With over 1.2 billion reported users and close to $200B in market capitalization, Facebook is undoubtedly the most ubiquitous social network today. For most users, the core value proposition of Facebook is simple – it is a means to stay connected with their friends (and acquaintances) and to share and learn about each others’ lives. And yet, over the years and over countless tweaks to Facebook’s NewsFeed algorithm (popularly known as EdgeRank), more and more users complain that they don’t get to see any updates from a majority of their friends. Indeed, the average user has over 300 ‘friends’ on Facebook, but thanks to Facebook’s determination of what’s relevant, they are likely seeing updates from only 20% (or less) of their network. What’s going on? Why is it that I have over 1200 ‘friends’ on Facebook, yet I never see anything from almost a 1000 of those? I used to believe they simply didn’t post as much, until I checked out several people’s profiles and saw major updates I would have liked to see, but never saw, despite logging in several times a day. Why is it that I see some stories over and over for days, and several never appear?

Keep it simple….

Several hours of tweaking Facebook settings, privacy controls and reading Facebook optimization controls told me one thing – it’s complicated by design. There is a lot on Facebook that’s simple and intuitive, but customizing your experience is definitely not. There is an option to sort your feed by ‘Most Recent’ but all it does is sort the pre-selected ‘Top Stories’ into reverse chronological order of any action taken by anyone, thus being not helpful at all as it doesn’t introduce new content and in fact increases repetition. You can unfollow or block users, you can tweak content settings for people and types of content individually, or you can organize your 1200 friends in lists you then follow (like really?). For the average user, it is too much to ask, but I’d venture to say that even for power users, it doesn’t really help much.

They have the edge

EdgeRank works in mysterious ways, and the best one can gather is that Facebook measures and ranks ‘edges’ connecting any one user to another user (or Page, Group, Brand etc) by the strength, time delay and frequency of their interaction. However, only active interactions count, i.e. liking, commenting, following or sharing. So if you passively enjoy reading someone’s updates but don’t actively ‘like’ them, chances are you’d stop seeing updates from them sooner than later. This is true for both your friends as well as pages you may have liked, unless of course they pay Facebook to promote the post. The problem arises when over time you see what you like becomes you like what you see, making your Newsfeed populated by the same subset of users and content types and effectively limiting the reach of content. And lest you figure it out, they tweak (and AB test) EdgeRank all the time. So you may not even realize that the reason some of your real world friends don’t comment on your exciting Facebook updates may be that they actually never got to see it, for no lack of intent whatsoever.

“Trust us, we know what you want to see”

Let’s face it, Facebook does know a lot more about us than we think. As long as you’re signed in, Facebook knows not just what you ‘like’ and who you stalk on their website, but also most likely what articles you’re reading and what websites you’re surfing for how long. Besides, information overload is a real problem. Between friends’ updates, activities, engagement content and brands, Facebook estimates they have thousands of news stories to show every user at any point. Surely some stories are better or more important than the other for every user. But by Facebook’s own estimate, only 0.2% of these stories are ever shown to the user. With no easy way to even access the remaining 99.8% and no straightforward explanation of how those 0.2% are determined, it is unsurprising that I see check-ins every time my dorm neighbour gets down to eat and I totally missed the news of wedding and first child of my high school best friend. And these were happy stories – considering Facebook doesn’t want users to not see many ‘negative’ emotion stories, I wonder what all I’ve missed that would have been relevant to know. Or not.

It’s all about the money, honey

All this brings me to the business of Facebook. It is not so hard to gather that the purpose of ‘optimizing’ your NewsFeed is as much to show you the most relevant updates from your friends as it is to show you ‘relevant’ sponsored stories by those that pay Facebook by creating real estate. Facebook marketing is, after all, a fast growing and rather effective (for now) channel for most brands’ marketing efforts these days. One can argue that, after all, it is a free service that Facebook is providing to the users and they deserve being compensated in some way for it by selling part of the user engagement it creates to the brands who want them. And these are brands the users want too, demonstrated if not explicitly by subscription then implicitly based on their behavior as Facebook understands. Perhaps the users shouldn’t complain so much, after all. Sure, they don’t get a perfect experience and sure, there are a few ethical questions because users don’t really understand how they are being manipulated. But what about the brands themselves?

