The Ultimate Beginner’s Guide to Data Quality and Business Intelligence

13 Min Read

The Ultimate Beginner’s Guide to Data Quality and Business Intelligence

Today’s marketers are becoming technically savvier.

The Ultimate Beginner’s Guide to Data Quality and Business Intelligence

Today’s marketers are becoming technically savvier. They understand the need to improve customer experiences or implement digital marketing strategies to engage consumers across channels. Customer retention and acquisition, Big Data, social media marketing, and content marketing are just a few of the goals and strategies in today’s marketing toolbox.

However, perhaps not so widely discussed are some important fundamentals – high quality marketing data.

As marketers we understand that without data, there are no insights. But managing the quality of the data and applying analytics are key to successfully implementing all these other great marketing goals.

When data goes bad, even the best laid out strategies are doomed to fail. After all, “Garbage In, Garbage Out”, right?

So let’s take a step back from the world of Big Data, Digital Marketing, and Customer Experience Management to focus on the basics – the data.

Data Quality by the Numbers

Let’s first check out some of these great stats by Halo Business Intelligence:

  • Nearly 40% of all company data is found to be inaccurate.
  • 92% of businesses admit their contact data is not accurate.
  • 66% of organizations believe they’re negatively affected by inaccurate data.

The implementation of a data quality initiative can lead to reductions of:

  • 10-20% corporate budget
  • 40-50% IT budget
  • 40% operating costs

And increases of:

  • 15-20% in revenue
  • 20-40% more sales


Dirty Data is Damaging

Unfortunately, many organizations take a reactive approach to data management, only taking action when something negative occurs. For example, multiple messages sent to the wrong person damages customer relationships, revealing a need to improve customer information. 

However, choosing NOT to fix dirty data can be extremely damaging. Consider these data quality horror stories:

  1. “Dear Idiot” Letter: As customer service representatives at a large financial institution dealt with angry customers, they began entering phrases into the salutation field such as “What an idiot this customer is.” When the marketing department decided to send a marketing campaign using the customer service database, letters went out as “Dear Idiot Customer John Doe.” Customer relationships were ruined and the company’s reputation was damaged.
  2. 17,000 Men are Pregnant: Due to incorrectly entered medical codes at British hospitals, thousands of men apparently required obstetric and prenatal exams. These seemingly simple errors caused disastrous results in billing, claims, and regulatory compliance.

While these examples are extreme, the negative impacts of dirty data on your business are very real.

How to Get Your Data in Order

As part of any new data initiative, a business needs analysis should be performed to understand what is required of data moving forward. A business needs analysis focuses on understanding business objectives, strategic goals and business drivers.

For example, what information is required to meet these objectives and how accessible is it to end users? Are data gaps occurring, limiting the availability of required information to support decision-making? What data issues may be impacting revenue, increasing costs, or causing inefficiencies in operations?

Documenting business objectives helps determine what data should be captured, how the data is related, and how data should be structured to create value.

Once a data management project is approved, data must be properly cleansed and integrated to ensure information is of the highest quality to drive smarter decision-making across departments.   

Begin by evaluating the quality of your data with a data assessment. Many vendors offer a complimentary assessment to help identify areas where data quality can be improved, what types of data may be missing, and other problems that may be affecting optimal data performance.

Data must also be integrated and placed into a central repository for a complete, 360-degree view of the customer or other business area. Data quality software automates integration processes and improves data quality by performing the following functions:

  • Parsing and standardization
  • Matching and linking data
  • Monitoring to ensure data continues to align with business rules
  • Enrichment through data appending

Business processes should also be established to ensure data manually entered into systems is of the highest quality possible. As we learned previously in our example of the pregnant men, many organizations experience data errors when information is manually entered, at a rate of 2% and 8%.  Even one wrong number entered incorrectly can cause a payment to fail, a wrong part number to be shipped, or apparently a man to become pregnant.

Data validation controls can be integrated into on-line forms, using rules to check the validity of data sets. For example, an on-line website form may require a visitor to enter data in specified formats. Or an IRS form may utilize controls to check that positive numbers are being entered into fields.  Training employees to be more aware of the importance of data quality is also a crucial step to achieve a company-wide awareness of maintaining high quality information.

Analyze Your Data – Create Customer Segments

With an integrated and clean database in place, customer analytics can be applied to target customers with the most relevant offers. Begin by creating customer segments. Customer segmentation refers to dividing customers into groups who share similar characteristics, such as age, gender, lifestyle, and so on.

