When it comes to data, the expectation is often raised that we will soon be able to leave part of our work entirely to the data. But just as a pilot remains in charge of the aircraft, no matter how much it is automated, we must also remain in the lead when it comes to data and data-driven working.
Meaning Data Driven
If you search for datadriven on Google, then tells Wikipedia you that data-driven means that “progress in an activity is achieved through the application of data, rather than through the use of intuition or experience”.
While the explanation is good, the term suggests otherwise. The term 'data-driven marketing' suggests that the marketing process is a machine in which you only have to put data in and vola: the right innovations and communication messages roll out. It suggests that you are 'guided' by data.
The practice is different. Besides the fact that such a machine – the great promise of big data – is only available to few companies, the main question is whether it is always the right solution, for example for your brand. Because the certainty we derive from data is sometimes a false certainty. That makes data-driven work complex.
Stay in control of the data
It's important that people stay involved, who understand what's going on under the hood. Only then are you able to learn from what is happening. And only then can you keep a grip on what is happening and intervene when necessary.
With the growth of data and the advent of smarter algorithms, we are increasingly letting software determine our soul and happiness. For example, online advertisements are automatically optimized. Bryan Melmed, Vice President of Insights Services at Exponential, a major ad network, explains how systems run when no one corrects them. In the example that he quotes, was advertised for an online clothing store.
By giving the algorithm full free rein and optimizing for clicks, gaming sites consistently came out on top. When optimized for conversions, 'people interested in clothing' showed the highest conversion. From this it becomes clear that people are always needed to understand what is really happening. In this case, people clicked on gaming sites very often and often clicked on advertisements by accident. The second is that clicks only say a small part about performance. Often other metrics, such as conversions, are much more accurate.
'You don't want a machine at the core making decisions for you.- Bryan Melmed
But there are also more and more companies that have applicants analyzed by an algorithm, based on their resume. The idea of smart software that analyzes millions of variables feeds the suggestion that the computer is many times smarter than humans and that they can let the data speak for itself. After all, who are we to dispute this immeasurable intelligence?
Algorithm example: Calvin Klein
On social media, algorithms determine who sees which content. Calvin Klein is a well-known American fashion house and brand, which sells jeans and lingerie, among other things. To stimulate the sale of bras, Calvin Klein ran an advertisement with a rather voluptuous lady to generate clicks to the webshop.
It is easy to imagine that many male Facebook users were spurred on by this ad and clicked on it. Because a relatively large number of men clicked on the ad, this signaled the system to send the ad more often to men than to women. This seems like a smart optimization, because men were after all more interested in the advertisement and would therefore form the target group.
In practice, however, it appears that women mainly buy their own lingerie. Men clicked through more often, but were said to convert much less. An undesirable result. If Facebook had had Calvin Klein's conversion data, it could have optimized on the number of sales or sales value. Then the system would not have optimized for men, but for women.
However, Facebook did not know that data. Facebook therefore optimized based on known data, but it did not know the data that it actually revolved around.
Be data-driven, not data-driven
It is important to realize that data can only quantify and not qualify. And that as a data-driven marketer you have knowledge of data, so that you can determine for yourself when data is of value and when it is not, and you can find the right people, tools and methodologies where necessary. Only then can you use the power of data. If not, there is a danger that you will let data make the decisions. Then the data is actually in charge of you.
That's why my advice is not to be data-driven, but data-savvy. To help you with that, I have the book Data-driven marketing written.
Image credits: John Spencer