'Back to the Future Day', the actual day that Marty McFly showed up in the future in the franchise's second film, is October 21st, 2015. If you found yourself a little guarded or skeptical, it's with good reason. False, doctored images of the dashboard of Marty's time machine showing previous, earlier dates have been circulating, and occasionally going viral for several years now. There's actually a cottage industry that's sprung up around it, including a debunking website: http://istodaythedaymartymcflyarriveswhenhetravelstothefuture.com/ and its companion 'bunking' tumblr blog, http://martymcflyinthefuture.tumblr.com/ which, in an impressive feat of internet trolldom, exists only to post, every day for the past several years, a nearly identical photoshopped image of the time machine purporting each particular day is "Future Day".
The false "Future Day" meme, beyond just providing an opportunity to rib your gullible friends, provides a unique and fascinating case study for those of us that care about what gets shared online and why. The meme itself contains a perfect storm of shareable elements that are worth digging into, but well beyond that, it gives us what almost nothing else on the internet can, a very clean baseline on why something goes viral. That Tumblr blog is the primary ground zero for this meme, and it posts exactly the same thing, through the same channels, every single day for several years, but it receives incredibly varying results, some days getting less than ten shares and likes, while gaining thousands on many other days, and topping out at nearly four hundred thousand on its best day. I've crunched the numbers for every day for two years to find out why.
Why it gets shared.
Studying things that get shared, there are a few reliable characteristics that this has in spades.
- Nostalgia: A great driver, particularly on networks of people we know, like Facebook. When we share it, we're talking about a common experience we've had with the people we shared it with.
- Information driven. It's a small soundbite of definitive, consumable information. The internet doesn't share that hand orientation generally correlates to having freckles, it shares that lefties have 40% more freckles.
- Image driven. Pictures work. Assume everyone on the internet reads everything on their phone held 2 feet away while walking. Pictures take up more real estate as people scroll, grab attention, and tell a quick story.
- Time bound: On par with nostalgia for most indicative, something that is topical makes it more interesting, and gives me a reason to take action.
Virality and Variance:
There was huge variance between how well that same daily Tumblr image performed, but with some clear patterns and correlations. It averaged 1,4333 'notes' a day, a Tumblr classification that counts both shares (reblogs) and likes. The average is largely swayed by a few heavy outliers though, and the median amount of notes it received per day was 237. Across 677 days, the distribution was pretty bell curved, with a few giant spikes. 10.6% of the time (72 days), the post received more than 1,000 notes, and 10% of the time (68 days) it received less than ten. It never received zero. 4.3% of the days it generated more than 3,000 notes, and about 1% of the time it generated more than 10,000 notes, with two clear outliers of 122,282 and 368,200 on May 3rd of this year.
What caused the differences.
There's a tremendous amount we can learn about the differences between the days it did well, and the days it did not.
- Day of week didn't matter much. Don't look at average notes per day of week, look at median. That rounds out the few outlier spikes, and those numbers become much more evenly distributed, with every day of the week falling within a 40 note range of the overall 237 median.
- Base of Repeat Posters: There's a small, retained base of frequent repeat sharers. They weren't responsible for any huge spikes, but they kept the steady baseline on average days.
- Breakouts feed more breakouts: Days over 3,000 tended to cluster, with a 3.2X likelihood for a breakout day to be proceeded by another day over 3,000. A likely explanation is that those exposed to the joke then perpetrated it. The site or image doesn't provide the punchline, your friends do, and the reveal is innocuous and pleasant enough that you want to try to put the shoe on the other foot.
- Depth of engagement: One of the key correlators to virality was shares as a percentage of initial engagement. The best performing posts had 68% of their initial engagement as shares and comments. For the worst performing posts, this number dropped to 58%. Depth of engagement (as measured by ratio of comments and shares vs likes) is absolutely critical for spread. Usually, depth of engagement is a good indicator of content quality, but sometimes the nature of particular content just lends better to a share or a like. If aiming for virality, create content built to share
- Broad and linear first level sharing: Looking into the first 100 shares per post, the median first level shares generated .85 reshares. If you were to visualize it as a tree, it looks a lot more like a linear stick than a branching limb. Don't ignore them though, these casual users, sharing amongst their friends accounted for 60% of the entire circulation.
- The role of influencers: 10% of the sharers created 40% of the second level engagements directly, but they're also the critical connective tissue that bring the meme to the vast majority of the smaller nodes. They're not a few giant lighthouses shining a beacon, they're more a few hundred hanging lanterns illuminating the path. On average, the top 10% cause 31 second level engagements per share, which then generate many stick' shaped linear share cascades. 31 is an above average amount of reactions for these influencers, who typically averaged 11 reactions on their three preceding shares.
- Huge channels: No post exceeded 10,000 notes without the assistance of an existing huge channel, such as a celebrity or large publisher resharing, but nearly 80% of posts above 3,000 notes did so without a huge channel getting involved. No huge channel reblogged directly from the original blog itself, they were all reached through a sharer with much more moderate reach. The right huge channel can be great for achieving an outlier, but they're not necessary for large reach, or a dependable part of everyday success, and they still depend on a broad network of average sharers around them.
What will happen on 'Future Day'? Will we see the first clean death of an enduring meme, or will we go through the wormhole and this meme will somehow find a way to march on? We watch together on the brink, and hopefully the lessons we're able to learn from this unique opportunity serve as some cold comfort when the date rolls over and the world cruelly reminds you that you still don't have a hoverboard.