Taivara Blog Post by Dave Cherry & Roy Slagle
Building a High-Performance Customer Data Platform
Why building a single view of the customer is so hard…and how to make it easy
As a retailer, imagine a scenario where a customer needs assistance.
Regardless of the contact method they choose (email, call center, social media, online chat or store associate) and using whatever identifying information that they choose to provide (mobile number, order ID, loyalty number, social media handle, physical address), you are able to confirm their identity, access their interaction history (orders, visits, issues) and efficiently serve them and exceed their expectations. This requires a
Because the customer has given us all the information asked of them or what they prefer to share, they expect the retailer to act like it knows them. If an email is clicked, the next text message sent should reflect that. The customer is dealing with a single brand, and all elements within that brand should act together as a single unit.
Sound impossible? Sound like a dream? Not quite.
What it actually sounds like is the baseline expectation of today’s retail customer – an expectation that most retailers are unable to meet.
Need a High-Performance Data Platform?
Sometimes you don’t have the luxury of time when building the next big thing. If you have an immediate problem to solve, we’ve found that bringing a cross-functional team together to tackle the problem head-on yields positive results fast.
Why is it so hard?
Achieving this single view goal state is very difficult for two primary reasons.
First, customers are dynamic – characterized by constant change. They change relationships both within their households and with retailers. They adopt multiple new physical and digital identifiers (usernames, social media handles, etc.) and change old ones. They change addresses – both email and home. And many use different combinations of the above data elements to establish their identity with retailers – often forgetting which they’ve used with whom. They also vary how and when they engage with retailers. McKinsey recently reported that more than half of all customer interactions involve multiple events and channels during their journey – and an outstanding, memorable experience must be seamless and last throughout the entire journey.
This causes a variety of out of date, conflicting, or missing data elements within the
Second, retail legacy systems were not designed with integration as a top priority. Each system – merchandising, marketing, customer service, Point of Sale, eCommerce, Fulfillment, Loyalty, and more – were designed with a specific purpose. And with a specific data model schema intended to efficiently fulfill that purpose. Connecting and linking customer records across systems at best was a lower priority – if a priority at all. This means that “customer ID” in one system may not align to “customer ID” in another system. This created an opportunity for IT to “solve” the issue through asynchronous interfaces, costly ETL processes and duplication of data (which often was out of synch) – with data mapping specifications that often went undocumented.
Now combine these two reasons and the resulting complexity and difficulty
How to make it easy
Let’s start with what is not easy – starting over. Replacing or upgrading legacy systems is expensive in both time and capital. And despite what many software sales reps may claim, there is no single, simple customer data platform that both integrates legacy data structures and allows for continued ongoing change.
But there is a reasonable approach with manageable investment and a much shorter timeline to benefit realization that can work for any retailer independent of their legacy environment.
So what are the keys to making it easy? Decision Modeling, Customization, and Iteration.
First, start with an important business decision or action – just one. Use a decision modeling approach to outline the key inputs, calculations, thresholds and criterion for determining an appropriate course of action. Following are just a few of the many
- Operational Efficiency (e.g. reduce contact center call volume and duration while increasing first call resolutions)
- Pricing/Promotion Effectiveness (e.g. understand elasticity and deliver personalized offers based on customer sentiment or behavioral insights)
- Customer Engagement (e.g. build loyalty to increase traffic or leverage influencers through contextually relevant personalized value offerings)
Next, understand your unique legacy environment and design a custom approach that maps similar data elements from different sources into a data model that a non-technical business leader can both understand and manipulate. Once this new data “fabric” is designed, feed data into it and use the resulting information to feed your operational processes and analytical models, delivering the data and insights needed to inform and improve decision-making for that specific action.
Start small and build a simple model that works. Then add to it. Then do it again. Iterate on the initial use case to optimize efficiency and accuracy and then move onto the next use case – which may very well leverage many of the same data elements. Continue to iterate to improve existing models, adding more relevant data that improves decision-making.
Do you need help building your own customer data platform, evaluating tech providers, or integrating with existing systems?
What else does it take?
Now that the game plan is understood, there are just a few additional ingredients that are needed.
To be successful, you must begin with a commitment to customer-centricity – an unwavering focus on prioritizing the customer and delivering the products and services that she values easily. Retailers must put the customer first, ahead of product. This represents a significant mindset shift in this traditionally merchant driven industry. Senior business leaders must be committed to this strategy.
Next, you will need an enabling technology. This challenge cannot be solved simply with upgrades, coding or other patchwork ETL/interfaces. But as stated above, this doesn’t mean new systems across the business and it also doesn’t mean implementation of a data lake (or similar data aggregation capability). The enabling technology that is necessary is a Customer Data Platform (CDP).
In 2013, marketing technology expert David Raab identified a segment of marketing platforms he coined CDP. Leveraging dynamic capabilities such as graph database schemas and a simple non-technical data mapping GUI (making the data accessible), a CDP permits a retailer to maintain legacy
There are about 2 dozen products that fall into the CDP category. Some key vendors include:
- Signal – a people-based platform
- Treasure Data Enterprise CDP
- BlueVenn which offers a multi-channel marketing automation toolset
- Segment – an API platform that allows non-developers to consume customer data from many disparate systems, and many others.
Implementation of any of these technologies takes patience, commitment, flexibility
There are several important factors to consider when evaluating CDP products – the first of which is whether to build or buy. Building in-house may make sense if your business has very unique customer data or marketing needs, but you’re likely to get more bang for your organizational buck by carefully selecting a vendor product. The spend on a custom software build does not stop when the product is built as it will require constant care and feeding throughout its service.
Data is at the heart of any CDP implementation. While the product’s sales pitch may present some very enticing insights into customer behavior at a fictitious retailer, you will need to perform some careful due diligence to ensure those same insights can be gleaned from your customer data. Most companies have disparate systems, multiple systems of record, and “dirty” data to contend with. Special consideration should be given to the following within your walls:
- What data will you require at what stages in order for the CDP to provide
- Where does that data currently
How difficult or possible is it to gain access to that data in real time or in scheduled batches.
- As custom development will likely be needed, what is the capacity and timeline of your IT organization to complete that
- How often does the structure of the source data change and what ongoing work will be required to update data
- What internal systems may make use of your newly aggregated CDP data and how can those systems ingest that data.
Work closely with your IT department during your due diligence to increase the likelihood of a successful implementation.
As a retailer, we know that you may also need assistance. Cherry Advisory and Taivara bring a depth of experience in retail innovation, advanced analytics, customer experience and technology deployment that can help you navigate and achieve success in attaining the new holy grail of retail – a single view of the customer
Executive Advisor, Cherry Advisory, LLC
Dave Cherry, Executive Advisor with Cherry Advisory, LLC, provides strategic guidance and executive counsel on retail innovation, customer experience and advanced analytics. He currently serves on the International Institute of Analytics Expert Panel, Women in Analytics Advisory Board and CBUS Retail Advisory Board. Dave holds a BS in Economics from The Wharton School at the University of Pennsylvania.
CTO & Practice Leader, Software Design & Development
Roy is the leader of Taivara’s Software Design & Development practice. He helps companies design, develop and launch innovative solutions for internal and external customers. You can reach Roy at https://taivara.com/roy