) has projected the total online advertisement spend could be in the order of $132B (+65% increase compared to 2011). USA and Western Europe is expected to account for 67% of the $132B by 2015.

Virtual Goods are expected to turn-over ca. $3B in 2012. The revenue potential from Social Networks and Mobile has been projected (see Lazard Capital’s Atul Bagga ppt on “Emerging Trends in Games-as-a-Service”) to be ca. $10B worldwide by 2015. If (and that is a very big if) the trend would continue the 2020 potential would be in the order of $60B (though I would expect this to be a maximum and very optimistic upside potential).

So how can a pedestrian get an idea about Social Media valuation? How can one get a reality check on these Billionaires being created en mass at the moment in the Social Media sphere?

“Just for fun” (and before I get really “serious”) I decided see whether there is any correlation between a given valuation and the number of Unique Visitors (per month) and Pageviews (per month) … my possible oversimplified logic would be that if the main part of the Social Media business model is to get a share of the Online Advertisement Spending there needs to be some sort of dependency on the those (i..e, obviously whats really important is the clickthrough (rate) but lets be forget this for a moment or two):

 The two charts (log-log scaled) shows Valuation (in Billion US$) versus Unique Visitors (in Millions) and Pageviews (in Billions). While the correlations are not perfect, they are really not that crazy either. I should stress that the correlations are power-law correlations NOT LINEAR, i.e., Valuation increases with power of unique and active users/visitors.

An interesting out-lier is Pinterest. Let’s just agree that this does per see mean that Pinterest’s valuation at $1.5B is too low! … it could also imply that the rest are somewhat on the high side! 😉

Note: Unique Visitors and Pageview statistics can be taken from Google’s DoubleClick Ad Planner. It is a wonderful source of domain attractiveness, usage and user information.

Companies considered in Charts: Google, Facebook, Yahoo, LinkedIN, Twitter, Groupon, Zynga, AOL, Pinterest, Instagram (@ $1B), Evernote, Tumblr, Foursquare, Baidu.

That’s all fine … but we can (and should) do better than that!

eMarketeer.com has given us a Online Advertisement Spend forecast (at least until 2015). In 2011, the Google’s share amounted to 95% of their revenue and for Facebook at least 85%. So we are pretty close to having an idea of the Topline (or revenue) potential going forward. In addition, we also need to understand how that Revenue translates into Free Cash Flow (FCF) which will be the basis for my simple valuation analysis. To get to a Free Cash Flow picture we could develop a detailed P&L model for the company of interests. Certainly an interesting exercise but would require “Millions” of educated guesses and assumptions for a business that we don’t really know.

Modelling a company’s P&L is not really a peaceful walk for our interested pedestrian to take.

A little research using Google Finance, Yahoo Finance or for example Ycharts.com (nope! I am not being sponsored;-) will in general reveal a typical cash yield (i.e., amount of FCF to Revenue) for a given type of company in a given business cycle.

Examples of FCF performance relative to Revenues: Google for example has had an average FCF yield of 30% over the last 4 years, Yahoo’s 4 year average was 12% (between 2003 and 2007 Google and Yahoo had farily similar yields ).  Facebook has been increasing its yield steadily from 2009 (ca. 16%) to 2011 (ca. 25%), while Zynga had 45% in 2010 and then down to 13% in 2011.

So having an impression of the revenue potential (i.e., from eMarketeer) and an idea of best practice free cash flow yield, we can start getting an idea of the Value of a given company. It should of course be clear that we can also turn this Simple Analysis around and ask what should the Revenue & Yield be in order to justify a given valuation. This would give a reality check on a given valuation as the Revenue should be in reasonable relation to market and business expectations.

Lets start with Google (for the moment totally ignoring Motorola;-):

Nothing fancy! I am basically assuming Google can keep their share of Online Advertising Spend (as taken from eMarketeer) and that Google can keep their FCF Yield at a 30% level. The discount rate (or WACC) of 9% currently seems to be a fair benchmark (http://www.wikiwealth.com/wacc-analysis:goog). I am (trying) to be conservative and assumes a 0% future growth rate (i.e., changing will in general have a high impact on the Terminal Value). If all this comes true, Google’s value would be around 190 Billion US Dollars. Today (26 July 2012) Google Finance tells me that their Market Capitalization is $198B (see http://www.google.com/finance?q=NASDAQ:GOOG) which is 3% higher than the very simple model above.

