Facebook Is Running Circles Around Everyone With ARPU
Summary
- COO Sheryl Sandberg has consistently focused on improving Facebook ad products.
- Facebook has high worldwide ARPU despite having a relatively low percentage of users in developed markets.
- Ads on Facebook and Instagram become more relevant and effective every year.
Introduction
My thesis is that Facebook (FB) has achieved a high average revenue per user (ARPU) due to their consistent focus and they are well positioned for it to continue climbing. One of the reasons I like the ARPU metric is that it discourages companies from inflating user counts. Bots and other fake users are all over the place; but if companies count them as real users, then their ARPU goes down.
The Facebook 10-K, Snapchat (SNAP) 10-K and Pinterest (PINS) 10-K all show annual worldwide ARPU directly. The Twitter (TWTR) quarterly releases don’t report it directly, but we calculate it based on quarterly ad revenue divided by quarterly monetizable DAUs. All 4 quarters are then added together:
Image Source: Author’s spreadsheet from 2020 10-K filings
There are 2 factors that show why Facebook’s lead is even wider than what we see above. The first is the way in which users are defined and the second is geography.
User Definitions Impact ARPU
Sometimes hyperbole is used rather than veracity when it comes to user counts but ARPU is a great equalizer. Being the denominator of ARPU, the user count is a key part of the ratio and ARPU goes up if smaller user counts are used. In the case of Facebook and Pinterest, they use monthly active user [MAU] counts which are larger than daily active user [DAU] counts. Obviously, this means the reported ARPU numbers for Facebook and Pinterest are lower than they would be if they used daily active users [DAUs]. Snapchat does in fact use DAUs while we use monetizable DAUs when calculating the ARPU for Twitter. As such, the ARPU above for Snapchat and Twitter is inflated relative to the ARPU for Facebook and Pinterest due to user definitions.
Geography Impacts ARPU
Looking at Facebook’s ARPU from 2015 to 2020, it is obvious that geography is a key factor. Per the earnings slides, Facebook has seen ARPU increases in all geographies over the years but the US & Canada segment is special; it is a pure-play on developed ad markets with high GDP per capita:
Image Source: Author’s spreadsheet based on earnings slides
Note that the yellow ARPU line above for Europe has continued to rise despite the General Data Protection Regulation [GDPR] implementation in May 2018. Facebook has said on earnings calls that their Europe ARPU would be closer to their US & Canada ARPU if it only included Western Europe and did not have Central Europe, Eastern Europe and Turkey.
Given what we see above, one would expect companies with a high percentage of users in the US & Canada to have high worldwide ARPUs. This is why it’s impressive that Facebook has the highest worldwide ARPU despite the fact that they have the lowest percentage of users in the US and Canada.
Image Source: Author’s spreadsheet from 2020 10-K filings
Continuous ARPU Improvement
In March 2008 it was announced that Sheryl Sandberg was leaving Alphabet (GOOG) (GOOGL) to join Facebook. CEO Mark Zuckerberg is a genius for bringing her onboard; her contributions towards ARPU have been extraordinary!
I believe that if another company other than Facebook had bought Instagram in September 2012, then the acquisition would not have been as successful. This is because Facebook had the right advertising know-how and a wide breadth of advertisers ready to try the new platform. Facebook does not break out the ARPU for Instagram separately, but I’m confident that if anyone else would have bought Instagram, then the ARPU there would not be as high as it is currently.
In the 1Q13 call (registration required), COO Sheryl Sandberg reminded us that she had learned many things in her time selling ads on the Internet for over a decade leading up to the call; one can tell she’s erudite in the subject of online advertising. She also mentioned a key competitive advantage at Facebook during the call:
I think with small businesses, we had an actually really deep competitive advantage, which is that people all around the world use Facebook. So when small businesses, who are historically way too busy to spend a lot of time using technology, start to use the Facebook platform, they're using something they already use as users. So once you have a time line or a profile, setting up a page is not a very big ask because you understand it and you're doing it anyway.
In earnings calls from 3Q14 to 1Q18, COO Sheryl Sandberg talks about the company focusing on 3 priorities that increase ARPU:
1. Capitalizing on the shift to mobile.
2. Growing the number of marketers using Facebook ad products.
3. Making Facebook ads more relevant and effective.
Prior to 3Q14, the #3 priority above was stated as “building our ad products” without specifically using the word “relevant.” By 3Q18, the #1 priority above pivoted to helping advertisers connect with people where they are.
