Efficient Frontier Insights

February 2011 Display Publisher CPMs: Social media is the flavor of the month

In our January Publisher Trends we noted that Video Publisher Impression share declined significantly from December levels. The story of February is that social media is the flavor of the month, with one-third of Display impressions in the US coming from this Publisher type. To put this into perspective, this is almost double the level of December 2010.

Figure 1 shows the different Publisher types compared against each other, and Figure 2 shows actual numerical performance per Publisher type and how they trended MoM (month over month).

Figure 1: US Display Publisher Category CPMs – February 2011

Figure 2: US Display Publisher Type Trends

Apart from the social wave, what is also readily apparent is the lower CTRs (click-through rates) in February across all Publisher types. We noted in our February 2011 Display Trends release that impression levels were up 13% MoM and 273% YoY (year over year). So Efficient Frontier advertisers are increasingly understanding the interaction between Display and other digital channels like SEM and social media, which is contributing to increased marketing expenditure in Display due to the overall digital ROI (return on investment) benefit. Additionally, while the proportional share of different segments might vary, overall volume is certainly growing significantly for EF US Advertisers in Display.

In the wider Display eco-system, with many high-quality publishers diverting evermore of their non-guaranteed Display inventory to private exchanges managed by Supply Side Platforms (SSPs) like AdMeld, Rubicon, PubMatic, etc., it is imperative that savvy marketers are paying the optimal amount for Publisher inventory.

Chris Jacob

The Google “Like”

Google has just announced that it is introducing a +1 feature that users will see as a small button next to a Google search result and ad. When clicked, the “+1” will be shared with everyone in the user’s social network. The data will also be used in aggregate to improve search results for everyone. These features have been explained in detail here and here.

Google and SoLoMo

“SoLoMo” has been used by many leading internet analysts to describe the convergence of social, local and media applications on the Internet. Google appears to be following the same path. Google integrated with Twitter a while ago to show hot trending topics as part of the search results. It has also prioritized local search and introduced several innovations and features in the past year. Google’s aggressive push on mobile advertising and the Android platform has given it a commanding 97% advertising spend share in the mobile space. The incorporation of the +1 button is the logical next step in their SoLoMo convergence. This will have an impact in two key areas:

#1 The core search product

Google’s algorithms rank search results based on an inferred calculation of preferences. In other words, based on your activity and connection to other pages, it gleans if a page is useful or not. While this approach has been remarkably successful, it is limited by the indirectness of the approach, i.e. the surfer is not telling you that he likes the page, but shows several signs that he likes it. +1 will give Google a direct answer from the user and is a strong and direct signal of intent and can only help their search rankings. It will also help Google to provide surfers with highly customized search and ad results, which will benefit both the user and the advertiser.

#2 Rich demographic information to advertisers

Until now, it was only the “like” button on Facebook that provided rich preference data at volume to advertisers. However, +1s will enable Google to tie in preference information with demographic data that it gets from the Google profile page. Google’s preference information has the potential to be even more useful to advertisers as Google searches are intent-driven vs. Facebook “likes”, which are more awareness-centric. In other words, Facebook data, on average tells you who is INTERESTED in your product or service, while Google data tells you who is intent on BUYING it.

This feature also promises to take offer and ad copy testing to a new level. One should be able to find out how many consumers like an offer or a price via the number of +1s. Like the “like” button, this promises to take market research to a whole new level.

Our take

At a first glance, the +1 button promises to do just what the “like” button does. While it will enable users to share information across their social graph, it will also give both Google and advertisers rich intent information that will help them create better more targeted advertising campaigns. Yet many questions remain. For one, people like to share content such as videos, news and articles. Will people share organic search results or a text ad? Second, advertisers can remarket to users who have “liked” something. Will one be able to do the same with +1s? Finally, but most importantly, is the question of social graph. Facebook has a social graph and Google does not. Google’s attempts to create one have failed so far. The crucial chicken and egg question is whether Google has enough social graph data to make +1 interesting for consumers and whether the feature can provide enough incentive for users to develop their Google social graph. On this front, Facebook has a distinct and definite advantage at this stage.

