Since the 2000s, it is common to hear that companies want to get closer to their respective customers to define themselves as customer-centric. Moreover, these companies have more and more individual customer data available through different channels. Thus, the marketing function moves gradually from mass marketing to segmented marketing to bring value to individual customers.
Companies want to simplify the switch to customer orientation by using the data harboured in their Customer Relationship Management (CRM) in order to construct a persona. We can define a persona as "a semi-fictional representation of your ideal customer based on market research and real data about your existing customers." (Kusinitz, 2018, p.1).
Why is it useful?
Persona’s can be seen as a tool which allows to come from a mass marketing to a fully one-to-one personalisation. Indeed, persona’s allow to cluster the mass of information that we have about clients into buckets of customers. Afterwards, it is easier for the companies to address issues of these ones. In other words, persona’s is a step between mass marketing and one-to-one personalisation.
How to make personas?
As explained above, a persona is built from actual customer data from a company. Thus, the starting point for the creation of a persona is to collect all the data available on its customers in order to build an exploitable dataset. However, it is important to know beforehand this dataset (outliers, heterogeneous data), which will be done via an exploratory analysis of data.
We must not forget to standardize our data so that we can compare them on comparable scales of measurement. In fact, the variables that make up a database generally have very different units, variances and standard deviations. Thus, it is important to standardize the variables by standardizing them so that they can be compared on comparable scales of measurement. We can describe normed variables as variables with a zero mean and a standard deviation of one.
When these preliminary steps are done, we can partition the data to highlight clusters. The best-known algorithm for clustering is called K-means. This algorithm will partition the individuals of a sample by grouping them based on similar characteristics. To do this, we want to minimize the intraclass variance while maximizing the interclass variance so that the individuals will narrow in on themselves when they have similar characteristics and are distant from other individuals when they are very different. However, one of the weaknesses of the algorithm is that we have to specify the number of clusters we want. To solve the problem, we can observe for which number of clusters the intraclass variance within the partitions realizes a significant increase when we decrease the number of clusters.The resulting graph generally takes the form of an elbow, hence its name (Robert L. Thorndike, 1953).
What can we do with persona’s once they are defined?
From a marketing point of view, when a company has a close relationship with its customers and knows each of them personally (& their respective needs), it has the ability to fully customize the offerings of its services and products. In addition, this firm can then determine different rates depending on the type of customer, but also offer adapted packaging.
From digital marketing point of view, these personas can enable us to perform user tests (UX Testings) to highlight the strengths and weaknesses of a website, and ultimately, optimize it. We can even think further by offering personalized content on the company's website for each persona built beforehand. There are already on the market different solutions to achieve this customization of website.
What does Semetis do?
At Semetis we create persona’s with the collaboration of our clients by organizing workshops with their teams and going through the analysis of all the data that we have at our disposal (CRM, Google Analytics, etc). As described in one of our previous articles: Web analytics, CDP's, the cloud & AI, An ecosystem perspective, we can build and manage a real data ecosystem so that, in the end, we can transform these data into insights that can be exploited for the company.