A CRM Data Quality Framework and Best Practices
- According to analyst firm Gartner, "Poor data quality is a primary reason for 40% of all business initiatives failing to achieve their targeted benefits."
Source: Measuring the Business Value of Data Quality report.
- CRM data is a perishable asset. Unless managed with a CRM data quality program the data will deteriorate at about 2 percent per month.
- There is a six step data management framework to maximize data value.
A CRM data quality program ensures customer data is accurate and complete, secured and available to those who need it, and in compliance with company policy and regulatory requirements.
Data quality is more than data integrity. Data integrity measures the accuracy and reliability of data while quality considers additional factors such as data completeness and value. Information accuracy is a prerequisite to effectively using information in customer facing business processes.
A 6 Step Data Management Framework
The following 6 step CRM data management framework brings structure, measurability and automation to create and maintain customer data quality.
Assess Data Purpose and Quality
The first step is really a two-part process to understand how data is used and to assess existing quality.
To identify the purpose of data you will need to meet with the users and get some initial answers.
- Who are the consumers of data?
- Where is data sourced and shared?
- How is the data used, and what data is needed to complete business processes?
- What data is most and least valuable?
- What data is in most need of improvement?
- What data is most and least used?
It's usually a mistake to overhaul all customer data at once. A phased approach that starts by prioritizing high value data required for essential business processes can make more sense.
Recommended upsell items; inventory availability; prior customer purchases; price lists
Substitute items; inventory availability; prior customer purchases; price lists
Total addressable market; historical sales by geo; account firmographics (i.e., company revenues, headcount, industry)
Company name; contact name; address; customer segment; email; opt-in confirmation
Ideal Customer Profile
Sales history; Contact Persona; Account profile attributes (i.e., company size, industry, geo)
Account demographics; firmographics; Contact behavioral attributes; Digital footprints; Personas; Social data
Lead score; Historical purchases; Online digital footprints; Conversion history; ICP
Contact title; role; industry; annual revenues; employee count; company growth rate; sentiment analysis
The next step is to assess data quality. For most companies, the CRM system is the customer system of record and you can use CRM software tools to identify data accuracy, completeness and consistency as well as detect data errors and anomalies.
It's a good idea to assess master data objects such as Lead, Account and Contact records before assessing transactions such as activities and opportunities.
A data quality assessment is a four-step process.
- Map. You need to understand all the places that store customer data and how they share, integrate or synchronize data. Otherwise, you will find yourself playing whack-a-mole.
- Measure. Data assessment requires measurement along the six data quality dimensions of accuracy, completeness, age, consistency, redundancy and usage. For example, what percentage of lead, account and contact records are accurate and complete? CRM software tools such as audit logs, (CRUD) updates, lists, views and reports can measure data utilization or view last modified dates to see how often certain data is updated.
- Causation. Identify the most common reasons and patterns for poor data quality. You will generally find a correlation between customer data sources and common data errors. For example, customer data collected from lead landing pages, webinar registration forms or whitepaper download forms often contain bogus contact data (especially telephone numbers) because these contacts wish to remain anonymous and not be contacted by salespeople.
- Valuation. Not all customer data is equal in value and some data may have no value. This step will prioritize data and avoid spending time and money where it doesn't make sense. Create a data value score for each data element.
One big caution from my three decades of doing this. The 'more data the better' approach runs contrary to an efficient CRM data program. Collecting customer data just in case it may get used one day generally creates far more cost than value. Excess customer data makes CRM software screens more arduous, leads to information overload and makes otherwise simple tasks such as creating lists, views, dashboards and reports more complex.
Excess data increases system administration effort and data quality program cost. Resist the temptation to amass large volumes of customer data and instead focus on the essential data that will directly impact customer use cases and company objectives.
Implement a Data Policy
Once you have a baseline assessment you can set data quality objectives and create a data policy to apply quality control procedures and internal controls to achieve those goals.
A data policy illustrates what good data looks like, how its measured and how its achieved.
CRM data objectives must balance value with cost. Achieving 100 percent quality data is not cost effective for most companies. A better approach is set data quality goals, often using thresholds, based on the value of data.
For example, customer name is essential and should therefore maintain a minimum of 99 percent quality. Customer industry is important, but not essential, so should maintain a minimum of 95 percent quality while the year the company was founded is either non-essential or designated a lower minimum of 60 percent quality.
Using the CRM data model is the simplest way to designate the value of all customer data in one place.
With specific and measurable data goals in place, the next step is to establish the procedures to achieve the goals. We've previously authored an article on CRM data policy design and benefits with plenty of examples so won't repeat that here.