Given the way Facebook treats ARPU as a priority, it isn’t surprising that the 10-K filings show vast improvement over the years:
Image Source: Author’s spreadsheet based on 10-K filings
Given the way ARPU and overall user counts have risen, revenue growth shown in the 2020 10-K is astounding. The 4-year Facebook revenue CAGR is nearly 33% going from $27.6 billion in 2016 to $86 billion in 2020. The diluted 4-year EPS CAGR is over 30% going from $3.49 in 2016 to $10.09 in 2020:
Image Source: 4Q20 earnings release
We see the same type of explosive revenue and EPS growth for many of the same reasons in the 2015 10-K:
Image Source: 2015 10-K
Dynamic Ads
I can’t be the only person out there who enjoys targeted ads. As a runner and a hiker, I’d much rather see ads about running and hiking than generic ads for things like asbestos. Some of the most targeted ads are dynamic product ads based on one’s time spent browsing and shopping at places like Amazon.
Global Marketing Solutions VP Carolyn Everson explains dynamic product ads at the December 2015 UBS Global Media and Communications Conference:
And so just to explain what dynamic product ads are, you can take a let's say a retailer and they can upload their entire product catalog. And then without having to create individual ads for millions of products which some of these obviously these retailers have millions of different product offerings, we can dynamically generate the right product and the right ad to the person at the right time based on potentially somebody leaving it in their shopping cart and they didn't purchase it or they were browsing for a particular product.
Anecdotally I’ve seen dynamic product ads based on my Amazon browsing history and shopping cart. This isn’t surprising given what COO Sheryl Sandberg said about another online marketplace, MercadoLibre (MELI), in the 3Q15 call:
Latin American e-commerce company Mercado Libre uses DPA [dynamic product ads] to re-market over 38 million products in over 13 countries.
Scale
An ecosystem with a large number of users and a large number of advertisers is well positioned to have high ARPU because more matches are possible in which ads are relevant:
Image Source: Author’s spreadsheet from 2020 10-K filings
Even back in 2012, Facebook had more than twice the number of users that any of these companies have today. The 2012 10-K shows the following MAUs for Facebook: US & Canada: 193 million; Europe: 261 million; Asia: 298 million; Rest of World: 304 million; Worldwide: 1,056 million.
Future Opportunities
A December 2020 Morningstar writeup shows the transformation of ad spending over the years:
Image Source: Morningstar
Advertising dollars will continue to shift online to companies like Facebook where the targeting and measurement tools are efficient.
Recent Challenges
Apple (AAPL) might go too far with their privacy crusade such that ad targeting suffers. I’m optimistic that Facebook will figure out how to continue with relevant and effective ads. CNBC reports that Facebook is testing ways to keep users from losing targeted ads when Apple moves forward with restrictions tied to privacy:
Image Source: CNBC
Image Source: CNBC
On February 17th, it was announced that Facebook was banning Australian users from sharing or viewing news. Facebook restored news in Australia on February 23rd and they explained their position in a post on February 24th:
It’s the publishers themselves who choose to share their stories on social media, or make them available to be shared by others, because they get value from doing so. That’s why they have buttons on their sites encouraging readers to share them. And if you click a link that’s shared on Facebook, you are directed off the platform to the publisher’s website. In this way, last year Facebook generated approximately 5.1 billion free referrals to Australian publishers worth an estimated AU$407 million to the news industry.
They go on to say that assertions claiming Facebook steals journalism are false. They point out that the concept of payment for linking can make things unworkable:
As Tim Berners-Lee, the inventor of the world wide web, warned, the Australian law could make the internet as we know it “unworkable,” arguing that it “risks breaching a fundamental principle of the web by requiring payment for linking between certain content online.”
I am optimistic that Facebook will be able to sort things out in other geographies without banning news for significant periods of time.
Valuation
Having a clear picture of normalized operating income and normalized free cash flow helps when thinking about valuation. The non-recurring $5 billion FTC settlement can be a bit confusing because of the timing. Here is the summary:
1. It impacted the FY19 income statement (Q1 & Q2).
2. It had no net impact on the FY19 cash flow statement (washed out).
3. It had no impact on the FY20 income statement.
4. It impacted the FY20 cash flow statement (Q2).
It helps to walk through the details so that comparisons between 2019, 2020 and 2021 are apples to apples.
The 2Q19 10-Q says the following about the way the FTC settlement impacts the G&A expense line in the income statement:
The increases in both periods were primarily due to legal accruals related to the FTC settlement, of which $2.0 billion was recorded in the second quarter of 2019 and a total of $5.0 billion was recorded during the first six months of 2019.