The coming months promise to be very interesting. Google and Facebook are becoming the biggest rivals in the Internet world. Google is attempting to attract not just intent-driven users, but also users who want to spend time socializing on the internet. Facebook is doing the exact opposite. Facebook does not have a search product, yet one can’t help but anticipate that there will be one in time. Bing is also in the fray, trying to garner market share from Google with its many innovations. All innovations in social, local, mobile and search are aiming towards a more relevant user experience. This is a big net positive for consumers as it means that the Internet is becoming a more relevant and interesting place. Advertisers benefit too as they now have unprecedented access to rich demographic data that is both actionable and measureable.

Dr. Siddharth Shah
Sr. Director, Business Analytics

 

Google Comparison Ads: Friend or Foe?

For consumers and performance marketers who may not be aware of Google Comparison Ads (GCA), it’s a product that enables users to compare multiple and relevant offers from various advertisers. Google Comparison Ads originated in the UK marketplace with the mortgage vertical and was first brought to the US in October 2009. After positive user and advertiser feedback, Google expanded the offering to credit card and banking products (CDs, savings and checking accounts) in late 2010. Thus far the mortgage product has already entered the paid model, but credit cards/banking products are still in an ‘alpha’ free trial period for advertisers.

As always, Google is leading the way in looking for innovative ways to provide more relevant information to the consumer marketplace, which should in turn help performance marketers reach prospects who are most interested in their products and services. At first glance, GCA brings speed, transparency, usability and privacy to the PPC (Pay per click) ecosystem. But is this complementary to regular paid search or is it in fact competing? Let’s take a deeper look at the pros and cons of this extension. 

How GCAs are a marketer’s “friend”

Customization and standardization. Google has opted to standardize the way information is presented and the type of material advertisers are able to provide within a product schema. This minimizes clutter, bait and switch offers and gives a fair assessment of each major brand. Google has also made this product for direct marketers only; no listing sites/aggregators are part of the current program. Outside of clean and concise information, marketers are able to create custom offers based on their goals and objectives to find the customers who will provide the greatest lifetime value to their business. Advertisers create ‘rules’ for specific card/customer criteria and if these inputs are met, they will bid the maximum lead price they are willing to pay to capture that customer. 

Targeting and Conversion Rates. Not only are direct marketers able to customize their offers, but this allows them to target users at a much more granular level. Prospects have a variety of inputs to ‘pre-define’ what they are looking for and their creditworthiness. Giving users the ability to search on a number of relevant attributes allows Google to serve up more targeted ads. This funnels more valuable leads to advertisers based on how users have ‘pre-qualified’ themselves. It’s a win-win situation as users receive an offer that matches their specific needs and increases the propensity for an instant ‘approval’.

Using internal Efficient Frontier data thus far, a major financial services client has seen an approximate lift of 120% in quality of conversion in comparison to regular Paid Search: 

Visibility and Custom UI. For advertisers who opt into GCA, there is a separate dashboard similar to that of Google AdWords. Like setting up a typical account, it is straightforward, painless and intuitive, and your Google Account Management team can facilitate this. Marketers can also set up a product feed via FTP to ensure that product offers are relevant and up-to-date with accurate financial credentials. Most importantly, you can use the toolset to create campaigns, rules, bids, etc. The workflow is imitative again of AdWords – it is easy to view a list of product offerings, set rules and conditions and of course review your lead information. Using tracking and lead IDs per product, Google not only identifies duplicate leads, but also only charges you once within that session upon ‘clicking out’ on the “apply now” button. Lastly, they allow third-party piggyback certified tracking to follow users through the entire conversion funnel. 

How GCAs can be a marketer’s “foe”:

Payment Model. Credit cards and bank products are currently in beta trial, in which all generated leads are provided free-of-charge to participating advertisers. Although that is delightful in the short-term, Google is rapidly approaching the paid environment. From what we know currently, position and placement is arbitrarily determined by a variety of hierarchical inputs, using credit cards as an example: APR, Period and Bid Price. With the ‘payment model,’ advertisers will need to bid above a floor price to enter the auction. The product that makes the auction will appear based upon the user’s inputted criteria and sort preference. Some performance marketers might find this counter-intuitive to bidding a premium for top placement, for if a user changes the filters and or sorting options, they may in fact be trying to achieve position 1, but in reality, secure position 3. Google has been urging advertisers to not look at Comparison Ads as a way to pay to be in top position, but instead to look at the bidding methodology as paying for what the lead is worth and thus they should be happy with the quality of the lead.  

