In the new world of business, many people agree that data-driven marketing informs our understanding about customers, gives us clearer data about customers’ preferences and demand and has customers buy our products and services consistently. But the question is how businesses should start data-driven marketing.

Today worldwide businesses attach importance to and seriously invest in data storage and analysis to remove limitations and increase growth potential for the sake of profit. Data-driven marketing is a marketing tool that helps present the products and services that impressively meet the demand of customers.

Data-driven marketing needs the processes of data storage, analysis and synthesis to understand thoroughly the needs of customers in all dimensions. To guarantee successful data-driven marketing, there are four things that marketeers need to know and understand.

1. Business Objectives

Businesses or marketeers should have clear objectives for their use of analyzed data, such as a objective to expand a customer base to cover 1 million buyers in three years. A clear objective is the first and foremost success factor in data-driven marketing. With clear objectives, businesses can determine the scopes of their data analysis that will ensure the achievement of their objectives.

After there is a clear objective, the next step is the use case generation which varies according to the types and contexts of business and has to suit respective business objectives. Examples of use cases are as follows.

• For an objective to enhance efficiency in business operations – Customers will be stimulated to consistently buy products and services to maintain revenue. For example, cross-selling can increase revenue compared with the purchase of a single product. The products or services that exactly interest individual customers can be introduced in a timely manner (personalized promotions/products).
• For an objective to improve customers’ experience – Customers will gain good experience and be willing to consistently buy products and services. For example, customer segmentation and the calculation of customer lifetime value can be applied to offer privileges to the groups of customers who generate high income for companies.

After initial use case generation, the next question is which use cases will help businesses complete their objectives. There must be use case prioritization based on these two factors:

• business impacts such as financial outcomes, new business opportunities, return on investment (ROI) and compatibility with the overall strategies of organizations; and
• the feasibility of, for example, data, systems, work processes and external factors such as the Personal Data Protection Act (PDPA), laws and regulations.

2. Data Readiness

The most important element of data-driven marketing is, of course, data. Therefore, businesses must evaluate their data to find out the quality of the data and check if there are enough data for analysis and what kinds of data they have.

Today there are more kinds of data than those in the past. Data can be roughly divided into three kinds:

1. unstructured data including picture, video and audio files
2. semi-structured data such as files of the XML (extensible markup language) format
3. structured data such as tables of data in databases.

The design of use cases based on data can really create competitive edge for businesses. Therefore, business organizations should consider how their data were collected, what were the sources of their data, whether there are any missing part and how the data should be used. For example, there can be gamification campaigns to gather data about customers’ behaviors and also develop interaction with customers. Customer relationship management (CRM) can be used to help efficiently collect data about customers. Besides, centralized data lakes or data warehouses can be applied to obtain data from various sources.

3. The design and development of data analysis models (Design & Development)

Before designing a data analysis model, businesses should first realize what they want to know and why they want the knowledge so that they can choose a suitable data analysis model. There are four basic data analysis models:

1. Descriptive analytics – the analysis of what happened in the past including sales and the behaviors of the customers who used to buy products;
2. Diagnostic analytics – investigation into factors behind what happened through correlation analysis to see, for example, reasons behind increasing sales, which can be promotional campaigns or not;  
3. Predictive analytics – predictions of what are likely to happen in the future based on the analysis of data from the past; and
4. Prescriptive analytics – analysis to find what should be done in the future.

Businesses should create data analysis models after having clear business objectives, choosing use cases and readying essential data. Data analysis models can be created by the following stages:

• creating features for data analysis processes (Feature Engineering)
• inputting created features to the Model Training stage to find the patterns of data that give insights which can be further developed and applied for business purposes
• evaluating models through tests including Confusion Matrix, F1 Score and AUC – ROC(Model Evaluation)
• adjusting parameters to increase the efficiency of models (Model Optimization)

4. Turning insights into execution  

This is another important stage which follows the reception of desirable data from an analysis process. This concerns how insights can be turned into tangible business outcomes.

Before launching a marketing campaign, businesses should draft its outline which describes what is its target group of customers, what will be presented to customers and how businesses can contact the group of customers. Then acquired insights are used to design a marketing campaign. Here are the basic tips that can be applied.

• Attracting target groups of customers – The spending behaviors of each group of customers which are acquired through customer segmentation and the customer lifetime value (CLV) can be used to determine which groups of customers our promotional campaigns should target. For example, customers with high CLV should regularly receive promotional campaign news and discounts or giveaways may be offered to customers with less CLV to encourage them to buy more.
• Searching for particular products and services worth presentation – Data are used to develop the personalized promotions that serve the demand of individual customers. For example, discount coupons on baby products are offered to pregnant women.
• Using the communication channels that reach the most customers – Today customers have numerous purchase channels. Businesses can collect data from these channels and find which channels are mostly used by customers. Then businesses know which channels will guarantee the best impacts of their content and campaigns and can thus direct their marketing campaigns through the channels.

In sum, successful data-driven marketing must begin with business objectives, followed by strategy formulation, data evaluation, the design and development of analysis models and the translation of insights into business outcomes and then data will help improve marketing effectiveness and create long-term growth for businesses.