This FTC settlement accrual expense was reversed out on the 2Q19 cash flow statement because it wasn’t actually paid out until 2020. In other words, the $5 billion net income line in the red box of the cash flow statement below would be $10 billion without the FTC settlement in the G&A income statement line. The cash for this FTC settlement was not paid in this period, so the accrued expenses and other current liabilities figure in the red box on the cash flow statement is $5 billion higher than it would otherwise be such that it shows $6 billion:
Image Source: 2Q19 10-Q
Again, the 2019 six-month operating income and net income were $5 billion lower than normal because of the FTC settlement, but it is washed out in the cash flow statement above such that the $17.9 billion net cash provided by operating activities subtotal was not impacted. The implications are the same for the 2019 10-K.
The 4Q20 release has a footnote as follows: "Free cash flow in the full year ended December 31, 2020 reflects the $5.0 billion FTC settlement that was paid in April 2020." The 1Q20 cash flow statement shows $1 billion for accrued expenses and other current liabilities. The 2Q20 cash flow statement shows $(3) billion for six months accrued expenses and other current liabilities. It only went down to $(3) billion instead of $(4) billion because there must have been $1 billion of other entries mixed in.
If we didn’t know anything about the FTC settlement above, then management’s 2020 FCF number of $23 billion would be less impressive relative to management’s 2019 FCF number of $20.7 billion. We know management’s normalized 2020 FCF is actually $28 billion which is 35% higher than 2019! Here it is from the 4Q20 earnings release with some red boxes added for key numbers:
Image Source: 4Q20 earnings release
The lens through which I view economic FCF is a little more evolved than what we see with the accounting numbers above. Of course I view the $5 billion FTC settlement as non-recurring, so I add it into the 2020 total. The accounting version of FCF subtracts all capex whether it is maintenance or growth. I try to figure out what is what and add the growth back in such that FCF goes up. This is largely offset by the fact that I treat share-based compensation like a cash expense so it lowers my economic FCF. The key numbers for these considerations are from the cash flow statement in the 2020 10-K and I’ve added red boxes for emphasis:
Image Source: 2020 10-K
For normalized FCF viewed through an economic lens, I start with the $23 billion from the 4Q20 earnings release and add the one-time $5 billion FTC settlement to get $28 billion. The $15.1 billion of capex for 2020 exceeds the $6.9 billion depreciation and amortization by $8.2 billion. I’m guessing that maybe half of this excess is maintenance capex and half is growth capex. As such, I add $4.1 billion to the existing $28 billion to get $32.1 billion. However, I treat the $6.5 billion share-based compensation like a cash expense, so I subtract it bringing my economic FCF total down to $25.6 billion. Given the growth and other considerations, I think it's reasonable to value the company at 30x this amount or a little under $770 billion.
The enterprise value (EV) is the $754.4 billion market cap [1] less $17.6 billion cash and $44.4 billion securities for a total of $692.4 billion. I’m leaving the $9.6 billion long-term operating lease liabilities and the $1 billion short-term operating lease liabilities out of this EV lens because we backed $0.6 billion of principal lease payments out of FCF and we want to keep it apples to apples when comparing the FCF valuation to EV.
[1] The 2020 10-K says they had 2,405,448,410 shares of Class A common stock and 442,221,541 shares of Class B common stock outstanding as of January 22nd, which comes to 2,847,669,951. We multiply this by the March 1st closing price of $264.91 to get the market cap subtotal.
I think the stock is reasonably priced; my valuation of around $770 billion is a little bit above the EV of $692.
This article was written by
Analyst’s Disclosure: I am/we are long FB, GOOG, GOOGL, AMZN, AAPL, VOO. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Seeking Alpha's Disclosure: Past performance is no guarantee of future results. No recommendation or advice is being given as to whether any investment is suitable for a particular investor. Any views or opinions expressed above may not reflect those of Seeking Alpha as a whole. Seeking Alpha is not a licensed securities dealer, broker or US investment adviser or investment bank. Our analysts are third party authors that include both professional investors and individual investors who may not be licensed or certified by any institute or regulatory body.
Recommended For You
Comments (172)
A friend who runs a significant ad budget started to cut back on digital advertising last year and it has worked really well for her company.
However, Google's announcement today that they will stop using individual browsing history to sell ads (even on their ad exchanges) is changing the digital marketing game. Third party cookies are already in the process of being extinguished. Apple and now Google are trying to stay one step ahead of legislation around the world that improves privacy. Facebook will have to adapt too. I am not sure how dynamic ads will work if there is no tracking of the individuals' browsing history across the web.
One word about the user growth. Facebook counts a lot of folks as users who may not be actual users. People who use Facebook to login on different sites are counted even if they are inactive on the Facebook pages. People who do marketing analytics are forced to open Facebook accounts so they can use Facebook analytics. Etc.