Thousands of advertisers have spent precious time and money over the years building up reach on Facebook pages, but sometime last year they realized that all of a sudden their messages weren’t being shown to all the users who had ‘Liked’ and previously engaged with their page, never mind to new users. So unless they pay for each posting, or the user is a dedicated follower who actively engages with every piece of content posted since the beginning of the change, Facebook’s reach for most brands is basically a myth and the promise of building an engaged community with two-way communication hollow. I wonder how sustainable this is, in the long run, especially as Wall Street maintains earnings pressure on Facebook and non-advertising revenue on the website continues to slip.

Bottomline, friends are not really friends on Facebook. Fans are not really fans. Don’t like the Likes too much.


read more

“Mood” is defined as “a pervasive and sustained emotion that colors the perception of the world. Common examples of mood include depression, elation, anger, and anxiety. In contrast to affect, which refers to more fluctuating changes in emotional “weather,” mood refers to a more pervasive and sustained emotional “climate.”[1] While “affect” is an external and observable expression of emotion, “mood” is internal.

Assessing another’s person’s mood, even when physically present with them, can be very difficult. When in person, one can resort to verbal and non-verbal communication to make this type of assessment. One might simply ask someone questions such as “how are you?” or “is something wrong?”, but we all know this rarely produces a ‘verbal answer’ which actually matches a person’s mood. When it comes to non-verbal communication, one might try to draw conclusions about another person’s mood based on observations of his or her affect.

However, even when one can assess someone’s affect – through tone, gestures, and general demeanor – this may be incongruous with their mood. To complicate matters further, what’s considered an appropriate level of affect to display to the exterior world varies across cultures, situations, and personalities. With it already being so challenging to evaluate another person’s mood correctly in person, with direct access to the physical cues which make up ~80% of communication, think about the challenges of evaluating someone’s mood online.

Well, this is precisely what Apple claims it wants to do – assess your mood, for purposes of ‘mood based advertising’ – in the ”inferring user mood based on user and group characteristic data” patent application (No. 13/556023) it filed this past January. Online advertisers already use a host of contextual factors – location, age, time of day, types of searches and purchases, and general browsing history – to target individuals, and knowing someone’s mood would add yet another powerful dimension to their arsenal. No one doubts how influential mood is in the way a person processes an ad, and how mood can impact purchasing behavior.

What might this form of advertising look like, though, you may be asking? Imagine if Apple could correctly assess in real-time whether you’re happy or sad – a brand such as Coca-Cola which wants to reinforce psychological association with happiness may only choose to show you ads when you’re happy, while a shoe brand may choose to show a certain ad to a lonely, sad young woman of a certain income category, who may be more susceptible to the ad at that time and engage in some impulsive retail therapy.

How does Apple intend to accurately measure something as intangible as mood, though?

According to its patent application, Apple’s system would collect and analyze a combination of physical, behavioral, and spatial/temporal data over a period of time to build a “baseline mood.”  This “baseline mood” will be used to assess ongoing data collected, to infer real-time moods by comparing against your profile using “mood rules” set along these dimensions. Behavioral data might include engagement with social media (what are you posting, looking at, and how often, for instance), online browsing, and engagement with apps (what and in what sequence), paired with age, gender, and spatial/temporal data such as location and time.

Physical data, and this is where it gets quite interesting, could include the likes of your heart rate, blood pressure, body temperature, or vocal expressions. Below are some of the diagrams from Apple’s patent application, giving a high level overview of their data process.

OE blog post image 1

OE blog post image 2

OE blog post image 3

What does it mean for advertisers?

If you think about traditional media, many companies held the belief that people in a good mood would respond better to advertising. Some psychological research suggests that in fact, people who are feeling low may be most vulnerable to advertising. With access to online advertising that can incorporate ‘mood’ as a metric, advertisers could potentially have more runway to test this hypothesis, considering the lower testing costs of online advertising versus TV advertising, for instance. Regardless of this type of testing, however, advertisers will jump on this type of data to further refine their desired targeting strategies – whether it is effective or not, harmful or not, remains to be seen.

