How marketers can improve and maintain customer data quality

According to IDG, poor data quality is one of the modern marketer’s most pressing concerns, with 66% of those polled citing data accuracy as a “top priority”. And little wonder, when you consider the importance of clean and accurate data, with successful customer relationships depending on having the right information to personalise communications. Furthermore, digital and physical customer communications that either don’t reach the intended recipient or contain inaccurate information can not only represent cost for no return, but could also damage a brand’s reputation.

Indeed, the impact of poor data quality has been demonstrated time and time again. Ovum Research, for instance, has previously calculated the damage as at least 30% of revenues.

Yet evidence suggests that marketers are their own worst enemies.

For instance, a Royal Mail Data Services study indicates that over half of all businesses have missing, incomplete or out-of-date customer data. According to the research, 63.3% of UK businesses reported out-of-date customer information, while 62.8% reported that their customer data was incomplete with gaps in certain data fields, with a further 60.1% admitting to having hardly any data for certain customers.

The report findings are not a one-off. Elsewhere, a recent Demand Gen Report revealed that more than 62% of organisations still rely on prospect data that is 20-40% incomplete or inaccurate. Additionally, almost 85% of businesses said they are operating CRM and/or sales force automation databases with between 10-40% bad records.

These findings are confounding considering that it has never been so important for customer data records to be clean and accurate.  

“Today’s customers expect a more concise and targeted communication model that relies on high quality data,” insists Steve Mepham, head of technical delivery at Celerity. “They expect to receive personalised and relevant communications that indicate not only that you understand your existing relationship with them but that you have also taken into account their preferences, which may differ by channel and communication type, e.g offers, news, generic & targeted advertising, service messaging, etc.”

“Today’s marketers are facing a huge data dilemma (and an unprecedented opportunity!) with access to greater volumes and variety of data than ever before. However it’s really the quality and governance of data that in our experience separates the best marketers from their competition,” says Vera Loftis, UK managing director of Bluewolf. “Well-integrated and accurate customer data is one of the best assets marketers have at their disposal to effectively personalise and engage customers, drive conversion rates, boost loyalty and trust, and ultimately maximise sales.

Today’s marketers are facing a huge data dilemma (and an unprecedented opportunity!) with access to greater volumes and variety of data than ever before.

“And yet, as Bluewolf’s most recent State of Salesforce Report revealed, nearly 80% of marketers don’t believe they’re offering the most relevant, personalised data to their end customers. Our report also finds that 60% of marketers cite poor or inconsistent data quality, or the lack of data altogether, as their biggest challenge to producing personalised campaigns.”

Natalie Khomyk, marketing manager at Data2CRM lists four categories of dirty data.

• Incorrect records. This is information that is false. For instance, the age of a client can’t be 150 years old.
• Inaccurate data. This is real yet incorrect information. Often, this is a mistake with the likes of postal codes or phone numbers.
• Inconsistent information. This is redundant data, such as duplicated customer records. It happens when you have no single data entry rules and different departments store the same customer under different names.
• Incomplete records. These are empty fields of data entry. This occurs when data is misinterpreted or doesn’t enter into the system.

When it comes to assessing the quality of your own data, Juanita McGowen, marketing analyst for Parker Hannifin, recommends approaching other departments to ask the following, initial questions around the quality of your marketing data:

• How reliable do your colleagues think the data is, and why?
• Has the data been reviewed and maintained since it was collected?
• How much (and what kind of) data do you really need to carry out marketing activity?
• Has an audit been run recently to check for duplicate information and consistency?

In addition to executing an audit to examine data quality, Loftis also recommends the following.

“When determining the quality of your customer data, it is important not to forget marketing basics and compare them against industry standards,” she explains. “If you are not hitting industry benchmarks for many of the following, for example, it probably means you have data quality issues:

• Email deliverability rates.
• Email bounce rates.
• Email open and click through rates.
• Unsubscribe rates.
• Contact field completeness.
• View of customer consistent between marketing, sales and service platforms.
• Ability to link marketing activity to ROI (opportunities, sales).”

But of course data decay is guaranteed for all marketing databases – contacts’ email addresses change as they move from one company to another, unqualified leads opt out of communications. Therefore, processes need to be put in place to ensure that after a data cleanse has taken place, existing data is not allowed to degrade. Indeed, reducing bad data is only half the battle - the other half is keeping it up-to-date. Therefore, the correct structures and processes need to be in place to ensure that the basic levels of data governance and quality are met. 

