Data in order - the key to keep your customers

One day, returning from work, I looked into my mailbox. Apart from leaflets, I found a letter from the company where I lease my car. This company mails me rarely because I decided to receive e-mail correspondence. I expected to find something nice in the envelope, because I have been using the services of this lessor for several years and they have never had any problems with me. But I did not receive any thanks or birthday wishes. The envelope contained a call to pay overdue invoices under threat of being entered in the debt register.

How is it possible? I always pay on time. I checked the invoice numbers and started looking for them, first in the mailbox, then on my letter shelf. I couldn’t find them anywhere. And so I called the call centre and it turned out that, despite the consent for electronic communication, invoices were sent by letter to an address that I no longer use. I could say that I am responsible to some extent for this  situation because I did not update my postal address. What’s more interesting, the letter threatening recovery actions came to the current address. Will I continue working with this lessor after such an adventure? Probably not.

An almost successful marketing campaign

We are living in the age of digitization. Business processes are becoming more and more automatic, and marketing communication are tailored to individual customer preferences. Instead of sending 10 marketing campaigns a year, companies can run 100 campaigns at the same time. That’s a positive aspect because a tailor-made campaign is the key to success in the era of huge marketing hype. On the other hand, running such campaigns is expensive. It requires special IT systems, data integration, employment of specialists. And what if at the end of this process in a super-matched campaign our client receives communication with a distorted name or surname? Will it matter for him that the offer fits him perfectly, although we were unable to remember his name correctly? Probably not.

A promotion that wasn't meant to be

On the other hand, an offer which we send to the customer can cause a lot of confusion. Imagine a customer who has recently bought a car. Such a purchase is a complicated process and takes several months. Our hero sought advice on the Internet, visited several showrooms, took some test drives and compared several offers. In the end, he had chosen the one “dream car” and bought it. He had just begun to enjoy it when he received an email offer with a discount of several thousand Euros for the car he has just bought. Suddenly all the joy of buying a car disappeared. Perhaps our hero will contact the showroom where he bought the car. The difference will be either refunded to the client, causing a loss, or the claim will be refused. But after such a refusal, would the customer return to the brand that “cheated” him? Probably not.

Source of the problems

We might indicate several other similar “probably nots” e.g. sending an offer that suggests that the customer wants to leave us, although he doesn’t. Another example may be the communication with the client using a channel for which we didn’t receive the marketing consent. Or communication encouraging a purchase of new products while the customer is in the process of making a complaint about a previous purchase. What do all these examples have in common? They have the same source – lack of proper data care.

Every large business organization uses several IT systems these days. If our client has gone through marketing and sales processes, received an invoice and communicated with the call centre, we have at least 4 copies of information about him in the organization. Practice shows that companies store on average over 10 copies of data concerning the same customer.

The mailing address is usually stored for each of those copies, which is an easy explanation for sending invoices and “soft” debt collection documents to two different addresses. If the information about the purchase is in a database which is not connected to the marketing automation system, we will not detect that the customer has already bought a product that we keep offering him. Data in the source systems may be of poor quality, which can be manifested by mistakes in spelling the name and surname. In addition, low-quality data will have a devastating effect on machine learning algorithms, for example on the indicator predicting the customers who will cut ties with the company shortly. After all, how to assess which marketing consent is current when they are collected in many IT systems at the same time?

How to solve this problem?

The solution to the stated problem seems obvious, we must take customer data from all the databases in the company and combine it. However, this issue causes a number of difficulties. Data may contain errors such as typos. Different IT systems may store the same data in different formats, for example a telephone number may contain dashes or spaces. Finally, the set of fields collected between the systems may be varying. Email field may appear on the web forms with a home address missing. In addition, customer data changes over time, so the stored data is not necessarily always up to date.

Given those difficulties, combining customer data from different systems cannot be equal with gathering all the information into a single database. Identification of records representing one person requires three steps: standardization, enrichment and construction of so-called “the gold record”.

Organizing data

The purpose of standardization is to bring records from various databases to a state in which they are comparable with each other. For example, the standardization of a personal identity number consists of cleaning it of unnecessary characters and correctness verification (length, checksum). However, if we want to standardize the customer home address, we compare the entire address (street, number, city, zip code) with a previously prepared dictionary.

Data standardization alone is usually not enough, and for records to be comparable, they need to be enriched. Additional information may exist in our data, for example, the Polish personal identity number contains information about the date of birth and gender. For the purpose of data enrichment, we can also use external sources, such as the Polish Central Statistical Office database, containing current data of companies operating in Poland.

Customer found

After standardization and enrichment, customer information is ready to be combined into the gold records. A single gold record represents all component records from various IT systems which we consider to be a single real customer. The records are combined basing on keys such as the personal identification number or a combination of several values like for example: name, surname and address. At the end of this process, we obtain a list of gold records and all their component records.

In the next step, we can enrich customer data with other information we possess about them. This process is easy, because we can easily attach information about sales opportunities to the CRM component records, and about recent contacts with the customer to the records from the call centre system. We can build our marketing campaigns and sales communication only on the foundation of a 360 degree customer view.

Start with the foundations

Many business organizations are ready to invest large sums of money in the IT systems that strengthen (by means of data analytics) business processes such as sales, logistics, marketing and risk assessment. However, there is little awareness that each of these business constructions, which allow challenging the world, requires building appropriate foundations first. The investments in digitization of processes pay off only when the data is appropriately prepared and we gained knowledge about our customers. Machine learning algorithms powered by well-prepared data will be faster and more accurate, and marketing automation systems will not only become more effective, but will also require less control. We should remember that our clients are at the end of this process and that they appreciate the fact that we properly care for their data. After all, the way we care for client data is a great reflection of our care for the clients themselves.