Optimization. Another major concern for advertisers and a justified one at that is the lack of Quality Score in the paid product. Google Project Managers have said that they will definitely be looking into this down the road, but there are no current plans in the immediate pipeline. Thus with no Quality Score component, larger marketers may have more leverage (investment) to inflate the marketplace and increase the minimum reserve CPC to obtain a particular position. Currently, Google is choosing not to pass critical ‘value parameters’ via the API to platforms and necessary keyword information for modeling and bidding appropriately based on conversion data. As a net result, it is imperative that agencies and performance marketers collaboratively work together to interpret conversion data and patterns to efficiently and effectively bid in a rules-based environment.

Google Comparison Ads Position 1 and CPCs. At this point in time, Google is only serving Comparison Ads against a limited number of upper funnel generic terms (i.e. “credit cards”, “mortgages”, “certificate of deposit” etc.). When a user physically enters this query, one will notice that Google automatically supplants itself in the first position, which pushes down the rest of the sponsored ads. 

Although this does not arbitrarily inflate CPCs in the sponsored auction, advertisers are losing real estate on the SERP and potentially a qualified prospect to a competitor via Comparison Ads. Yes, the product was created to improve the user experience and provide valuable leads to performance marketers, but it is critical for agencies to monitor the SOV gained and or lost by the introduction of this tool. And of course, without the auction control of QS, and limited visibility to optimize bids, will that potential prospect via GCA be more or less expensive than a standard paid search event? The other unidentified variable is how many keywords Google will decide to include in the list to enable Comparison Ads.

Like any new product, there is still a large amount of mystery about Google Comparison Ads. Google has been a bit ambiguous about providing transparent details on the auction marketplace in GCA and causality within the sponsored results. To be fair, any product will continue to evolve over time to adapt to a capitalistic marketplace. Efficient Frontier is proactively diving into our client data, so look for future blogs on this topic as we better understand the finer points of this Google product.

-Mike Fogarty
Account Manager

 

 

Statistics Are Like a Bikini

“Statistics are like a bikini – what is revealed is interesting but what is hidden is crucial.” – Aaron Levenstein, Associate Prof. emeritus of Business, Baruch College

In my latest Search Engine Land column, I discuss how to make your statistics skimpy yet meaningful. This is because statistics tell us about general trends and properties about our data but often they hide information that would give key insights. The goal of every analyst then is to make statistics skimpy – but at the same time, make them as revealing as possible while not overwhelming the decision-maker with unnecessary numbers.

In this post, I focus on working the average and median statistics. While widely used to describe trends and data, they have a few limitations, which can give you a wrong impression of what is going on. Here's a link where I show you how: http://searchengineland.com/how-to-make-your-statistics-skimpy-yet-meaningful-68740

Dr. Siddharth Shah
Sr. Director, Business Analytics

Facebook 101 – Bids, Budgets, & Segments, Oh My!

When looking at a Facebook marketing pie, some of the main ingredients are the bids, budgets, and segments. Here are some tricks of the trade we recommend.

Bids
Facebook provides a suggested bid range when advertisers upload their ads. Our recommendation is to start at the lowest end of the bid range, and gradually increase if you have a more conservative budget. If you have a more flexible/aggressive budget, start at the higher end of the range, and work your way down. The logic behind this plan is that the higher initial bid will give you better intent in Facebook’s algorithm.

Budgets
Facebook offers two budget options (set on the account level): lifetime and daily. Lifetime is more akin to accelerated on Google, while daily is similar to the standard option on Google. The lifetime budget option will bleed through your budget immediately, so our recommendation is to go with the daily budget option, so you can ensure that ads are running throughout the day. Use the daily budget as a cap because it will all be spent by day’s end.

