Facebook Books: The Future of Social Reading

Facebook today announced that their next big priority categories are movies, books, and fitness. Never mind that Amazon and Apple realized those were priority categories years ago, and in the case of Amazon, as early as 1998 with their acquisition of IMDB. What struck me about the declaration is not that they are late to the game but rather that their definition of success is so modest.

Success for Facebook in any given category is to open it up to mass content sharing - gaming, news, and music were all cited as success stories and indeed, more content --> more engagement --> more ad revenue. The Facebook flywheel is a well oiled machine by now with likes and recommendations. However, i can't help but feel that it could be so much more...

In the case of books, Facebook seems to have missed an opportunity to innovate with the acquisition of Push Pop Press. Whether that was because of IP issues is unclear, but instead of solely thinking about recommendations on what books to read, it seems that there's an opportunity to optimize the activity of book reading itself and completely redefine the category. In other words, a recommendation engine focused on intra-book comprehension AND inter-book discovery would be much more useful and valuable.

For example, it's well established that people learn in different ways. Further, research shows that our preferred learning style is situationally dependent versus an immutable fact. So, what if the future of books is adaptive where the content is dynamically served, based, in part on the feedback loop we create? Metrics to consider could include;

• Descriptive Metrics - average words read per minute, dwell time per page, highlights per 1,000 words, average session length, dictionary usage, books read per month, etc.
• Interactive Metrics - built into the books could be optional polls and pop-up quizzes that test comprehension, the result of which would influence the content to follow.

Since it's Facebook we're talking about, all of the above metrics could of course be shared back out to each user's news feed, so instead of just knowing that Jason is reading A Scots Quair, I could also share notable quotes from the book, and publish the results of various quizzes I'm taking as I read. We could go one step further and Facebook could create dynamically assembled book clubs where I can chat with other people who are reading the same book as me at the same time. 

The other big opportunity is to view interactive books as a launching-off point to dynamically linked 3rd party content. Continuing with the A Scots Quair example, Facebook knows my hometown is Dundee, Scotland which is near the setting of this book. So, when I highlight the word Dundee in the book, I get a set of personalized search results - maybe an offer to buy Dundee University gear or a link to their Facebook page, whereas someone else highlighting the word who's not from Dundee gets more tourist-oriented content.

Regardless of monetization opportunities, the key point is that we're talking about an optimized reading experience based, in part, on Facebook's social graph. There are undoubtedly many who would argue that reading should remain a solitary activity, but one only need to look at how Twitter has augmented the TV viewing experience to understand that the future of reading is social and dynamic. My bet is that it will be intra-book in addition to the more generally accepted inter-book experience.

The Trouble With Social

When Seattle hip-hop duo, Common Market, penned Trouble Is, they defiantly asked "...this is hustle biz, what the trouble is?" While for them, the "...trouble is love don't want you," in social, the trouble is the hustle has been focused on all the wrong things. Paul Kedrosky said it best in a recent tweet and judging by the number of favorites and retweets, it obviously struck a chord.

Whether the payload is news, apps, games, or search results, the prevailing approach today is to mine someone's friends, apply some simplistic model that presupposes that we are like our friends, and call it good. While it's well established that similarity breeds connection, it's worth remembering that our homophily merely occurs at a higher rate than between dissimilar people. In other words, while our friends are similar to us in many ways, we all have outliers, and it's these outliers that social search has done a really poor job of addressing so far. This is true of not just friends but also interests.

The question though is how hard someone like Paul is willing to work for his content? Most techies believe that curation is the answer and the frustration expressed by Paul is solvable with more data and better algorithms. However, what they tend to discount is that curation is always going to be imperfect because no algorithm can anticipate my need without fail. Hypothetically, if such an algorithm were possible, would a tech company even implement it? If we found what we wanted 100% of the time on Google's first page of results, what would happen to their revenue? Could any increased CPC compensate for the loss of impressions on subsequent results pages? I doubt it. I believe Google's in the business of optimizing, but not too much. Speculation aside, the problem is that we're complex and sometimes unpredictable. Our needs are fluid and we push ourselves to discover the unknown. 

So, if curation nor search are the silver bullets, might social search be? Facebook would have the world believe that we are like our friends because that's the sandbox they currently have to play with and they need brands to believe that. Almost everyone else in the space is following them with their friend based recommendation engines. Bing, anyone? At Harnu, we're taking a different approach by trying to encourage new connections. Want to know what's going on in Syria right now? Talk to a Syrian. Get a local's perspective. I don't know about you, but my friends are not useful for this class of discovery. That's why we built Harnu. 

For us, while we may be taking a novel approach to social discovery, our uphill battle is going to be achieving scale for the foreseeable future. Facebook obviously has the scale. The question is whether they'll take a novel approach to social search or serve up the same old friend recommendations as everyone else. If they expand results to consider people that are not friends and let people connect with these strangers, then folks like Paul Kedrosky may get their wish as a massively long tail of people and content starts to come into play with the trade-off that Paul would have to do a some content sifting himself.