How does the valuation picture look for Facebook (pre-Zynga results as of yesterday 25 July 2012):

First thought is HALLELUJAH … Facebook is really worth 100 Billion US Dollars! … ca. $46.7 per share… JAIN (as they would say in Germany) … meaning YESNO!

  • Only if Facebook can grow from capturing ca. 6% of the Online Advertisement Spend today to 20% in the next 5 – 6 years.
  • Only if Facebook can improve their Free Cash Flow Yield from today’s ca. 25% to 30%.
  • Only if Facebooks other revenues (i.e., from Virtual Goods, Zynga, etc..) can grow to be 20% of their business.

What could possible go wrong?

  • Facebook fatigue … users leaving FB to something else (lets be honest! FB has become a very complex user interface and “sort of sucks” on the mobile platforms. I guess one reason for Instagram acquisition).
  • Disruptive competitors/trends (which FB cannot keep buying up before they get serious) … just matter of time. I expect this to happen first in the Mobile Segment and then spread to desktop/laptop.
  • Non-advertisement revenues (e.g., from Virtual Goods, Zynga, etc..) disappoints.
  • Need increasing investments in infrastructure to support customer and usage growth (i.e., negative impact on cash yields).
  • The Social Media business being much more volatile than current hype would allow us to assume.

So how would a possible more realistic case look like for Facebook?

Here I assume that Facebook will grow to take 15% (versus 20% above) of the Online Ad spend. Facebook can keep a 25% FCF Yield (versus growing to 30% in the above model). The contribution from Other Revenues has been brought down to a more realistic level of the Virtual Goods and Social Media Gaming expectations (see for example Atul Bagga, Lazard Capital Markets, analysis http://twvideo01.ubm-).

The more conservative assumptions (though with 32% annual revenue growth hardly a very dark outlook) results in a valuation of $56 Billion (i.e., a share price of ca. $26). A little bit more than half the previous (much) more optimistic outlook for Facebook. Not bad at all of course … but maybe not what you want to see if you paid a premium for the Facebook share? Facebook’s current market capitalization (26 July 2012, 18:43 CET) is ca. $60B (i..e, $28/share).

So what is Facebooks value? $100B (maybe not), $50+B? or around $60+B? Well it all depends on how shareholders believe Facebook’s business to evolve over the next 5 – 10 (and beyond) years. If you are in for the long run it would be better to be conservative and keep the lower valuation in mind rather than the $100B upside.

Very few of us actually sit down and do a little estimation ourselves (we follow others = in a certain sense we are financial lemmings). With a little bit of Google Search (yes there is a reason why they are so valuable;-) and a couple of lines of Excel (or pen and paper) it is possible to get an educated idea about a certain valuation range and see whether the price you paid was fair or not.

Lets just make a little detour!

Compare Facebook’s current market capitalization of ca. $60B (@ 26 July 2012, 18:43 CET) at $3.7B Revenue (2011) and ca. $1B of free cash flow (2011). Clearly all value is in anticipation of future business! Compare this with Deutsche Telecom AG with a market capitalization of ca. $50B at $59B (2011, down -6% YoY2010) and ca. $7.8B of free cash flow (2011). It is Fascinating that a business with well defined business model, paying customers, healthy revenue (16xFB) and cash flow (8xFB) can be worth a lot less than a company that relies solely on anticipation of a great future.  Facebook’s / Social Media Business Model future appear a lot more optimistic (the blissfull unknown) than the Traditional Telco Business model (the known” unknown). Social Media by 2015 is a game of maybe a couple of hundred Billions (mainly from advertisement, app sales and virtual economy) versus the Telecom Mobile (ignoring the fixed side) of a Trillion + (1,000 x Billion) business.

Getting back to Social Media and Instragram!

So coming back to Instagram … is it worth paying $1B for?

