Your customer database is the foundation for the work that you do. Whether you’re building marketing campaigns, testing and improving the product experience, handling customer support tickets, or running predictive analytics queries, you’re likely leveraging customer attributes, insights, and engagement history when making decisions. Given the importance of customer data, therefore, you’d think that many organizations prioritize customer data quality management over everything else. Unfortunately, that’s not always the case.
Although customer data has become more available, teams frequently struggle with quality components such as data consistency across systems, data accessibility, data plan violations, and more. In fact, a 2018 Harvard Business Review study found that only 3% of surveyed companies had acceptable data quality standards.
Organizations with sound data quality management processes, on the other hand, are able to make data-driven decisions with more confidence and launch more efficient targeted campaigns. This post will walk through why customer data quality matters, what standards you should be setting for your customer data quality, and the steps you can take to manage data quality.
Why does customer data quality matter?
In today’s customer-driven era, experiences that are personalized to the user and connected across channels have become the norm. For marketers, this introduces a challenge — any mistargeted messages can have a huge negative impact on the customer experience, leading customers not only to not engage but possibly to take on a negative perception of your brand.
As customer data is the foundational asset with which you make targeting decisions, a lack of data quality at the start can lead to poor end results. Additionally, if you’re leveraging ML predictive intelligence to drive your targeting, the quality of recommendations generated will only be as good as the customer database that your models are trained on.
With a system in place for data quality management however, you’re able to deliver targeted campaigns faster, and with more confidence. For any paid campaigns, the increased targeting accuracy will make your campaigns more cost-efficient, allowing saved budgets to be reinvested.
What does a high-quality customer database look like?
When managing customer data quality, here are a few key characteristics to focus on.
Consistency, accuracy, and completeness
As customer events, attributes, and insights are collected from different digital touchpoints and cloud feeds, data points are often formatted differently. For example, one tool may be implemented to collect a user’s first name from your iOS app as firstName, while the same tool is implemented to collect the first name from Android as First_name. Such inconsistencies can make it difficult to build audience segments and run queries within that tool’s database. To maintain customer data quality, it’s important to ensure that customer data points are consistent across user profiles, no matter which channel they’re collected from.
Accuracy and completeness of customer profiles are also critical. If the data points attached to each customer profile are not correct or up-to-date, there will be a detrimental impact on downstream campaigns, analysis, and modeling. Furthermore, if each customer profile contains only a first name data point, you’re not going to be able to use your database to make many substantial decisions. It’s important to identify and collect the data points that are important to your business applications while maintaining user consent at all times.
Teams across the organization from marketing, product, analytics, engineering, to support, all need to leverage customer data to make decisions. Even the most robust customer database is not worth much if key stakeholders cannot access it when they need to. When access to data is democratized, multiple teams are able to work off the same, high-quality data set, and make strategic decisions.
Proper event collection
As your customer data set grows, manual data quality updates can become an extremely laboursome task. Additionally, even if you do invest the time to cleanse your customer data once, it’s not the best use of data engineering resources to perform data cleanups again and again as you collect new data. Having a feedback system in place that helps developers ensure they have proper event collection at run time and that implementations are correct across platforms is important to managing customer data quality in the long run.
What can you do to improve data quality?
Ensuring high customer data quality requires a quality management system that consists of cross-team collaboration on data planning, data point validation, identity resolution, and developer support.
Cross-team collaboration on data planning
The first step of determining if your customer data points are correct is defining what “correct” looks like. To do so, it’s important to have key stakeholders from across the organization collaborate on establishing what data points you need to collect for each team.
Marketing will need certain user attributes and events to build audience segments for their campaigns, Analytics and BI will need a subset of data to test and train their models on, and Product will need granular user events to understand engagement. The sum total of desired data points can be tracked in your data plan, a living document outlining what’s being collected, how it’s being named, and what it’s being used for. Your data plan can be tracked in an internal system, or you can use pre-built data planning functionality in your Customer Data Platform.
Once you have your data plan in place, it’s important to identify incoming data points that validate the rules set in your data plan. Using A Blueshift partner, like mParticle, it’s possible to include the specific data plan ID in your implementation, making it easy to track data plan violations in the mParticle UI.
With a system in place to surface data plan violations, it becomes much easier to review validation reports and track data quality. Remember to keep your data plans up-to-date as your data strategy evolves over time!
When you have users engaging across multiple digital touchpoints (email, website, mobile app, paid social, etc.), it’s important to have a framework in place that allows you to merge the actions made by the same user across different channels and devices to a single, persistent user profile. Without this unified view of customer event history and attributes, it’s difficult to ensure that the messaging you’re delivering to a customer is up-to-date with the latest actions they’ve taken and to ensure that messaging complies with regulations such as GDPR and CCPA. Having an identity process in place that allows you to automatically tie engagements across channels to a single user identifier, and to match anonymous activity to known profiles when that anonymous user makes themselves known, makes it easier to deliver consistent, cost-efficient experiences across channels.
Support developers to minimize violations
For any data-driven team, the ideal state is to have no data plan violations ever, and therefore perfect data quality. While this is a lofty goal, you can support your chances of success by enabling your implementation team with handy developer tools that support proper event collection. With your data plan in place, developer tools such as Linting tools and mParticle’s Smartype make it easier for devs to ensure proper event collection at run time by automating code completion and quality feedback based on the standards set in your data plan.
Improving customer data quality is an ongoing process that requires systematic management and cross-team collaboration. While doing it successfully is a challenge, having the right tooling in place makes data quality management a lot easier.
Customer Data Platforms, for example, allow you to collect customer events and attributes from across digital touchpoints once, tie that data to persistent customer profiles, and forward that data selectively to your favorite marketing, analytics, and data warehousing tools. Having a CDP in place as your single point of collection makes it easy to address data quality as data points are ingested, ensuring that once those data points are forwarded downstream to your other tools they are accurate and formatted consistently.
To learn more about how our partner, mParticle, makes it possible to simplify your data pipeline and enables data planning, data point validation, and developer tooling, you can explore their data quality features here.