Any number of segments can be created as long as each segment is:

Large Enough to be Profitable:  The segment should not be too narrow, making it cost-prohibitive to reach these customers. The segment should be worthy of the marketing efforts.

Accessible:  The consumers must be able to be reached through channels already established, such as a website or store, or by new channels that can easily be created.

When creating customer segments, a company must consider a wide range of customer characteristics, such as:

  • Behavioral:  characteristics such as product usage and frequency of purchase.
  • Demographics:  factual characteristics, such as age, gender, occupation, and income. For example, are the majority of your customers female or male? Where do they live? Are they single or married?
  • Psychographics: values, attitudes, lifestyles. Psychographic information answers questions such as what motivates your customers to buy your products and services? What are their key values? What are their hobbies and habits?
  • Value-Based: actual or potential revenue of customers and the costs of maintaining relationships with them. Analyzing these attributes allows marketers to allocate resources to the most-profitable customer groups.

Based on any combination of these characteristics, companies can develop key customer segments and develop marketing strategies designed to generate the most profit from each unique customer group. A company may want to enhance loyalty, increase customer value, or provide products and services to a particular geographic area.

Consider the following examples:

  1. Geographic segmentation
    Marketers may create segments based on geographic location, such as neighborhoods, cities, states, regions, or postal codes. A local retailer may want to only target consumers within a certain radius of the store. Or a large retailer selling seasonal products, such as winter coats, will concentrate more marketing efforts on areas with colder weather.
  2. Life Stage segmentation
    Life stage segmentation requires looking at a combination of demographics and psychographics characteristics to determine where consumers are in their life cycle. Different marketing techniques will appeal to different segments. Targeting a college student will be much different than targeting a young family or senior citizen.
  3. Customer-value based segmentation
    By creating segments based on transactional history, such as average spend, products purchased and frequency of purchase, companies can devote more resources and create best-in-class offers for their most profitable customers. A bank may want to increase the credit card limit for high-spend, high-value customers. Or an on-line retailer may offer free shipping to profitable customer segments.
  4. Buying Frequency segmentation
    Creating a segmentation based on buying frequency is good for targeted marketing campaigns, such as re-engaging past customers or rewarding frequent shoppers. For example, by creating a segment of customers who have not purchased in the past six months, a special incentive can be sent to this group.

In the following example, a regional retailer has identified three segments
and the marketing strategy for each.



Use Modeling to Predict

Similar to segmentation, predictive modeling allows marketers to develop very precise, targeted campaigns. Both techniques examine the characteristics of customers and prospects, however modeling takes this one step further by also predicting future behaviors. Modeling is the practice of forecasting consumer behaviors and assigning a score based on the likelihood of completing a desired action, such as purchasing a product. For example, which customers are most likely to spend the most across a 6-month cycle?

Check out the following examples of how a predictive model may be used:

  1. Response modeling
    This type of modeling is used to identify customers or prospects most likely to respond to marketing offers. By collecting and analyzing data on individual customers, marketers assign scores to individuals representing their likelihood to perform a desired action, such as a product purchase. Highly targeted offers can then be sent to those with the highest scores.
  2. Propensity to Purchase Model
    This refers to the likelihood of a customer to purchase a particular product. By better understanding a customer’s purchase propensity, companies are more likely to up-sell or cross-sell these customers. Types of data that are often included in this type of model include purchase behavior in the last 12 months, amount of spend, frequency, and other actions such as increased clicks on a website. By understanding a customer’s likelihood to buy, consumers in the market can be targeted with the right offers at the right time.
  3. Customer Churn modeling
    By better understanding what is causing customers to stop using your services or switch brands, processes can be put into place to mitigate churn risk.  A customer churn model evaluates customer behaviors such as purchase frequency and date of last purchase. A possible indicator that a customer may be thinking of leaving is decreased purchasing frequency. By implementing triggers based on changing customer behaviors, customer retention strategies can be put into place before a customer switches brands.

Data quality and business intelligence are critical for success in today’s economy. More companies are increasingly investing in data management and business intelligence solutions to maintain high quality data and target multi-channel consumers. And with better data insights, marketers are better able to focus on today’s data-driven, technically-savvy marketing strategies.


Share This Article
Exit mobile version