Creating a 30 second YouTube movie that goes viral is the holy grail of marketing. So how is it done?
Ensuring the success of a viral-produced movie is still largely hit-and-miss. Some of the more well-known ingredients that boost the chances of viral success – such as babies, pranks and stunts – seem to have great success on some occasions, but turn into catastrophic failures on others.
There are many little tricks that can help-kick start a viral-produced movie, but these only work if the movie already has the potential to spread on its own. Even if a formula for success is chanced upon, the issue of how to attach a brand to the movie remains.
As soon as people sense the movie is actually an advertisement, the virility of the movie typically gets cut short.
An algorithm developed by my colleagues and I (though adolescent in development) is the first step towards identifying a production framework that ensures the success of viral movies.
Viral transfer of information through networks is essentially word-of-mouth buzz on steroids. Since the emergence of the internet as a social communications tool, word-of-mouth communications has taken centre stage.
In a study I reported in 2010, we found strong evidence suggesting the effects of traditional advertising are weakening. Increasingly, consumers rely on the opinions of others online rather than what corporations are telling them.
The key to understanding how viral messages spread is understanding what motivates people to transfer information to others, and what makes others act on that information.
Richard Dawkins, the British author and evolutionary biologist, drew similarities between packets of information that spread down through generations and Darwin’s theory of evolution. He named these packets of information “memes”, after the biological concept of genes.
According to Dawkins, a packet of information needs to be “fit” in order to transfer or replicate. In the same way genes need to be fit to survive extinction, so too do memes need to be fit in order to be transferred.
Survival therefore rests on the degree of fitness. Figuring out what makes a packet of information “fit” is the basis of developing an algorithm that can explain and predict messages spreading virally.
Four elements need to be in place for a branded movie to become viral: congruency, emotive strength, network-involvement ratio, and paired meme synergy. These four are the basis of the branded viral movie predictor (BVMP) algorithm, as explained below.
- Congruency. This refers to the consistency of the BVMP theme with brand knowledge. Judgments about brands are shaped to a large degree by past knowledge. Our minds are full of a series of associations that are attached to a brand, and our feelings towards a brand are shaped by these associations.
Harley Davidson for most people is associated with freedom, muscle, tattoos and membership. Our attitude towards these feelings shapes our attitude towards the brand. If we place a high value on freedom, muscle, tattoos and membership, then likely we will form high value towards the Harley Davidson brand.
As soon as we witness associations with the brand that are inconsistent with our brand knowledge, we feel tension. A Harley Davidson scooter designed for inner city commuters would be incongruent with the brand knowledge of many people.
The BVMP therefore remains weak and exposed to extinction.
- Emotive strength. We process thousands of packets of information every day. The weaker ones (most) earn a single thought in short-term memory, and then become extinct almost instantaneously.
Our minds couldn’t cope with elaborate processing of all information in a day; if we did then likely we would go insane.
A BVMP needs to evoke a stronger response than these other packets of information to survive. Stronger responses are tied to emotion – information that evokes a strong emotion sinks into long-term memory and benefits from multiple episodes of processing. Emotion may come in many forms, in different combinations and in contrasting strengths.
Disgust and fear are powerful, and are relatively immune to extinction. Sentiment can be equally as powerful, but is more dependent on the network-involvement ratio. Humour and happiness are weaker, and tend to turn reliance on to the other three elements for survival.
- Network-involvement ratio. Another way of describing this element is in terms of how relevant the message is to the seeded network. Transferring packets of information is not enough if the receiver is not motivated to accept the information.
In turn the receiver must then transmit. The internet has taught us that acceptance of a message is necessary, but can exist at any transmittal node without transmission. In other words, there only needs to be one point of transmission (e.g. YouTube), but every node (person) must process (view) the information.
The BVMP needs to be relevant to most of the nodes in a network in order for it to spread, and the network needs to be large enough for competing network nodes to also process the information.
University students are a large network of nodes sharing similar sparks of involvement. Thus, a movie of a student spray painting a message on the whiteboard during class will most likely spread quite efficiently.
The office-workers-who-have-a-degree network is fairly close on several dimensions to the university student network, so node transfer to this new network is also likely to be efficient. The degree of BVMP theme involvement within the networks is high in comparison to the network size.
- Paired meme element synergy. The three elements above are necessary but insufficient for BVMP success (see the Paired Meme elements table below). In our analysis of successful BVMP we noted certain patterns of memes that only appeared effective when paired with certain other memes.
“Impromptu entertainment acts” appeared to work when paired with “Eyes Surprise” (unexpectedness causing the eyes to widen). When paired with ‘"bubblegum nostalgia" (evoking familiar historical sentiment), the BVMP doesn’t work. “Anticipation” (dramatic build – what will happen?) works with “Voyeur” (captured/recorded by chance, e.g. on an iphone), but not on its own, and so forth.
Of course it’s entirely possible more than one pair could be combined; we have only begun to look at BVMP success when meme elements synergise in pairs.
Our algorithm comprises four elements, but is fine-tuned to predict word-of-mouth only in the context of website experience. Modifying the algorithm to predict word-of-mouth more generally would become extremely complex.