It is fairly expensive for marketers to ask every Rama, Shama, and Bhama if they want to buy the newest gadget. They shun the dragnet for a line baited with the right kind of messaging that can get a consumer to bite. The trick is to trawl in the right spot. This is where market segmentation comes in handy.
Bundling like-minded consumers into the same groups expands the probability of scoring the sale. Thanks to the digital revolution, marketers can now explore zettabytes of data to develop these segments. Algorithms can find correlations that escape recognition by a human, information that marketers can then use to develop market segments.
On paper all this sounds great, and experts certainly agree on the basics of segmentation. However, where they differ is how to use the technology to keep marketers from over-segmenting.
Human touch
No matter what, keep a human in the loop
The problem starts with a flawed dataset. Single datasets are risky because they either provide incomplete or faulty data. To avoid this, the dataset needs to be appraised by validating it against other sources. This is where the conflict between analysts and data scientists begins. The analyst looks at the primary data; and the data scientist, for accuracy, creates multiple searches of data.
The data scientist knows the data and the science, but there is a need for someone who knows the business and the brand as well. It all boils down to a human who will question the data. Therefore, while machines may be great at finding answers, they don’t know to ask specific questions.
Find the bull’s eye, then call the shot
The main challenge is creating segments.
Trying to create characteristics based on internet activity is not reliable; the data will contain a lot of noise. There may be a a lot of internet cookies, but there is no transparency in how these cookies gathered the data; so proceed with caution as the quality of the segments may vary.
Right direction
Make sure that you identify the segment in a reliable way
The goal is to dodge false leads.
First, run the analysis on half the data set, then run it on the other half. Some specific segments based on certain parameters that show up in the first run-through may be outliers, and therefore can be discounted if they do not appear on the second run-through. The precisely-grained segmentation may distinguish customers based on their product preferences, purchase frequency and spending levels. Such demographic details as gender, location and age may be relevant too.
At the juncture, there are two theories that come into play — fake and silly. The more finely-grained the segment, the more likely it is to be fake. Similarly, it is silly to cut a line so fine in the specifics, that you may not see data with those characteristics again. Here, cross-validation is imperative because you want the analysis to be assenting, not exploratory.
Though automation can check data twice, only a human can provide insight. Therefore, understanding consumer behaviour and the data is what leads to good and valuable insights.
Coming back to market segmentation, it is important to understand that a segment has to be “not fake” and must also exhibit ‘future behaviour’ that is predictable. For example, look at the “early adopters”, those people who are the first on their block to buy a new gadget. They will buy an iPod, followed by a smartphone, then the Apple Watch, and so on. The segment generates repeat sales.
Consumer behaviour
Not knowing consumer behaviour can leave a marketer in the dark. The market segment appears subjective when the marketer is unaware of the “secret recipe” that drives the segment. The outcome is one of two: either the marketer is blamed for not doing a good job when the campaign flops, or they come to distrust the big data.
Therefore, do not treat the technology as a black box by throwing data into the juicer and drinking the extract that comes out. Someone on the team has to understand the fundamental forces that shape a segment or you will never know when you are going off the radar.
The digital marketing ecosystem is not as simple as we believe. It is common knowledge that customers expect to to spend less, but gain more value. This puts pressure on the marketing team. Not only do they have to be the good samaritans of the campaign, but must also ensure that there is little or no room for error and wastage.
Over segmentation
Let’s look at the million dollar question: How much segmentation is too much?
Generally speaking, as a marketer, you have to keep personality segments at the bottom-end. With the personalities, you create specific content tactically targeted to those buyer types.
So far, so good.
It looks like marketing hasn't changed. Right?
You would think so, but that isn’t the case.
Today, with artificial intelligence, machine learning, natural language processing and a swarm of other technologies, we can explore big data shrewdly. It is possible to create several “micro-segments” relying on AI to do the cherry-picking.
Over a period of time, AI will filter those segments and keep only the most effective ones. Ad makers will create ads that reach only those market segments in a language that resonates with those personalities. IBM's Watson is perhaps an AI platform that most are familiar with. There is also Salesforce’s Einstein. Inevitably, other platforms will follow suit.
The rise of AI does have the potential to disrupt entry-level marketing jobs. Algorithms can handle the workflows of e-mails and social platforms, predicting which consumers will open and respond to email pitches, and even crank out personalised messages after comparing them against the segmented data. As long as the outcome and the key result areas reach the expectations of your boss, they are going to keep using technology.
And mind you, technology is only going to get better.