Most likely though, this would have the potential to alienate a lot of Facebook users unless they specifically opted in to that sort of feature. However, it'd also have the potential to dramatically expand Facebook's social graph. In my case, instead of the 460-something friends I have today, I'd set out to add at least one friend from every country in the world - something I'm able to do today on Harnu in the quest for perspective and knowledge that none of my friends is talking about.

As an example, on Harnu, I learned from someone in Belarus last week that it's now illegal there for anyone to gather information without accreditation. Crazy but true...

В последнее время дошло до того, что в стране запрещено незаконно собирать информацию. Только при аккредитации. Вам будет смешно, но это так.

More Camels

Right now, we're in the middle of a complete redesign of Härnu based on early feedback (both implicit and explicit) from our users. So, as the guy on point for bringing users, there's not a whole lot for me to do but plan assiduously and sit tight for the next iteration.

Having said that though, I've been dabbling in posting Twitter and Facebook updates on behalf of Härnu and what I've learned won't surprise anyone - photos are gold! 

A word on virality

Before I get into details, Facebook defines virality as "The percentage of people who have created a story from your page post out of the total number of unique number of people who have seen it." In other words, it's the number of people "Talking about this" divided by the "Reach" as defined by Facebook.

So, with that out the way, here's a quick rundown of what I've learned so far:

Photos Trump Text

One day recently, I posted a Facebook update about a mother using Härnu to find her adopted son's biological brother in Ethiopia. This was a use case we could never have imagined but it was one that we thought was incredible and were really pleased that our efforts could play a small part in helping. This post was all text.

The next day, I posted a quick Facebook update about Indonesian users now on Härnu. I included a stock photo of Bali to highlight my point.

We saw a virality score of 2.41% for the post about the mother and a 4.55% score for the post about Indonesian users. In other words, the post about Indonesian users was 1.9X more viral than the one about the mother looking for her adopted son's brother in Ethiopia.

Hypothesis #1: Posts with embedded links and / or photos are more viral than those without

Easy and obvious enough right? What came next though was a much better illustration of that point. One of our users from the Philippines whom I've been chatting with, sent me a picture of him and a camel (he's living in Saudi Arabia just now). He also told me that people eat camel meat there. In retrospect, not very surprising. Kind of like horse being on the menu in much of France right? I thought this was pretty interesting, so I posted this to our Härnu Facebook page (note I started with an attention grabbing question):

This post had a virality score of 12.1%. Our most successful one yet and 2.65X more viral than the post about Indonesia and 5X more viral than the one about the mother.

Hypothesis #2: Photos that are more personal and authentic are more interesting than stock photos

Hypothesis #3: Posts framed as questions are more viral than simple text updates

Note I'm using the term "hypothesis" here and not yet "conclusion" - we're still learning and a sample of one does not a trend make. That said, I followed up this last post about a camel with another one with a photo, posed as a question, and tried to piggyback on the Queen's Jubilee in England.

How viral was this one? The answer is not at all. Turns out no-one cares about the Queen, or at least, none of our users does.

Conclusion: While we have yet to achieve Konyesque virality, in the meantime, our winning formula appears to be MORE CAMELS ;-)

For the experts out there, what kind of content has worked best for you on Facebook & Twitter updates? What kind of virality scores do you typically see? Do you think it even matters?

Intent is King

When Härnu's Facebook page reached 100 Likes, Facebook offered us a $50 credit to try out Facebook ads. Now that we've dipped our toes in the water, I thought I'd share some thoughts on the experience.

Wanting to get a really firm idea of not just clicks of the ad on Facebook but actual conversion as measured by new Härnu user sign-ups, I started off with a really niche ad - For the image, I selected our logo, wrote some text in English and then targeted it to users in Norway and Sweden. I layered in some demographic constraints such as wanting to target people between 18 - 35 but that was about it. Oh, instead of having users click thru to our Facebook page, I set it up to have the ad click thru to Härnu's home page.

Well, after running it for about a week and paying $0.67 per click we didn't observe any incremental user sign-ups attributable to the ad. Next, I thought maybe I should try writing the ad in Swedish to make it feel more localized. So, I did and this time restricted the ad to just Sweden, dropping Norway (sorry Norway!). Well, in the past month, I've seen 50K impressions, 18 clicks, a clickthru rate of 0.035%, and zero incremental users attributable to the tactic.

Thinking of ways to spruce up the ad, I replaced the logo image with a screenshot, I re-wrote the ad in English trying to be more direct and impactful, and this time I targeted it to NZ, Australia, and Canada. The thinking there was that maybe it was a cultural thing and English speaking countries might be more apt to try the service out. Well, this time I've observed a click-thru rate of 0.030% and zero incremental users attributable to the tactic.

Conclusion - Facebook ads are not working for us.

