This answer might satisfy a statistics test, but for marketers, the real issue with establishing a relationship between two things is understanding how the story your marketing data tells. Correlation studies should show marketers how to make better campaign decisions.
Unfortunately, most marketers experience the issue of showing a correlation between precise marketing efforts and campaigns with precise outcomes. You may be able to show some connections, but the relationship is tenuous and far from effective in really driving home actionable insights
Marketers look can look for relationships between any two data points. For example, marketing analysts might wonder if more Facebook shares of a particular URL lead to a higher Google search ranking. Perhaps marketers want to know if emails with red call to action buttons get more clicks. Maybe the question is — “do posts with images get more shares across social media?”
There are thousands of questions marketers might ask, and while correlation is not always causation, discovering important relationships through correlation data is undeniably valuable.
Correlations are best suited to detect linear relationships between variables. If a relationship isn’t linear often the correlation will be missed in analyses (based on often used Pearson correlation).
So correlation ≠ causation but the reverse can also be true. A causal relationship between variables without a clear (linear) correlation.
But when you see a correlation this could mean: causal relation, random chance or a third factor in play which needs to have a causal relation with both variables independently causing the correlation. You can learn a lot looking for this third factor!Roel Willlems – https://roelwillems.com/
When you apply an incorrect understanding of causation as it relates to marketing campaign successes you can end up making costly marketing decisions to increase budgets or reduce budgets but actually end up failing to impact your desired result.
Identifying correlations can help marketers::
In short, correlations give insights that help marketers make actionable decisions, fine-tune marketing strategies, and ultimately, grow businesses.
Moz recently conducted some interesting in-depth correlation analysis of data commonly available to search engine marketers. While they openly acknowledge that correlation is not causation, the results yield some insightful information about keyword strategy when it comes to what actual hands-on work gets prioritized in your SEO campaign.
Freelance SEO consultant Dave Smart posted an epic thread on twitter, about how understanding types of correlations can help you filter and apply information gleaned from “correlation studies” properly.
1. Random coincidence: They align, but that’s just because they happen too. Fun stuff, but increasing mozzarella consumption isn’t going to increase the amount of qualified civil engineers.
2. ‘Reverse’ correlations, where the trend is connected to the thing you are studying, but that’s because it’s affected by it, it doesn’t control the thing you are studying.
I used to work in aerospace labs, & a very sensible boss taught me a lesson that stuck with me here: Don’t heat the aeroplane. We used to do stuff for fast jets, & one of the things that happens when something is [bouncing off through the atmosphere at plus 1 Mach is… they get hot from the friction. There was a clear chart of leading edge temps & speed with near +1 correlation.
On that data alone, if you confused the symptom for the cause, to make a fast aeroplane, you just need to heat it up. The hotter the faster. (That’s not a good way to make a fast plane faster by the way).
These kind of correlations can be useful for working out things you need to do, in the plane example make sure the plane can stand the heat.
In a web[SEO] example, making sure the site loads just quickly with 1,000 users on at the same time as opposed to just 100 at the same time becomes important IF you started ranking in 1 for a heavily searched term!
3. Actual Causation! That’s what we want, but we can only determine this is what’s happening if we change the variable highlighted & that has repeatable results.
Some are easy, like does adding a noindex tag affect ranking position? Yes, it’s an easily measurable thing with a strong binary outcome.DAVE SMART
To study it is simple. Ask a different question, “Do clicks from different locations affect google’s understanding of the relevance of a page to that term / city?” & it’s a whole lot muddier.
You might never determine this actual causation no matter how big your study as there’s way to many unknowable variables, as we don’t have access to the same picture google has, we don’t know if there’s something else doing this and we’re actually seeing a false or reverse correlation.
This type of data analysis is not reserved to just keyword research. Marketers can use it for social media, paid search, email marketing, design, and more.
You name it, you can conduct correlation analysis to see if any relationships stand out to you. The key to identifying causation via correlative study is to limit the amount of variables that go into your experiment.
If you want to get answers, you can’t run multiple experiments at the same time. You need to be able to track precise data points. Don’t change 4 or 5 seperate factors if you’re trying to make an A to B relationship.
The formula for correlation (r) is:
But, I’ll stop right there. No marketer has time to sit around and do math by hand all day. That’s where marketing data analytics software like Tapclicks comes in handy.
Instead of finding the value for x and y to find r, you simply pick your variables and let the software do the hard work for you. That way, you can spend your time using the data to make better marketing decisions.
I was able to configure two widgets with a data view for Google Analytics, that shows the last 6 months data on Bounce Rate, how many people visit just one page, and Conversion rate of visitors turning into customers.
We can, from this data, begin to imply that there is SOME potential correlation between these data-points as the decrease in bounce rate falls in line with an increase in Conversion rate. However, that’s just the beginning of your process, as you would want to test that relationship by changing some variables.
You might change the medium/source and find that your bounce rate for a certain category of traffic like paid ads is much lower, and you started an ad campaign that increased your conversion rate.
That’s a MUCH different insight than “Improve bounce rate to improve conversions”.