What does it mean for consumers?

With privacy already being invaded in countless ways, who exactly will welcome this with open arms?

What data should be available to advertisers? How nervous will people be about Apple keeping this kind of data safe? What if it gets into the wrong hands? What happens if Big Brother knows my mood, location, age, physical characteristics, and interests – all in real time?

The challenge of recognizing ‘mood’ is something medical professionals have not yet untangled to their satisfaction. This begs the question – would something so potentially innovative not serve the world better in the realm of medicine rather than in advertising?? Maybe I’m in the “mood not to see ads” – what then? What is known of the potential effect such advertising could have on my mood? What will be done to protect against misuse, or even understand the scope of what that means?

The use of biometrics, were mood-based advertising become a reality, also makes me think differently about the iWatch.  Initially, I wondered what was so new about this product – how will it serve customers, who may already have an iPhone, in a truly new way? Is there enough to make someone go out and purchase the watch, in addition to their phone? Now, I see this is a first method for Apple to build its capability in collecting biometric data. It is indeed going to serve them quite differently, compared to any of their other products.

Final thoughts

How will Google respond to this? Some say they are venturing into the realm of incorporating data on our behavior in the physical world through acquisitions such as Nest, but have they done anything in the way of detecting and incorporating mood? Which one makes people more nervous, and is taking matters ‘one step further’? Other players such as Microsoft filed for their own patents for similar ‘mood-based advertising’ systems, which would rely on data collected online, as well as data collected through the Kinect sensor.

While there is no ‘real-time mood based advertising’ currently in use, and its use may be quite far off in time, the tale of The Apple and the Moodreader provides some important food for thought.

Sources:

USPTO - http://www.google.com/patents/US20140025620

Business Insider - http://www.businessinsider.com/apples-mood-based-ad-targeting-patent-2014-1

The American Psychiatric Association - http://bit.ly/1zoDMfk

Apple Insider - http://appleinsider.com/articles/14/01/23/apple-investigating-mood-based-ad-delivery-system

GeekWire - http://www.geekwire.com/2012/happy-sad-microsoft-system-target-ads-based-emotional-state/


[1] Source: The American Psychiatric Association


read more

Although some may argue there are several more, I see the evolution of advertising being characterized by movement across two axes:

  • CHANNEL: Word-of-mouth –> Print –> Radio –> Telephone –> Television –> Google (no, the internet). Yes there are some others that don’t fit perfectly into this flow
  • OBJECTIVE: Customer Acquisition –> Lead Generation –> Awareness –> Branding –> Differentiation –>(what I can only call emotional exploitation because there’s no other way to explain some of the Google ads that don’t really talk about Google)

Now comes this new form of advertising called “native” that sits in a new dimension altogether – it can be used across any channel and accomplish a number of objectives. WHAT IS IT? According to a number of sources that I pieced together (adWeek, eMarketer, comScore, Mashable, and my personal favorite for the hard-hitting facts, Wikipedia), Native Advertising “matches the natural form and function of the environment in which they live.” Form = feels like natural content. Function = behaves like the native experience.

(Funny fact, a lot of brands admit to not knowing exactly what it is)

eMarketer_Ads That US Business Pros Consider Native

 Why Will it Be Big?

The facts: (1) Print is dead…we won’t even go into the detail although some argue it’s still alive…mainly print publishers. (2) Overall display click-through rate (CTR) is down to 0.1%. (3) Overall search advertising CTR is declining YOY even though CPC continues to rise. Net/Net – print advertising is dead (worth reiterating) and digital advertising is dying. Why? Because users have become accustomed to ignoring banner and search advertisements. The future is native advertising – ads that do not even appear to be advertisements from the user’s perspective. I’m no expert, but the expert’s agree that native advertising growth will dramatically outpace display (and other) advertising:

US Digital vs  Native Social Advertising

Still Significant Room for Improvement

Recall the two defining characteristics of native advertisements: (1) it matches the form of the medium and (2) it matches the function of the content. Pop-quiz: do these ads meet that criteria?

