What can marketers learn from the Church of the Flying Spaghetti Monster?

Correlation vs causation in an email

Did you know… that global warming is caused by a decrease in the number of pirates in the ocean? That’s what the Church of the Flying Spaghetti Monster believe!

Just look at the ‘proof’ – as the number of pirates decreased, the global average temperature has increased!

This is a silly example but the same misattributions happens in many areas of culture, science and certainly in marketing. With the availability of big data, the abundance of reporting and analysis tools, increasing time pressure and increasing importance of proving ROI; it can be easy to look at results and jump to conclusions. Seeing a correlation and assuming a causation is one of the most common examples of this.

Recognising correlations

We’re presented with correlations all the time, to the extent that people often want us to assume that one factor caused a particular result.

Take this example from the Daily Mail – a blatant jump from a correlation to causation! Note that the original study clearly stated that the finding shouldn’t be seen as causative!

'But I’d never do something like that!' I hear you cry

Recognising a correlation for what it is, a starting point for testing, is the hardest part. As humans, we’re wired to recognise signals and draw conclusions from what just happened to what will happen next.

As an Account Manager in email marketing, I still find it hard to stop myself making this mistake despite having this drummed into me since way back when I started my Psychology degree! I see correlations that I don’t understand and ask my clients to explain them and it’s difficult to challenge them and myself not just to jump to the easiest conclusion.

For example…

Opens and clicks often correlate in email marketing results so people assume that the clicks automatically follow the open. But this isn’t always the case as with some types of campaigns, people may get your emails and always intend to click (e.g. an email from your favourite shop on payday, with a subject line promising a discount).

In this case, the click rate is actually causative of the open rate. Plus, of course, there are many reasons people will open an email but not click.

The importance of considering all related factors

You might send out lots of product-based email campaigns and see how the content of the subject lines relates to open rates. You see that subject lines about perfumes tend to have higher open rates than subject lines about electronic devices, so you conclude that people are more interested in perfume than electronics.

This might not necessarily be the case if you have not considered all factors. What if the subject lines about iPads and the like are sending those emails to junk?

The recipient preference is not the determining factor here, it’s the deliverability of the email. 

The perfume description in the subject line is not causative of the higher open rate.

An example from when I worked in SEO

I’ve been out of the SEO biz for a while, but recall it being a challenge to truly move beyond simple correlations. A typical example:

You’re working in a B2B business selling products. It’s the start of a new financial year and you are given some new budget, so you pay a writer to write lots of relevant articles and place them on high-PageRank sites with links back to you. You rise up the SERPs, traffic to your website increases, your sales go up and you conclude that the increased sales were a result of the increased investment.

But what if…

• Your sales were going to go up anyway – it’s a new financial year and your clients also have new budgets to spend
• Your rankings were going to go up anyway – Google released an update favouring your SEO activity last year
• Your company’s other marketing channels are also busy spending their new budgets. This would mean more brand awareness, more links, more traffic to your site and a rise in SERPs.
• Your competitors have also spent money on article links but built more than you - the Google update actually penalised against this kind of behavior, but maybe you were affected less than others?

Takeaways

I hope that this blog isn’t stating the obvious, or preaching to the choir! I feel it’s something worth reminding ourselves about as often as we can because these mistakes really are made all the time. I see it often at work and every day in the media. Remember- the ongoing health crisis caused by the supposed MMR-autism link was based on thousands of people believing a misinterpreted correlation as a direct causation.

So what to take away from this? Three simple steps:

• 1. Recognise a correlation for what it is – a starting point
• 2. Identify possible factors/influences (e.g. deliverability, time of day, season, other marketing activity, creative/content changes, competitor activity)
• 3. Test, test and test again!