With so much emphasis on analytics for business decision making these days, I often find myself reminding marketers about the importance of the quality of the data they are using—not just the quantity.
To develop data-driven insights, marketers first need to address the question of what data is valuable to their enterprise. Are you using all the data available to your organization about your customers? Are your data up-to-date? Are disparate databases linked together to give you one view of the customer? Have you filled in missing data with high-quality third-party data?
For many, just getting access to current and comprehensive customer data is a problem. Having the time and resources to clean, enhance and organize data to derive actionable insights in a timely manner can seem like an analyst’s pipedream.
Analysts and researchers know that the quality and coverage of data are the biggest factors driving the accuracy and effectiveness of analytics. What kind of model, which outcome to measure, what software to use—all are important considerations, but experience and testing show that the biggest determinant of good analytics is good data. And the biggest barrier to ensuring quality inputs to our decision-making process is corporate culture.
This is where the C-suite must step in. Organizations need to adopt a data-driven culture. The data geeks and modellers know what to do, and the marketers know the questions that need to be answered. But it’s only corporate leadership that can tackle the barriers and resistance that stand in the way of assets being shared across an organization. Enterprise policies regarding data gathering, storage and access are intended to protect consumer data, but they also must ensure compliant and responsible use by those who access the data. Sharing across silos is essential to understand cross-sell and upsell opportunities, but too often the left hand of an organization doesn’t know what the right hand is doing.
Contrary to consumer mythology, there is no such thing as a “digital” customer as distinct from a “traditional” customer anymore; there never was, actually. There’s just “the customer” who researches, shops and engages with brands in many varied and complicated ways. Why do we have digital marketers and CRM managers in different divisions of an organization? Digital agencies, email providers, DM providers, media buyers—too often, these specialists work at cross purposes or, at the very least, collaborate only on small subsets of the total picture. Data integration leading to one view of the customer is essential to good analytics.
These challenges and imperatives are top of our agenda at Environics
To create good quality data, our approach is to let the data speak: to know what can be modelled reliably and what can’t. We have developed a series of best practices over the years and are happy to share our methods with users in great detail. Just ask us. We never produce an estimate for a postal code when we feel we don’t have sufficient data or a reliable scientific method for creating it. If our experts won’t approve the methodology, we simply won’t release the variable.
With limited resources, many Canadian marketers want to focus their time on analytics—not on data gathering or data prep—so we continue to identify new sources and new methods to ensure that:
- The demographics are as comprehensive as possible, despite recent census challenges;
- Disparate media measurement sources can be used through the common lens of PRIZM, our segmentation system that uses demographic, economic and geographic models; and
- These high-quality data can be easily accessed for custom areas and easily integrated with customer and transaction data.
It’s a big job. Nearly a third of our staff of 100 employees is involved in data production—even more right now as we prepare to release our annual data update this March. We do not believe in quick-and-dirty approaches or just repeating processes from one year to the next without review and innovation.
The world of consumer marketing is changing: the talk of disruption and constant change is not just talk. So whether you’re thinking about your own data or data from third parties that enhance your data, remember the objective of predictive analytics is to simulate reality as best you can in your models and processes.
While best practices are essential, the foundation for good analytics is the highest quality data possible.