Segments
The real fun with Facebook is in the segments, which are set on the campaign level. The segments are essentially all of the likes and interests, as well as demographic information, available to target. You can target based on all the basics, like age, gender, relationship status, interested in, schools, work, etc, as well as the many likes and interests available. Since this pool is so large, our recommendation is to start with a handful (i.e., 20 – 50) of these demographics and likes/interests you’d like to test, and then refine, refine, refine. If a particular segment works well for you (for example, women ages 25 – 35 interested in traveling), then target them with more specific ads/images. If a particular segment converts poorly, feel free to cut it. The best results I’ve seen so far have been from starting broad and then refining segments.

It’s also important to align yourself with a company that can handle auto-splitting. Auto-splitting is automatically splitting your segments so you can set unique bids on those ads. For example, if you have two ad variations and want to target both men and women between the ages of 18 – 20 who are interested in hiking and biking, through auto-splitting, you can take those two ads and actually create 24 unique ads with unique bids. See below:

This not only gives you more insight, but more control over your bids and performance. There will be more to come in terms of likes and interests and some interesting tools our engineers are working on, so stay tuned. For now, enjoy these insights and have fun getting started.

-Karen Maciolek
Senior Account Manager

Making the most of the AdCenter platform

Over the last year as Bing and Yahoo have migrated to share the AdCenter platform, the two engines have seen a decrease in both spend and click share. As quantity has decreased, quality has increased with RPCs (revenue per click) up. The widely predicted CPC (cost per click) increase has materialized but largely because of improved quality, not because of the increased number of competitors as was expected. This quality improvement is the key feature of the YaBing platform and has come about because of a culling of the partner network – volume has shrunk dramatically while relative quality has increased to be on par with Google.

Shrinking Volume

The alliance transition completed in Q4 2010, at which point Efficient Frontier saw a drop of 5.5% in share of spend going to both Bing and Yahoo. Throughout 2010, Bing held largely constant as Google essentially took share from Yahoo, furthering Google’s dominance of the market.

These aggregate numbers mask a significant variation within different verticals. Looking at four of Efficient Frontier’s largest verticals, we see as much as a 20 percentage point difference between retail -- a Google stronghold (Yahoo/Bing 10%) -- and Finance (Yahoo/Bing 30%), a long-time Yahoo power. The aggregate directional changes we’ve seen will have to be looked at in the context of your specific business to assess the opportunity.

The volume drop on Bing/Yahoo can be seen clearly through the transition in the chart below. Starting in September and through the end of the year as traffic was being switched over from Panama to AdCenter, the percentage of clicks coming from the two engines dropped steadily – 21% overall.

At the same time we saw the quality of the traffic improve significantly. Revenue per click showed a steady upward trend over the four-month period.

The source of both the volume decrease and quality increase is significant changes to Yahoo’s partner network. As discussed in other posts, a huge part of Yahoo’s traffic comes from partner sites. These have traditionally converted at a much lower rate than traffic from search.yahoo.com. Our data shows a striking increase in the monetization quality, as measured by revenue per click (RPC) of the partner traffic.

In March of 2010, revenue per click from Yahoo partners averaged 40% of the Yahoo’s overall RPC. This compares to Google, whose much smaller set of partners (including Ask.com) converted at 80% of the overall RPC. After the alliance was completed we saw Bing-Yahoo RPCs at a similar level – about 80% of overall CPC.

As a side note, this improved traffic stands to benefit Bing and Yahoo as they no longer have to subsidize poor partner traffic. AdCenter provides great tools for managing the partner network, but we believe that many if not most users will set a single bid on the AdCenter platform. That bid will be based on the average revenue per click across partners’ sites as well the “owned and operated” sites (i.e. search.yahoo.com and Bing.com). If partner traffic is pulling down the average conversion rate, advertisers will lower their CPCs accordingly. In essence, bing.com and search.yahoo.com will be subsidizing the poor network traffic. Additionally, Bing and Yahoo have to share some percentage of the click value with these partners further undercutting their revenue. The smaller, higher quality partner network we see today, combined with tools for controlling bids on the partner network just makes good business sense both Microsoft and Yahoo.

So what should one do?

• Look at your data. As illustrated above, the relative performance of the different engines and their respective partner networks varies greatly by industry.

• Make sure you are present on all parts of the AdCenter platform. This means full keyword coverage as well as content. Why spend hours trying to eek more performance out of existing keywords when you can get a significant volume lift by just having a full presence on search keywords and content themes. Content on AdCenter in particular is a significant opportunity that is ignored by too many people.

