6 tips on how to take care of the quality of customer data

6 tips on how to take care of the quality of customer data

Every twelve months, the amount of data on the Internet grows by nearly 40%, which translates into an increase of 30 GB of data per second. Just 20 years ago, this much information made up the entire network. How to find true and accurate customer data in this flood of information?

First: define the structure of your customer data

According to DBMS Institute research, only 21% of companies know how to acquire and process their data for marketing purposes. Therefore, it is first necessary to determine what data specifically has value. It is best to divide them into 3 categories: demographics, interests, behavior.

Demographic data include basic information about what or who the subject is, information about the customer's status and legal form, the contract they have, or contact information such as phone number, email address, profile information on LinkedIn or Facebook.

Data on users' interests can include the types of products they buy, the industries they operate in, the likes they get on social media or the forums they frequent online.

Behavioural data is best defined by information about their activity e.g. In your online store, use of the service, product or SAAS services.

This list of data is immediately available if you have e.g. CRM or a system enabling online sales. Below is an example of an online bookstore data model.

Second: identify reliable data sources and verify their accuracy

Typically, the primary sources of data are: the CRM system and data entered by sales and service, data from the transaction system – where users create accounts or perform a service, and data from social media and the company's website itself. Validating key data can have a beneficial effect on subsequent contact and engagement. The best example is the use of double opt-in model in lead campaigns and for registration in our online service. Not only does the customer consciously give permission to leave their contact, but in the process we confirm the existence of their contact information, which protects us from placing false information in our resources. And to these are exposed entrepreneurs using affiliate systems, where it is currently estimated on the basis of DBMS Institute research that 38% of the data entered do not meet the validation rules set by the advertiser. That is why it is worth connecting the customer registration process with a Facebook account, which requires a cell phone number during registration and has a validation via SMS.

On the other hand, if you already have a contact database, it's worthwhile to verify the data once a month in a service like „Check database”, thanks to which, using the API, you can determine whether the e-mail address is in a correct format or whether the selected e-mail address exists on the mail server.

Beware of the so-called "data bias". hard rejections, which can cause your servers to be put on blacklists.

Third: interact with your customers

Observing customer behaviour is cool, but interacting with them is even cooler. Last week, we informed them about a new blog post in the DBMS newsletter, and thanked those who read it personally. However, there was a mistake in the communication of the personalized message. Gentlemen's name was not displayed in the email. On the same day we received the following information from one of them.

This simple exchange of information made us aware of an important point. Our customers are our allies, so we can practically improve the quality of communication in this case.

Fourth: use data from different sources

According to the Data-Driven Marketing report prepared by the Netsprint group, the most frequently used data are the so-called. 1st party data, i.e. clients' own data (taken from their transaction platforms and client management systems) – used by 75% of respondents. However, only 47% of respondents use 3rd party data, i.e. external data from third parties – either directly from the owner of the data, or from aggregators and data warehouses.

This data is available in the form of ready-made segments to which the advertiser directs the marketing message, which can be used in SMS campaigns, or email marketing and display campaigns. However, the fewest marketers (21%) use the so called "social network". 2nd party data – own data coming from the partner (usually the publisher) and made available to the advertiser for the campaign. This is a big mistake, because in them lies the knowledge of the behavior of people who have at least once come into contact with our brand.

So the question is how to extract it? First, it's a good idea to give each contact an appropriate identifier and analyze their behavior through events such as. The most frequently used data are the percentage of openings in relation to the sent messages on the website, thanks to which if the recipient appears on the website or fanpage we will be able to give him a personalized message or thank him for his interest and give him a nice gift. Secondly, it's additional fuel to give scoring value to a specific behavior.

Is it all? Of course not – The main goal of campaigns using data especially 2nd party and 3rd party is to acquire leads and generate sales online. As many as 81% of marketers in result-oriented campaigns like to use their own data in retargeting, reaching with ads users who have already visited their website or online store, but did not convert. On the other hand, among 3rd party (external) data, advertisers mainly use for effectiveness actions:

– purchase intent – what products the user put into the basket, but did not buy,
– precise interest profiles – what topics did they intensively search for and read in the internet,
– geolocation – where he/she has physically been or is at a given moment,
– look-alike – data allowing to find on the Internet and reach with advertising users similar to those who visited the advertiser's website or made purchases there.

What are the benefits of this approach?? First of all, understanding the broader context of customer behaviour: from the reaction to an email or a web article, to a return visit to our website or fanpage.

Fifth: measure the economic impact of improving data quality

The process of measuring the effects of data improvement depends on the priority placed on it. In the case of online businesses, this may be indicated by the following. the percentage of openings in relation to the sent messages. This parameter, a little underestimated by marketers, indicates how often the recipient interacts with the brand and whether the message actually reaches them, which has a direct impact on sales. On the other hand, in activity-based businesses in the so-called "data-driven" environment, data quality is a key factor. "real world" metrics correlating your business with data may look different. E.g. in the debt collection industry it is important to have contact with the debtor. If the underlying contact information is not of the right quality, it is to provide a minimum of 30 percent of the right contacts, in relation to the debts purchased. Such a company may have a problem reaching out to a debtor.

Choose 2-3 KPIs that you observe on a daily basis that have a direct impact on your business. It can be the level of interaction, it can be the level of use of the service or the number of correction invoices. It is important for different departments to be able to comment on the origin of the numbers and their reliability.

Sixth: build a culture of working with data

The growth of the internet and data consumption entails processes related to data quality assurance. We still hear a lot about „democratizing data”. All in all a good slogan, and a very current „buzz word”. But there is a catch. Companies do not need every employee to have access to data and information. What they need is more employee access to good data and valuable information. And data quality and analysis efficiency is more a matter of processes (data cleaning, data governance, dictionary management) and people (analysts, domain experts who understand the context of the data) than tools.