I'm sure there are ways we can make this better, and for a new entity, it doesn't help having so little name recognition. However, while I'm certainly not a creative expert, we could see a 10X improvement and it'd still only be a 0.3% click-thru. I've read that average click-thru on Facebook is somewhere around 0.04% - 0.05%. Contrast that with Google AdWords average of ~2% and I'm left scratching my head asking what am I missing?

Just this week, GM decided to cut bait with paid advertising on Facebook and I have to say I can't blame them. For a while now, we've seen Sheryl Sandberg tout the virtue of brands being a part of a user's life on Facebook, while shying away from metrics like CTR, never mind conversion metrics such as sign ups or sales lift:

"Facebook executives argue that the click-through numbers are not that meaningful; they say that people remember ads better and are more likely to make purchases when their friends endorse products." - Business Week, May 2011

If Facebook, with all of the mountains of data at its disposal can't hang its hat on hard metrics, and instead has to rely on some qualitative intercept study talking about recall, then they have a huge problem.

Frankly, I've been one of those sales guys representing ad inventory (not at Facebook) that every metric suggests is not working and it's your job to torture the data and come up with a story to get brands to buy. When it's a new venture, it's easy to make claims that this is a whole new class of ad inventory, and for a while, you feel like you've convinced the brands, but eventually every publisher has its day of reckoning and if the ads aren't working, the media agencies and brands eventually figure that out.

You might still be the smartest guy in the room, but at some point it just doesn't matter anymore. Large amounts of money are being spent and brands want results. You can talk brand equity, you can talk reach, you can even put some new spin on it around building fans but it eventually has to come back to sales, and that's where Facebook is stumbling pretty hard right now in my opinion.

When I am in product research mode, I go to Google or Bing - not Facebook. I triangulate my options using search engines and may subsequently check with one or more friends if it's a big ticket item. However, Facebook is now surfacing those friend recommendations off-Facebook via Bing! In other words, what Google likes to call the Zero Moment of Truth simply never happens on Facebook.

If Facebook's future contemplates the bulk of its revenue continuing to come from ads, surely it must wrest control of those ad dollars away from Google, and it's going to have to find a way to pivot its site to capture those moments of intent.

As brands shift their perception of Facebook from newfangled thing that they're are willing to experiment with to a mature part of the ad spending eco-system, no amount of squishy story telling by its ad execs, even if (s)he's a modern-day version of Hans Christian Andersen,  is going to convince major brands that Facebook deserves to be part of its marketing mix unless it can demonstrate that those dollars are resulting in incremental behavior.

Right now, this little advertiser is not convinced. Losing GM is one thing. Losing the long-tail of advertisers would be quite another...

The Value of Ephemeral Connections

The big news this week is Facebook's $100B+ IPO. Some deride it as a passing fad - not much more than a glorified address book. Others see its Social Graph as the next great monopoly - a dynastic operating system of a social generation. Whatever the case, 900M users gives Facebook almost limitless options and a better than 50/50 chance of succeeding in anything it decides to pursue.

Indeed, with the recent tie up of Bing + Facebook, we're now seeing the first credible social search product. Of course, Google's been at it for a while with Search Plus Your World but it doesn't have the scale of 900M Facebook users behind it.

For example, now when I go to Bing and search for Maui hotels, I now see 3 things: 1) Organic search results - in this case mauihotels.com shows up first; 2) Paid search - Ritz Carlton shows up top; 3) The option to ask my friends on Facebook, or alternatively, where there's a match, Bing suggests friends who may know based on Facebook profile & timeline info.

All of that taken together is much more powerful than simply a paid search result from Ritz Carlton for example, right? However, the success of Facebook enabled social search hinges on two things 1) Our continued sharing of every aspect of our life and 2) The assumption that we are like our friends.

I'm not about to argue with Harvard researchers on the first assumption. However, the second assumption suggests room for further opportunity. That is, while I of course have things in common with my friends, there also many differences - enough to suggest that simply knowing who my friends are is not always enough to infer what I'm likely to consider relevant.

So, if enduring relationships can't always guarantee relevant content, might there be value in ephemeral connections? After all, it happens all the time in the offline world - random encounters at restaurants, in bars, at the movies, in the supermarket, at the park with kids etc. result in conversations with strangers that influence whether or not you buy a product, watch that movie, or go to that bar. These are not friends in the classic sense. Rather, they're people in passing that you struck up a conversation with and trusted enough to listen to what they had to say and valued their opinion, but you're not necessarily likely to add them to your Facebook friend list any time soon.

With that in mind, social search will become much more powerful when it progresses beyond simply mining my friends' likes and comments to actually putting me in touch with potential trusted sources wherever they reside be they friend or stranger. Add in a layer of anonymity so I don't need to become Facebook friends with every ephemeral connection, and now we have a product that truly taps into the potential of the Social Graph.

That vision is very much at the forefront of where we're taking Härnu - a recognition that sometimes seeking out a stranger's opinion or getting a local's perspective can be much more valuable than those available from existing friends.