Paul Ballew, former global chief data and analytic officer at Dun & Bradstreet has suggested that best practice for data governance should include:

• Defining your data standards, including the metrics for adhering to those standards.
• Ensuring data quality at the point of origin and at key checkpoints as data flows through your organisation’s systems and databases.
• Adopting a unique, persistent key that identifies each entity, such as a customer, and the corresponding data that relates to that entity.
• Establishing a nomenclature and taxonomy to identify, categorise and organise your data.
• Implementing a rigorous data maintenance strategy to update constantly changing information.

Ensuring that rigorous steps are in place to ensure that data entry is as robust as it can be is an important step. According to Royal Mail Data Services, bad data entry is a leading cause of data issues, with less than half of businesses automatically validating the customer data they capture on their websites or as it enters internal systems. 21.4% still have no validation process for customer data captured via the web, while 15% fail to check data that is entered into their internal systems.

“When it comes to accurate data capture, validation and verification at source is the best place to start,” says Mepham. “The earlier in the value chain that this can be achieved the better. If this is not possible, then the matching and merging of that data in the ‘back-office’ needs to include relevant data management.

“For address information, the use of PAF, NCOA and various suppression databases is recommended to assist in understanding changes that customers may not notify you of. This is particularly important for those customers who are less engaged and therefore less likely to keep you informed.

“Where possible, ask the customer to supply, verify or update their own details, either through messaging or easy to access and use web-based applications/forms. Encourage the customer to confirm their details and provide additional information about communication, lifestyle and other preferences.”

Data maintenance

There will also be a need for ongoing data maintenance, something that can sometimes be overlooked after an initial crusade to rid the company of dirty data.

“The reality is that the need for ongoing data maintenance is often forgotten about when data quality plans are put in place,” says Jean-Michel Franco, product marketing director at Talend. “Data quality follows the same principles to other well-defined quality-related processes: it is all about engaging an improvement cycle to define & detect, measure, analyse, improve and control quality. This doesn’t happen at one time or one place.”

Within each organisation responsibilities around data need to be clear, including how data is shared, for what purpose and who maintains records.

Indeed, while the appointment of a chief data officer to lead ongoing data maintenance is an increasingly popular step, the responsibility for data management doesn’t stop at their door. 

Phil Hutchison, operational marketing director at Neopost, adds: “Within each organisation responsibilities around data need to be clear, including how data is shared, for what purpose and who maintains records. Data should be validated at each point of use, its accuracy checked and any remedial action needed taken. In addition, duplicate record entries should be weeded out to prevent communications being issued twice to the same contact. This not only avoids unnecessary costs, but also prevents an unprofessional impression being made on the recipient.”   

Mepham adds: “Regular data cleanses are recommended to ensure optimum quality. This involves the current database being extracted and passed through a variety of verification, validation, matching and merging processes, before being reloaded.

“Ensure that data is not kept beyond its shelf-life. Understand the value and relevance of the data you have and apply housekeeping to remove data which has lost its value over time, for example that which was captured but never used, cannot be properly verified or ceased to be relevant to your business today.

“Once systems become automated and continual data governance processes are in place, data will become more self-managing. However changes in business practice and rules, the acquisition of new data sources and changes in customer behaviour all need to be taken into account when considering data quality, as do changes in the data supply systems themselves.”

As summarised above, there are many approaches to defining best practices for data cleansing and management. As a quick takeaway, Loftis provides the following checklist of recommended steps to improve and maintain data quality:

1. Identify a data steward. Choose a specialist who defines and oversees policies, processes, and responsibilities for administering an organisation’s data.
2. Perform a data assessment. Identify issues with the data in order to plan cleansing and enrichment strategies.
3. Standardise salesforce and marketing automation platform data fields. Ensure data fields have consistent definitions and formats across applications.
4. Validate the field values. Confirm that data falls within defined limits or acceptable values.
5. Enrich the data. Improve and refine raw data with additional information like DUNS numbers, industry and address information, geographic coordinates, or alternative emails.
6. Selectively replicate and synchronise data. Instill a process that transfers only the most valuable data between applications and systems.
7. Cleanse data before mastering. Creating a single customer view is always more successful when the data is cleaned first.
8. Enforce data hierarchies and relationships. Not all data has a flat, one-to-one relationship, so maintain affiliations and relationships between records.
9. Trap data entry errors. Manual data entry is highly error-prone, so capture errors at point of entry to preserve data quality.
10. Repeat. Establish a process that regularly evaluates the effectiveness of all the steps.