What exactly does a top-notch customer data foundation look like? To find out, we surveyed and interviewed analytics leaders from a wide range of industries, company sizes, and stages of digital maturity. We found that the end goal is an always-on, comprehensive view of the customer that connects all data from all systems, at scale, and feeds the entire organization with the insights, segments, and data it needs.
Why did we go to the analytics leaders to find this out? That’s because they’re the ones most often charged with figuring out ways to not only manage a huge volume of customer data but also to harness it for rapid business growth (cue image of an analytics professional surfing a tsunami of data while sipping a glass of champagne).
Analytics leaders are bilingual: they’re fluent in the lingo of year-over-year business growth, and they also speak the language of technology. They know the levers to pull when driving growth and they have a keen sense of what’s technologically possible now as well as where the tech is headed in the future.
For these reasons, it’s increasingly the job of the analytics leader to inform and shape the right data foundation for the entire organization, in partnership with IT stakeholders and business leaders.
But as one analytics leader from a Fortune 500 retailer put it, “We can all see the top of the mountain and we all know exactly where we want to be. But it's really hard to break that down into a series of steps for how we actually get there.”
The analytics leaders we talked to focused on 6 key areas. Here’s a summary of what they said:
1. Data needs to be complete, and how you co-locate it matters. Top analytics leaders eschew rigid databases and ETL (extract-transform-load), preferring raw ingestion and scheme-free environments. This provides the flexibility to keep their data foundation up-to-date as new sources are added and data structures change. It also helps bypass a lot of manual data prepping that analysts are often stuck doing before they can generate insights.
2. Records have to be unified, and deterministic identity resolution doesn’t cut it. AdTech vendors and advertisers have known for years about smarter, more flexible probabilistic approaches to managing identity and increasing targeting reach. The rest of the organization, however, has been stuck trying to build a fully deterministic view of customers through loyalty programs and other types of identity capture. (Quick and dirty definitions: deterministic goes for exact matches, as in, email address equals email address therefore it’s a match; and probabilistic uses machine learning to find likely matches across many types of data even when they don't match perfectly). The problem is, there continue to be treasure troves of customer information that can’t be deterministically linked together, even with the most carefully thought out programs and business rules. Forward-thinking analytics leaders are turning to machine learning to get a more complete and accurate view of their customer base.
3. A single version of identity isn’t enough. This one sounds a bit mind-bending, but stick with us. If you embrace the messiness and complexity of customer data, you’ll quickly see that brands need multiple views of customer identity to support their full range of use cases. For example, you need a probabilistic view to gain a true understanding of your customer base, in order to answer even the most basic questions like “how many total customers do I have?” This view should be balanced between precision and recall (read the full report for more details on what this means) and is optimal for many types of personalization use cases like product recommendations and segmentation. However, with operational use cases like sending receipts you actually do need a deterministic view of customers. Every brand has this today. Finally, some brands might want to have a third view tuned to maximize campaign reach for use cases like targeted advertising and lookalike campaigns on Facebook or other platforms.
4. Brands are hungry for customer insights. Bringing together data and unifying it is a great first step. But analytics leaders are also tasked with finding usable insights from all of that data, for use by a variety of teams. Most analysts are baking intelligence into their data foundations, with predictive models and custom attributes built for marketing and analytics use cases.
5. The right data foundation helps the entire organization. Having a smart, unified view of customers isn’t just an analytics problem. It’s critical for marketers, advertisers, customer support, and even merchandisers and product development teams. In the last year or so, it’s also become a key component of compliance use cases like the right to be forgotten à la CCPA and GDPR. For these reasons, analytics leaders are building hyper-connected customer data foundations that support a wide range of use cases.
6. Your data foundation is a circle, not a line. Knowing your customers, making better decisions, and delivering more seamless and personalized experiences across all channels isn’t something you do once and forget about. You need to create a feedback loop from customer insight to segment to experience and back to customer insight. Top analytics leaders are creating connected data foundations that they can use to consistently tune their strategies from end to end.
For each of these areas, the details matter. That’s why we wrote up an in-depth report that dives into all the considerations outlined above, with how-to insights, quotes from top analytics leaders, and survey results, all of which will help accelerate and inform your creation of a best-in-class customer data foundation. Download the full report now.