 Form = maybe, although I argue that the “sponsored content” feels pretty unnatural. Function = probably: since I’m reading an article like I would any other)

 Now think ALL the way back to the objective-evolution of advertising. The one common denominator of all steps in the chain was that each type of advertisement, served through any channel, was meant to translate into incremental sales. Now, regardless of form and function, do these “native” advertisements accomplish that goal? NO. Who cares that the Atlantic article was sponsored by IBM. If you read the whole article, you would see that the “sponsored” area at the top is the only place that IBM is mentioned. Yes, it talks about Big Data, which is near and dear to IBM’s heart (if they have one), but for me, a data-point of 1, after reading that article I am not left thinking about IBM.

 So What?

 “native advertising” is still not NATIVE ADVERTISING, and there are opportunities to get in while it’s HOT. Current examples may match form and function, but there are very few examples that actually accomplish the goal of an advertisement – to sell or create awareness that will eventually sell.

I believe that the next decade will bring about a number of platforms that were incepted and built to cater to native advertisements for a very specific form and function of content. To be successful a native ad will not only fit the two traditional criteria for native-ness, it will also actually drive sales. Because we are increasingly using different platforms for every possible life “use case,” to that criteria, a native ad must be created for one and only one platform to be successful. And since native ads cannot be transferred across platforms (at least those with any difference in form or function) and still be effective, platform will eventually begin to cater to the native ad.

References:

  • eMarketer
  • Hub Spot
  • comScore
  • Mashable
  • adWeek


read more

When I heard the news that Snapchat rejected Facebook’s all-cash $3-billion dollar takeover, I had a lot of questions. First of all, why did Snapchat reject such an offer given its non-existent business model? Secondly, what does Facebook see in Snapchat? How can Snapchat add value to Facebook’s business?

On first glance, it could seem like Snapchat is merely a fad. Users can send pictures or videos to their friends and decide the time frame within which their friends could view the sent message. Following this logic, it seems as if Snapchat should be able to learn a lot about their users based on the pictures that users have uploaded. However, is this really scalable? Unlike words, it is a lot more difficult to categorize and collect user information based on pictures in Snapchat’s current format. Also, Snapchat allegedly deletes messages that have already been viewed by the recipient, which makes information gathering even more challenging.

I think Snapchat has three main value propositions to its users:

  • Users can select their audience – Unlike Instagram or Facebook where an uploaded photo is visible to all your friends or friend categories, Snapchat allows its users to select who the picture goes to. This allows users to communicate in a more focused manner.
  • Users can communicate in the medium that most accurately depicts their message – Every message is a story and stories can be told differently depending on the content. Snapchat allows its users to express their story through art, text and media.
  • Time commitment required is minimal – Snapchat is accessible through a mobile device and each message lasts 10 seconds at most. Time to access messages is limited to 10 seconds, and time to craft a message is also shortened because users know that this is short-lived.

Although these are all valid propositions, it is still difficult to justify why it should be worth $3 billion because none of these value propositions are revenue generating. A closer look into Facebook’s user base shows that 78% of them are mobile users, which means that they directly compete with Snapchat for attention. Furthermore, Snapchat surpasses Facebook’s photo uploads by 50 million images on a daily basis. As Snapchat takes user time away from Facebook more and more, Facebook’s value proposition to its advertisers is starting to erode. Is this worth $3 billion dollars? To Facebook, it probably does – especially so since Facebook’s IPO has faced a lot of investor scrutiny with regards to whether they can provide a reasonable shareholder return.

Whether Snapchat is eternal or ephemeral really depends on how well it can sustain its user base. From a features stand point, there is definitely a lot that Snapchat can do to constantly keep its users hooked.