• Manage the partner networks. Partner traffic is huge and even though Yahoo’s network has improved, it requires close monitoring. Exclude underperforming domains and consider bidding on it separately. Do not turn off the partner network unless absolutely necessary – remember presence!

-Shay O’Reilly
Senior Business Analyst

 

The Real Value of Facebook Ads - Exploring the "Likeberg"

Some view the Facebook ads marketplace as biddable, and therefore label it a lower-funnel opportunity like search (i.e., direct marketing). Some have argued that Facebook ads are not directly tied to consumer intent, use images, and should be characterized as upper-funnel (i.e., brand marketing). I'm here to argue that because of the new form factor of social media, Facebook ads can't and shouldn't be characterized as upper or lower-funnel.

Let's start by examining the upper-funnel potential of FB marketplace ads. The Facebook ad is a small, static image with limited text. The small stature of the creative unit is reason enough to doubt its potential to alter materially brand awareness, brand perception, or purchase intent.

Yet there is evidence of substantial impact from seeing ads that reference friends who like that ad/product/service, or seeing an ad that is a direct endorsement by a friend via sponsored stories. Facebook touts a study that the ads drive a 4x lift in purchase intent. Clearly, there is some potential for Facebook ads to address upper-funnel metrics.

Now let's take a look at the lower-funnel potential. The text-heavy, CPC-biddable, keyword-targetable Facebook ads remind us of search. But the signal from Facebook's keyword targeting -- primarily leveraging the data from users' profile, such as geography and likes -- is not nearly as powerful an indication of interest as the clear intent of a search query.

The result of the variance between Facebook's affinity targeting and search's intent targeting is that Facebook ads tend not to perform like search. (CTRs and conversion rates tend to be lower.) However, if optimized across targets, bids and creative, the ads can perform like search and deliver strong lower-funnel/direct marketing performance.

While there is value in both upper- and lower-funnel metrics with Facebook, the larger value is in what happens socially as a result of a successfully placed ad. Using a recent buzzword, it is the "social amplification" rather than just the ad itself that creates the most value.

The average Facebook user -- of which there are more than 600 million -- has 140 friends. When a brand is "Liked" by a Facebook user, the action of "Liking" that brand is broadcast to all friends in their newsfeed. If that "Like" is "Shared" (a feature that is now merged with the "Like" button), the action is broadcast to all friends' newsfeed in the form of a story. Therefore, the mere action of "liking" both on and off Facebook is very valuable. But that is just the tip of Like-iceberg, or as I like to call it the Likeberg.

Facebook allows all fans and connections of fans to be targeted directly with marketplace ads. When possible, Facebook adds details about friends who "Like" the brand to ads delivered to connections. Adding the brand "Like" to the ad unit itself adds a powerful recommendation element to the ad.

Facebook's Sponsored Story Ads take this a step further using the copy of the individual's "Like" as the ad itself. Sponsored Stories are essentially a turnkey way to broadcast recommendations with the control residing with the recommended (brand) vs. the recommeder (user). Clearly, "Likes" are the fuel for Facebook ad campaigns.

We're just beginning to descend into the core of the Like-berg. All "Likes" give brands the ability to publish to the newsfeed of fans. Think of newsfeed publishing as a sort of opt-in email list that has arguably has more buy-in and greater visibility than email, especially with certain populations such as the youth category. Brands can offer coupons, promotions, or offers as well as leverage FB elements like polls and apps to drive engagement. The key to extracting value from news feed publishing is getting the right value-mix to customers.

Each consumer who becomes a fan provides a rich set of data to analyze. This data-set becomes all the more rich as they engage with applications. This data can both inform targeting and tie directly into Facebook ad optimization.

Obtaining "Likes" probably sounds pretty good to most marketers. Unfortunately, we can't put an actual value on social amplification. Yet. There are companies like ChompOn that have quantified the value of a "Share" in terms of daily deals. Also, Context Optional has rich data that quantifies the virality of application engagement driven by Facebook ads (ranging from 3-12X). But attempting to value likes on comparative metrics to existing valuations seems a little useless. The bottom line is that the value of a like could be nothing if a brand doesn't extract value, or invaluable if used to its full extent.

