Blog

The Top of Funnel Butterfly Effect

This post is based on a Comment I recently posted on Rand Fishkin's Whiteboard Friday video, titled How Can the Value of Top-of-Funnel Channels be Measured?

This really struck a chord with me after seeing Kelvin Newman's presentation at SearchLove 2014 in London.

Trying to predict the ranking, traffic or business outcome effect of certain Top of Funnel actions is like trying to forecast the weather for in 6 months. There seems to be a parallel with Chaos Theory. According to WikipediaSmall differences in initial conditions yield widely diverging outcomes for such dynamical systems, rendering long-term prediction impossible in general.

Let me illustrate this with an example: during a conference you might be queueing for food and randomly start a conversation that could ultimately lead to a massive business deal, a PR opportunity, etc. This might not have happened if you would have started queueing 15 seconds later or at another catering buffet. A small difference in this initial condition can thus create a very different outcome.

With SEO maturing to encapsulate more traditional marketing elements such as branding and PR, this creates a situation in which SEO professionals are still trying to find causality when correlation is probably the best we will ever achieve in this context. I must admit that I also want to dissect and understand processes in an IF THIS THEN THAT manner, but unfortunately this isn't always a feasible approach.

According to Kelvin's analysis, we are going to have an increasingly difficult time trying to reverse engineer (aka understand) Google since it is becoming more of a machine learning company; referring to the acquisition of Deepmind.  Some of the points he was making were truly fascinating:

  • When machines teach themselves; they solve problems in ways we can't understand or reverse engineer.
  • Algorithms can react in peculiar ways when they come into contact with other algorithms, and the real world
  • Even with billions at stake, the world’s greatest scientists cannot predict the outcomes of a complex and chaotic system with precision (such as the weather)

So, what now? Shall we just forget about SEO, change careers, go sit on a mountain, herd sheep while enjoying nature in all its glory? Even though some days this might sound like a good value proposition... all is not lost! 

Even though we might not be able to reverse engineer Google, we are aware of what Google is trying to achieve and there certainly are things that we can do in order to optimise our chances for a positive outcome. In this context it is interesting to see how David Sotimano, from Distilled, used machine learning to try and figure out why some Distilled blog posts performed well (as to organic traffic), while other posts didn't perform well at all. Results were not conclusive but the approach certainly showed an opportunity for the future as to defining the base conditions maybe?  

What is required, but not necessarily sufficient?

How does this all translate in practice?

In a previous life I have done loads of marketing and PR for brands such as Eidos, Vivendi Universal Games, Blizzard, etc. It was always notoriously difficult to predict the outcome of a specific action/activity. But also, the more you get yourself out there and the more initiatives you undertake, the more likely it is for Serendipity to emerge.

Over 2000 years ago the philosopher (and statesman) Seneca managed to encapsulate this extremely well: Luck is what happens when preparation meets opportunity.

Well I guess that in this context we need to achieve the optimal preparation with our Top of Funnel activities. So it is indeed about trying to figure out which actions are most likely to help us achieve "luck".  I refer to this as: Which are the base conditions or activities needed?  SEO Professionals are actually extremely well placed to play an important role in defining the minimum required promotional activities. Being used to large data sets and analysing the hell out of everything, SEOs can very likely help define increasingly large parts of the marketing spectrum within an organisation.

Reverse Engineering a Hit?

I often think that what we are trying to achieve with a piece of content, is what musicians are trying to achieve with their music... a hit = going viral. Is there a sure recipe for creating a musical hit?  Not really, but there probably are some things you really need to do in order to optimise your chances.

So this is probably what we need to define for ourselves? What can we do in order to optimise our chances towards a positive outcome? What is the minimum effective dose of digital marketing that we should be implementing?

Many questions, for which the answers will be different for every company and/or industry.  So in conclusion: analyse, find correlations, understand your minimum required marketing and... don't assume that Serendipity can be taken for granted.