In the meantime, Snapchat could benefit from cashing in on its large user base. One way to do so is by requiring users to login using their Facebook account. That way, Snapchat could access more detailed user information and target users more accurately. This would be a good value proposition to companies who are looking to advertise. On the other hand, user disruption could be limited by only showing ads after receiving 10-15 Snapchats.


read more

The interesting world of affiliates

I want to touch on the topic of affiliates again as I think they play such an integral and interesting role in the online economy. While we learned one side of affiliate marketing in class, I’d also like to discuss the affiliate world I encountered back in 2008, on the Facebook Ads team.

When I first started at Facebook, Facebook had just launched its very first monetization product, Facebook Ads. It was a new platform, used a simple interface, and while it had a limited number of targeting options, it was already using the unique user data that Facebook has today around people’s interests (tv shows, movies, books, etc.) Nevertheless, advertising was a brand new idea at Facebook and to Facebook users. Advertisers were still glued to Google AdWords along with other popular online ad platforms. Facebook needed to break through the community and establish its foothold in the world of ads. So, what did Facebook do? Facebook went after the affiliate community.

Going after the affiliate community

Why did Facebook do this? Affiliates often times ran low quality ads. The Facebook Ads system was flooded with bad ads from weight loss, to the new acai diet, to singles ads with inappropriate photos of women’s chests. These bad ads were all affiliate ads. To increase quality, Facebook could have gotten rid of these ads completely and just went after established businesses and brands. However, they would have missed out on a lot of opportunities that the affiliate marketers were bringing to the Facebook Ads platform.

First, it wasn’t that Facebook didn’t want to go after established businesses. Those relationships took time. Nobody trusted social media at that time, and people still don’t even understand it today so you can imagine the push back and convincing that needed to be done. However, affiliates were swarming to Facebook. And here is where the interesting world of affiliates comes in.

Affiliates, while they can be “dirty” and “sketchy”, they are SMART. They know online advertising like a dog knows how to bark. They understand clicks, impressions, CTR, optimization, etc. They know different ad platforms and the intricacies of how they operate. In addition, affiliates are in a volume business. They upload hundreds and sometimes thousands of ads at a time. This kind of ad volume allows them to learn quickly and iterate. In particular, they are always on the lookout for new ad platforms. Why? New ad platforms offer new and more inventory, which allows them to make money. So, when Facebook Ads came along, they jumped on the opportunity.

Now the ads they put on Facebook, for the most part, were not high quality. This isn’t to say that all affiliates use misleading ads. On the contrary, there are a lot of very high quality affiliates out there that choose to put up higher quality and more relevant ads. However, those ones aren’t always the first movers. It’s the sly guys that get in first to test out the waters and push the boundaries. As a result, Facebook was flooded with bad ads. And instead of blocking them from the site entirely, Facebook chose to work with them.

Facebook’s affiliate relationships

Facebook’s Ads team quickly established relationships with affiliates. I was on that original team that did this, and as much as I didn’t love working with them, I found their world fascinating. Some of the most successful affiliates using Facebook Ads were 20 years old or younger! They were college students making money on the side and paying their way through school. Some of them were older and experienced in affiliate marketing – those guys had an affiliate following. Some were dedicated parents. Some threw lavish parties in Vegas penthouses with the cash they were making.

The best thing for Facebook though was that in partnering with affiliates, Facebook was able to build a robust ads system that laid the foundation for its advertising model today. The sheer volume of ads coming from affiliates on an hourly basis was enough to test the ads system’s durability. Engineers could use the data to build an ads system that could handle that level of ad volume. On the operations side, it allowed for Facebook to build a more efficient and automated ad review system. From a legal standpoint, Facebook was able to test advertising policies and iterate on them to pave the way for the high quality ads from more recognizable brands. And, the amount of feedback Facebook was able to receive from affiliates about the pros and cons of uploading ads, using the ad creation flow, using the online ads tools, and reading through ads policies allowed Facebook to fine tune its product.

Today, those kinds of affiliate marketers are no longer on Facebook because over time Facebook’s ad quality rose with more brands coming on board and Facebook going after a broader range of ad types. However, we must never forget that these original affiliates were the ones that actually helped build the Facebook Ads system. They were the volume drivers pumping money into Facebook, making money for themselves, and ultimately giving Facebook the ad volume it needed to scale to what it is today.


read more