I would be remiss not to mention one last "Like" wrinkle. To have your news feed posts show up to your fans, they have to regularly engage with some element of your experience: polls, apps, offers, product likes, etc. A good way to drive this re-engagement is with Facebook marketplace ads -- a brilliant revenue move by Facebook.

To bring this full circle, the full value of Facebook advertising is much larger than the immediate impact of the ad. To not consider the social impact of marketing on Facebook is to miss the forest for the trees. And in my experience, marketers are much more interested in growing a forest of revenue opportunities than cultivating a single sprig of sales.

Justin Merickel

V.P.Marketing and New Product Development

GOING BEYOND THE US IN DISPLAY

A big question many U.S. advertisers have when expanding their businesses beyond a domestic offering is what price differential they can expect with their advertising. In the search engine marketing (SEM) world, many companies including Efficient Frontier release reports and insights about international trends. However, there is far less insight when it comes to Display.

Of course, non-U.S includes more than 190 countries that speak numerous languages, and includes first-world and third-world nations, so it cannot be taken as an absolute, applicable to every foreign market. However, the trend data should give U.S. advertisers, and indeed international advertisers, a sound directional insight as to what they can expect cost-wise from their Display advertising, which would significantly aid their forecasting and planning.

In Figure 1, we show the relative CPM (cost per thousand impressions) Index of the U.S. and non-U.S. Display markets through 2010, with the U.S. CPM index for each month used as a baseline across the EF client base.

Figure 1: EF U.S. vs. Non-U.S. Client CPM Index Trends 2010

Throughout 2010, the non-U.S. CPMs in display have been about 70% cheaper than the U.S. CPMs, with that gap widening as the year progressed and macro U.S. economic indicators improved.

This covers all verticals, so naturally there could be differences if we compared this across a given vertical. We will document more detailed insights in that vein in the coming weeks and months.

Another interesting point to note is that just as non-U.S. CPMs were about one-third the price of U.S. CPMs, the CTRs of these ads for non-U.S. was about one-third the U.S. CTRs for 2010. It is difficult to draw a definitive conclusion from that statistic alone beyond that, U.S. consumers have a greater propensity to interact with Display marketing messages.

Ultimately, the level of inventory, nature of the competitive landscape and sophistication of the Display marketplace infrastructure outside of the U.S. is still a couple of levels behind, and this plays a significant part in the Display inventory pricing differential. This is quickly changing as all major players in the Display ecosystem are rapidly expanding their operations, footholds and product offerings to Europe and Asia in particular.

It is for this reason that any partner an advertiser chooses in this modern age not only understand the dynamics of advertising globally, but can also execute that marketing effort to the same optimal performance standards they would expect in their domestic market.

Chris Jacob

Facebook and Google – An Apples to Oranges Comparison and Why Current Metrics Don’t Work

Google and Facebook’s battle for eyeballs, talent and advertising dollars prompted us to look at our own data. We looked for answers to questions on everyone’s minds: (1) How does Facebook measure against Google in terms of spend, CPC (cost per click) and other key metrics? (2) What do these metrics tell us about Facebook’s potential ability to grow?

Metrics



We looked at the Facebook campaigns of many of our clients across several key metrics and have summarized our findings in the chart above.

 Spend

In this advertiser set, the average Facebook budget was 6% of the budget on Google, a figure that, when extrapolated out to Google’s entire revenue base, suggests Facebook’s revenue was $1.85 BN in 2010 (Google’s revenues was $29.32 BN in 2010).  Our extrapolation lends credence to the rumor that Facebook’s revenue was $2 BN last year.

 CPC, CPM and CTR

CPCs on Google and Facebook are virtually identical. However, the CPMs tell a completely different story. On a CPM basis, Facebook ads cost only 5% of that on Google. Put differently, advertisers are able to get 37% more impressions on Facebook while paying only 6% of their Google budgets. 

Our preliminary take

A side-by-side metrics analysis between Google and Facebook on traditional search/display metrics is not appropriate. Google is an excellent advertising medium for reaching in-market consumers who are clearly articulating their intent with their search terms.  Facebook cannot deliver the same kind of user-intent based targeting, at least not today.