Let’s remind ourselves that Instagram is a Mobile Social Media Photo sharing platform (or Application) serving Apple iOS (originally exclusively so) and Android. Instagram has ca. 50+M registered users (by Q1’2012) with 5+M photos uploaded per day with a total of 1+B photos uploaded. The Instagram is a through-rough optimized smartphone application. There are currently more than 460+ photo  apps with 60Photos being a second to Instagram in monthly usage (http://www.socialbakers.com/facebook-applications/category/70-photo).

Anyway, to get an idea about Instagram’s valuation potential, it would appear reasonable to assume that their Business Model would target the Mobile Advertisement Spend (which is a sub-set of Online Ad Spend). To get somewhere with our simple valuation framework I assume:

  1. that Instagram can capture up to 10% of the Mobile Adv Spend by 2015 – 2016 (possible Facebook boost effect, better payment deals. Keep ad revenue with Facebook).
  2. Instagram’s  a revenue share dynamics similar to Facebooks initial revenue growth from Online Ad Spend (possible Facebook boost effect, better payment deals. Keep ad revenue with Facebook).
  3. Instagram could manage a FCF Yield to 15% over the period analysed (there could be substantial synergies with Facebook capital expenditures).

In principle the answer to that question above is YES paying $1B for Instagram would be worth it as we get almost $5B from our small and simple valuation exercise … if one believes;

  1. Instagram can capture 10% of the Mobile Advertisement Spend (over the next 5 – 6 years).
  2. Instagram can manage a Free Cash Flow Yield of at least 15% by Year 6.

Interesting looking at the next 5 years would indicate a value in the order of $500M. This is close to the rumored funding round that was in preparation before Facebook laid down $1B. However and not surprising most of the value for Instagram comes from the beyond 5 years. The Terminal Value amounts to 90% of the Enterprise Value.

For Facebook to breakeven on their investment, Instagram would need to capture no more than 3% of the Mobile Ad Spend over the 5 year period (assuming that the FCF Yield remain at 10% and not improving due to scale).

Irrespective;

Most of the Value of Social Media is in the Expectations of the Future.

70+% of Social Media Valuation relies on the Business Model remaining valid beyond the first 5 years.

With this in mind and knowing that we the next 5 years will see a massive move from desktop dominated Social Media to Mobile dominated Social Media, should make us somewhat nervous about desktop originated Social Media Businesses and whether these can and will make the transformation.

The question we should ask is:

Tomorrow, will today’s dot-socials be yesterday’s busted dot-coms?

PS

For the pedestrian that want to get deeper into the mud of valuation methodologies I can really recommend “Valuation: Measuring & Managing the Value of Companies” by Tim Koller, Marc Goedhart & David Wessels (http://www.amazon.com/Valuation-Measuring-Managing-Companies-Edition/dp/0470424656). Further there are some really cool modelling exercises to be done on the advertisement spend projections and the drivers behind as well as a deeper understand (i.e., modeling) of the capital requirements and structure of Social Media Business Models.

In case of interest in the simple models used here and the various sources … don’t be a stranger … get in touch!

PSPS (as of 28-July-2012) – A note on Estimated Facebook Market Capitalization

In the above Facebook valuation commentary I have used the information from Google Finance (http://www.google.com/finance?q=facebook) and Yahoo Finance (http://finance.yahoo.com/q?s=FB) both basing their Market Capitalization estimation on 2.14B Shares. MarketWatch (http://www.marketwatch.com/investing/stock/fb) appear to use 2.75B shares (i.e., 29% high than Google & Yahoo). Obviously, MarketWatch market capitalization thus are higher than what Google & Yahoo would estimate.

, , , , , , , , , , , ,

16 Comments

Machine Intelligence Blog

It's not Magic! It is mainly Linear Algebra Applied Creatively!

AI Strategy & Policy

Artificial Intelligence Strategies & Policies Reviewed. How do we humans perceive AI?; Are we allergic to AI?, Have AI aversions? or do we love AI or is it hate? Maybe indifference? What about trust? More or is it less than in our peers? How to shape your Corporate AI Policy; How to formulate your Corporate AI Strategy. Lots of questions. Many answers leading to more questions.

Things I tend to forget

if I don't write it down, I have to google for it again

Wireless End-to-End

A blog serving the wireless communications industry