Home Business Data is king, but you need to know how to use it

Data is king, but you need to know how to use it

by Ana Lopez
0 comment

Ori Mendi is a serial entrepreneur and the founder and CEO of Kueeza publishing platform that serves more than 150 million monthly users.

“Data is king”, “data driven”, “data, data, data.” These three expressions tell one story: Data is the most important thing parameter for your business and should guide all your decisions. But where do you start?

Step 1: Identify the right idea

First of all, you need direction, because data won’t help you find an answer if you don’t know what the question is. Suppose you have a website and you want to increase the conversion rate (the percentage of visitors who take a certain action, such as clicking on a banner, making a purchase, filling out a contact form, etc.). Without a direction, idea, or innovation that is likely to work and deliver results, you can immerse yourself in hundreds of activities and trials that drain endless resources from your team without getting anything in return. That is, your operations will technically be “data-driven”, but you will still lose due to poor resource allocation.

The world is full of opportunities. For me, the best way to identify the right opportunity is to get the proof of concept. This evidence could be the success of a similar product at a company close to mine in several characteristics, or perhaps a strong intuition shared by several people in the organization (and not just in my own head). If you see such signs, it is safe enough to continue.

Step 2: Understand the data

Data is data and can often be seen as synonyms of facts. But data is highly dependent on your point of view and the depth of your analysis, and in many cases your first impression isn’t necessarily the right one.

Let me give you an example of our work at Kueez. We ran an experiment on a network of websites we own: in addition to the existing advertisers already appearing on the websites, we launched a new video ad element and waited for the results. We saw that it generated a lot of money and enthusiasm, so we kept it. We worked according to the data and that was a success. After all, data is king, and if it makes a significant percentage of sales, then it must be significant, right? All right, let’s take a look.

Step 3: Understand the data in an alternate world (A/B testing)

In the case I described above, we uploaded two separate versions: a basic version and a custom version with video ads. Despite the promising results, in terms of high revenue from the video ads, the experiment showed that there was actually no added value. While the video player brought in a significant percentage of revenue, it did so by cannibalizing the other advertisers. So the average total revenue per visit was basically the same in both versions.

The insight here was that despite the data showing enough revenue from the video player, the overall picture showed that this experiment had no value other than perhaps diversifying the advertisers.

Step 4: Do an in-depth analysis of the A/B tests

To so conducting an in-depth analysis, we must first understand who our customers are, that is, create customer segmentation. The world is divided into different types of customers: Someone who uses an iPhone behaves very differently than someone with an android phone, a Portuguese speaking customer is different from an English speaking customer, and an English speaking customer in the US is different from an English speaking customer in the Philippines. Advertisers also have different effects on different customers.

In certain customer segments, we found that our average revenue per visit was significantly higher in the modified version, while other segments showed the opposite. In many segments, the experiment added significant economic value to us, so we chose to keep it for those segments. That is, customer segmentations allowed us to identify the places where the experiment worked well, so we could make an improvement in those places while avoiding harm in the places where the experiment didn’t pay off.

So even though the overall (average) results of the A/B testing showed no change or particular value, we still found great value in the product. Only the value was reflected in certain segments and not in others, canceling each other out in the total result.

What lesson can you learn from this story?

This story may have annoyed you, just as the experiment annoyed us. But it was worth it in the end, because despite the desire to boil things down to a simple yes or no answer, in most cases the reality is much more complex than that. The question is not “Does it work?” but “Where does it work?” and sometimes also “When does it work?” because what works now may not work in six months.

To make optimal data-driven decisions usually means not being bound by a straight yes or no answer, but instead regularly working with detailed data (divided into many segments) and responding to questions accordingly. Don’t be blinded by averages – while they are the easiest to measure, they rarely give you the full picture. These foundations will help you make the right decisions and adjust them regularly.

businessroundups.org Business Council is the leading growth and networking organization for entrepreneurs and leaders. Am I eligible?

You may also like

About Us

Latest Articles