Enforce Data Quality at the Source
Maximizing data quality is a two-step process of correcting data anomalies at the source and periodic data maintenance after the fact.
Eliminating poor data entry at the source is achieved with a combination of data policy guidelines so users enter data correctly and technology automation to prompt or resolve data irregularities.
Simple CRM software capabilities to prevent bogus data from entering the system include spellchecks, dropdown lists, basic duplicate data detection and routines that automatically populate data fields based on values in other fields.
With minimal effort you can also assign field validation rules to enforce field specific data entry formats. For example, you can avoid too few or too many digits in a zip code, prevent alpha characters in a telephone number or assign other constraints which enforce properly formatted data. Another technique is to create valid data value ranges to prevent users from entering nonsensical values.
With some development effort, you can further reduce data entry errors with fuzzy logic matching for Leads, Accounts, Contacts and other primary CRM data fields.
One other consideration is that when a specific piece of data is needed for synchronization or important downstream business processes, you may want to make that data field a required field when the record is created.
I typically have field level validation rules present a prompt or a queue card referencing the specific data policy rule and preventing the record from being saved until correctly entered. Online prompts or notifications that enforce your data policy guidelines, field level data integrity checks and automated data resolution routines go a long way to prevent inaccurate data entry.
Master Data Management (MDM) is a more advanced software application to eliminate data entry errors and bogus data at the source. It centrally manages shared master data, such as customer records, by consolidating, standardizing, persisting and distributing common data among shared destinations so that it is consistent and accurate.
CRM projects tend to consider customer data at a project or departmental level and overlook other systems outside their purview that also manage customer data. This results in overlapping customer data that delivers different and inconsistent results for the same queries. MDM software enforces valid data entry at the source and synchronizes data among systems so there is one version of the truth. MDM software is generally most needed by large organizations with multiple lines of business, geographically dispersed operations or multiple systems storing customer data.
Review and Resolve Bogus Data
Customer data deteriorates so no matter how well you enforce data entry at the source, periodic data maintenance services are needed to review and cleanse data on a regular basis.
Data hygiene processes often start with exception reports to find erroneous or incomplete records. Another method is to use the filters in your CRM software to identify records with bogus or missing data. For example, you can create views or lists where designated fields have no values (data = null) and then measure the extent of incomplete data by user.
What are often called merge-purge processes are used to eliminate duplicate records.
Duplicate records may stem from slight differences in spelling of lead, account or contact names or other primary fields as well as inconsistent rules (or the lack thereof) across data collection channels.
Most CRM applications have simple dedupe routines. If properly configured they can identify much of the redundant data. MDM systems are more sophisticated for this purpose.
A periodic purge and archive process should also be considered. It's good practice to keep your CRM software clean of unresponsive contacts. So if your email broadcasts report hard bounces for select contacts it's likely those contacts have moved on and the email or maybe the contact should be archived. Eliminating bounced emails also improves your ISP reputation score, thereby ensuring more of your emails do not succumb to spam filters and increasing email deliverability rates.
Data cleansing improves data quality while data enrichment increases data value.
If marketers had certain additional buyer information, they could increase campaign conversions. If salespeople had more customer information, they could increase sales win rates. If product managers knew how much certain customer segments were willing to spend, they could optimize product pricing. There are many use cases that support appending additional data to the customer record. Data enrichment most often acquires data from a third-party vendor.
Your customer insights are the place that will identify the most impactful data to acquire. The data enrichment process is a bit different than the data quality process, so we've described the customer data enrichment process in another post.
Educate, Monitor and Improve
The final step is to make data quality an enterprise-wide program. IT cannot do it alone.
The data policy should designate data owners, data stewards and data consumers. Owners generally have the final say in how data is acquired, structured, used and archived. Stewards, sometimes called custodians, are accountable for data policy compliance and quality objectives. The data steward ensures that data policy processes are understood and followed.
CRM users must understand why data quality policy and rules are needed and the negative impact if they don't do their part. Communication is key to getting users to see the value of keeping accurate records. Once they understand the value to themselves and the business, they are more likely to get on board.
If users repeatedly fail their responsibilities, it's the steward that coaches performance improvements. If such coaching fails, it gets escalated to the data owner.
As with all technology projects executive sponsorship is absolutely required. In the event of slow or challenged user adoption, a change management program may be needed.
What gets measured gets managed. To achieve and maintain data quality we normally use a few exception-based CRM reports, variance alerts and a CRM data dashboard.