At Blitz Digital, our business is built on data. We use it every day to gain valuable insights into how our client’s digital advertising campaigns are running, where the most traffic is coming from and which audiences are converting into paying customers.

As seasoned professionals in the data biz, we know that getting data analysis right isn’t as easy as it seems. Even with the latest analytical tools, it’s still possible to get it wrong. While many tools will make gathering data easy, it’s still what you do with that data that matters.

What Data Can Do For Your Business

Misinterpreting data can lead to false conclusions, damaging the success of your digital advertising. Worse still, misinterpreting data can also lead to false claims about what you are reporting.

These are a few common mistakes humans make when interpreting data.

Using Data From A Small Data Set

Taking data from a small data set is a surefire way to jump to false conclusions. And it’s not just digital marketers making this mistake. Recently, we’ve seen reporting on coronavirus numbers that have failed to take into account the wider context of the data they are analysing.

In July of 2021, 399 vaccinated Israli’s tested positive for the Delta strain of coronavirus. In the same period, 320 unvaccinated Israli’s also tested positive for Delta. These numbers were significant because Israel had reportedly vaccinated large portions of their population.

However, if we take a step back and consider the broader statistics, we see a somewhat different picture.

In August, with more data we can see that unvaccinated people (around 20% of the population) now make up 50% of the newly infected people. That number is expected to grow as infections and hospitalisations among the vaccinated plateaus. Furthermore, correct reporting of the case numbers in Israel showed that the elderly were eight times more likely to experience a severe case of covid than their vaccinated counterparts.

An increase in sample size decreases sampling errors

By analysing the data too soon, it’s easy to jump to the wrong conclusions. In this case, that conclusion was that vaccinated or unvaccinated you were just as likely to contract and spread coronavirus. While the statistics from July were alarming, they were only one part of a much wider situation that was yet to unfold.

Taking a snapshot of the data from Israel led many people to misinterpret what the data was proving. By zooming out and looking at the bigger picture, we can see what having a vaccinated versus an unvaccinated population does to the spread of coronavirus.

In clinical medical trials, baseline imbalance is where the characteristics of patients can interfere with the trial of a medicine or treatment. Having too small a sample size means a person can easily skew the results. An easy remedy for this is by increasing the trial size. As the size of the trail increases, the absolute size of imbalance reduces. It’s the same for digital marketing.

Paying too much attention to outliers

Small sample sizes and outliers are highly correlated. Outliers are the random events that tend to happen, which can skew our data in one direction.

Take for example, if you gathered a group of ten people and wanted to find out their average income. If nine of the sample earn between 50 and 70 thousand dollars, but your tenth sample is a millionaire, the average pay across the group will skew much higher. Obvious huh? Unfortunately, when it comes to outliers in digital marketing, they aren’t as easy to spot as this example. What’s needed is a careful eye and a curious attitude.

So, how can you spot outliers in your analytics? Take this real-life example. We ran a digital marketing campaign for a client and saw a huge surge in traffic coming from the campaign. We could have stopped there and reported that the campaign was a wild success. Instead, we put our detective hats on and grabbed our magnifying glass!

On closer inspection, we saw that the increase in traffic was coming from an influencer account. Furthermore, even though our campaign did well, the traffic from the influencer wasn’t what made it a success. The boost in traffic didn’t make much of a difference to the number of people who were converting into customers. This traffic was visiting and then dropping off.

By focusing on our objectives (converting customers, not website traffic) and taking a second to look at where the extra traffic was coming from, we were able to report on the data that mattered to our client.

The Importance of Zooming Out

We know that a significant boost in data can be exhilarating. It can be easy to see a jump in traffic and equate it with an increase in customers. However, if we were to ignore where that traffic is coming from, we risk making assumptions about other parts of our business – like how many sales we might make next month. This can very really mess up your bottom line.

Instead, asking questions and zooming out on the data in front of you is the best way to determine whether the trends you are seeing are significant. Disinformation relies on distorted data. Learning how to interpret data will do wonders for your business and stop you from jumping to conclusions. When presented with unusual spikes in data, ask yourself:

• Is this an outlier?
• What is driving this data?
• Does it back up the long term data consensus?

Take a step back and consider outliers as what they are: the exceptions, not the rule.

If your business is struggling with data analysis, it’s time to call in the experts. At Blitz Digital we thrive on all things data. We can help you make the most of your data and make sound decisions based on real insights.