Facebook does deliver something Google cannot today: virality.  And it is the power of the viral network effect that makes Facebook so powerful and valuable.  The primary means a brand has to achieve a viral effect is through consumers’ liking their company, products, or even their ads.  The action of liking a brand brings a host of very valuable exposure, data, and communications opportunities including:

a.     A rich data set about their customers and prospects

b.     The ability to communicate to consumers who “like” the brand via ads and regularly published news feed updates

c.     A means of transmitting a brand or product message when the on-average 140 friends are exposed in their news feed to that individual’s “like” or “share” (repeat the same action if any of these friends also like/share)

d.     An expanded advertising target set via the ability to target all “friends of likes” with ads

Despite the obvious limitations of traditional metrics, the signs are highly encouraging for Facebook. The fractional CPMS and huge impression volumes would be very attractive to brand marketers who can reach a large audience more effectively than ever before. Indeed, as Mary Meeker mentioned in her Web 2.0 talk, CPMs on Facebook appear to be undervalued.

CPCs can also be argued to be undervalued on Facebook. While CPCs on Google and Facebook are very similar, today it is unlikely that the metric captures the true value of the data, targeting and virality of Facebook. In a basic example, we firmly believe that the value of a like on Facebook is far above the cost of acquiring a like today. Due to these factors, combined with the relative immaturity of Facebook ad markets, we believe Facebook CPCs are indeed undervalued.

We see Facebook as a medium that provides enormous value in a way we haven’t seen before. While their reach provides huge opportunities in a traditional direct response sense, the real value is in the social outcomes of a response that, in turn, produce revenue. We anticipate the marketplace adopting new metrics to understand, analyze and optimize to these “social amplification” variables as Facebook continues on its path of rapid growth.

Dr. Siddharth Shah
Sr. Director, Business Analytics

$50, $65 or $100 billion; Just how much is Facebook worth?

Although Facebook has recently been valued at $50 billion and $65 billion, a consensus value on the company remains unclear. Some analysts think it’s grossly overpriced while others think its value is bound to be higher. While we shall steer clear of a potentially contentious opinion on Facebook’s true value, we thought it would be an interesting exercise to assess the expectations on Facebook’s revenue and growth to justify its varying valuations.

Rather than build a traditional valuation model, we have used an unorthodox approach. Using Google as the benchmark, we have assumed that Google and Facebook have a similar low discount rate and similar margins. These assumptions allow one to use revenue as a proxy for earnings.

Google has a market capitalization of $200 billion, i.e. four times Facebook’s value. Yet Facebook’s revenues in 2010 were 16 times lower than Google’s. All things being the same, the only way one can justify the current Facebook valuations is if we expect higher growth of Facebook’s revenues than the Google benchmark. 

Using a discounted cash flow model, I calculated Facebook’s required growth based on an estimated growth of Google’s revenues. For instance, if the market expects Google’s revenues to grow at 20% a year for the next 10 years, Facebook’s revenues would have to grow at 46% a year.

The red line shows how much Facebook’s growth would have to be relative to Google’s growth. In the wide range of reasonable scenarios that we have considered, Facebook would have to grow between three and five times faster than Google over 10 years to justify its $50 billion valuation. When we repeat this exercise for different valuation scenarios, the following trends emerge:

1. Between $50 billion and $100 billion, at a 10-15% expected Google growth rate, the pressure on Facebook’s growth increases by a multiple of the expected Google growth. Thus, if we expect Google to grow at 15%, Facebook would have to grow by 36% for a $50 billion valuation and 47% for a $100 billion valuation.
2. By our estimates, Facebook would have to grow three to four times faster than Google over the next 10 years to justify its present valuation. If we assume that the expectations on Google’s revenue growth are between 10-15% for the next 10 years, then the expectation on Facebook’s growth are 30-50%. Google’s history is instructive in understanding if this could be done. Between 2004 and 2010, Google’s Q4 revenue growth was 43% compounded annually. It did so by rapidly growing at first and then steadying to a slower but robust growth mode. Facebook appears to be where Google was in 2004-2005. Thus, if Facebook is able to rapidly grow over the next four to five years and then slow down, it should still be able to meet its lofty growth expectations.

Dr. Siddharth Shah
Sr. Director, Business Analytics