Customer data platform sit firmly within the realm of customer data management, alongside traditional approaches of customer databases and Customer MDM as well as emerging datalake approaches:

1. Customer Databases which have traditionally been used as staging areas to prepare data for Marketing Automation campaigns
2. Customer MDM which have traditionally been used for cleansing and generating customer golden records
3. Customer Data Meshes and Fabrics, emerging domain driven data management approach for datalakes
4. Customer Data Platforms (CDP), emerging platforms focused on preparing and activating customer data

We find that often the line between these approaches and platforms end up being blurried causing implementation anti patterns, significantly reducing the value of the implementation.

‍#1 — Using a CDP as a Datalake

Due to the way Customer Data Platform have evolved, some of them have incorporated features traditional from a datalake to handle some of the process and management of customer data. Some of these solutions replicate the same infrastructure as traditional data platform, leveraging technology like Spark under the hood.

We have seen some of our clients leverage these data platform to directly collect customer data, query or analyse them or replicate the same type of data management that happens in their datalake.

We have seen our clients undergoing that path create data silos’, be faced with significant bills to pay, be forced to replicate customer data onto multiple systems increasing their cyber attack radius and make it more difficult to manage or be faced with a slowness to experiment.

In addition CDPs solution are typically not at the forefront of datalake innovations, and offer more restrained capabilities than pure data platform solutions such as Databricks or Snowflake for instance. Composable CDPs solutions exists that can directly connect to these data platforms solutions and make them actionable without creating any additional integration complexity. Some of the more traditional Customer Data Platforms have also started going towards that route offering “Bring your own lake” functionalities.

#2 — Using CDP for reporting on customer data

Talking to some of our clients that have successfully implemented Customer Data Platforms, we often see that attribution to some of their customer knowledge is directly attributed to the presence of a CDP.

“We know this because of the CDP”

For some of our clients, the CDP program is the first time marketing and technology teams have gathered together to collect, model and report on customer data, at least to that extent. Most of the effort in setting up the CDP is spent gathering the data across touchpoints, working on defining the deduplication rules, harmonizing the data and doing data quality assurance. For some of our clients the implementation of the CDP is the first project that forces them to do this.

While some CDPs offer a set of pre-built reports, these often do not offer the flexibility as fully fledged reporting tools like Tableau or PowerBI coupled to a general data platform. In most of our clients’ implementation the deduplication rules tend to be fairly simple and could have been easily implemented as part of a query or an ETL process for reporting and analysis, however the CDP implementation program is often the driver for putting the effort of “IT” and “marketing” together.

#3 — Using CDP just as an orchestration tool

One of our clients had an extensive datalake offering real-time processing and integration onto their Marketing Automation platform. In order to take advantage of their customer data and move towards omni-channel marketing, they decided to deploy a Customer Data Platform to orchestrate journeys across touch points.

The client wanted to be able to leverage a marketer friendly interface to provide decisioning and journey mapping capabilities for omnichannel campaigns, however the implementation proved of limited value as integrations needs and limitations such as API calls restriction of downstream systems limited some of the use cases and with some of the data being replicated to multiple systems (CDP, Marketing Automation for instance).

Most marketing automation system offers some decisioning capabilities and omni-channel communication capabilities, although sometimes at an added price tag, they offer an initial way to bootstrap customer focused communication decisions. In our view, more complex decisioning capabilities should sit within the center of the enterprise and provided an interconnected tissue of integrations to the systems powering the company, such as Inventory Management, Order Management and CRM for instance.

#4 — Not using the right data models for their CDP

We have seen some of our clients not taking full advantages of the capabilities of the tool they purchased, due to the wrong data modelling approach or decision.

We have seen some of our clients rollout CDPs with dimensional / relational data model capabilities, but only end up loading customer aggregated data onto the tool, severely impacting the flexibility and potential usage of the tools.

Some of our clients have been struggling to maintain some of the data models they had implemented, for instance one of our clients faced issue updating their data model with additional attributes due to time outs from the UI.

Different types of CDPs have different approach to data management, some use flexible and dimensional schemas, some are working more on a semi structured and event driven approach. During implementation it is therefore important to understand their approach to data management and limitations.

#5 — Layering Multiple CDPs

Having worked extensively on the Customer Data Platform space, WiseAnalytics.io has seen a lot of implementations of CDPs. Sometimes multiple implementations at the same client, sometimes multiple implementation of CDPs as part of the same program.

Some of our client saw another CDP as the best way to tackle the limitation of their chosen CDP, chaining them together. Customer data platforms tend to have major focuses on different areas some are better at data integration, some are better at data management or customer deduplications. This type of pattern is far from being an exception, and a number of CDPs have native connectors to other CDPs.

However, these native integrations are sometimes not available or used. Before we came in, one of our client, was pumping data from their data platform onto a first CDP, using it for customer deduplication, exporting it back onto onto their data platform, remodelling their data and re-exporting back onto a second CDP.

These chaining of CDPs and steps significant increase latency and render the entire data integration chain fragile and should be best avoided. Most often than not CDPs are not a best of breed tool for most purpose, they provide an integrated solution to the management of first party consumer data for marketing purpose but are often not best in-class solutions to specific problems. A CDP should never be the solution for another CDP’s inadequacies.

#6— Using CDPs as a MDM

Some of our clients have been looking to use CDPs as Customer MDMs. There are a large number of difference between the implementation, latency and deduplication features between CDPs and Customer MDM.

1. CDPs are traditionally built for enabling real-time identity resolution, although it is not the case for all the solutions marketing themselves as CDP
2. The identity resolution and deduplications features of CDPs are typically not as involved as CDP
3. CDPs mostly have a focus on mutable data rather than immutable data
4. Most Customer MDM offer advanced features for Data Stewardship, allowing to manual treat customer matching exceptions and merge profile manually

CDPs are not meant to fully replace Customer MDMs, they do not offer the same set of functionality and do not cover to the same extent Identity resolution and deduplication. They do however offer summary capabilities, which might be sufficient for a number of marketing use cases.

#7— Everything being a Single Source of Truth (SSOT) / Golden Record

Most of our client still firmly believe in the single source of truth “Golden record”, and see the customer data platform as the tool to help build and act upon the Customer SSOT. This belief has been firmly ingrained through decades of defensive data management practices and marketing around the CDP.

However, in the age of data privacy, regulation and customer intimacy, a single source of truth is often a limitation to make the best usage of your available data. We strongly believe that firms should look to embrace multiple versions of the truths (MVOT) to get the most of their first party customer data. MVOT are versions of the truth build from an initial SSOT embedding context, relevance and purpose. An enterprise might for instance wants to leverage a different set and consolidation of information for Marketing Automation campaign than for Media campaigns.

MVOT is however, not something that is well supported by most Customer Data Platforms. This is an area where we have seen our customers leveraging composable CDPs to be at a great advantage on, being both better supported in terms of offloading the business logic and transformations to build upon MVOT to an external system (typically a datalake / data warehouse) or in terms of their pricing models.

#8— Relying exclusively on CDPs to activate their customer data

Most of our clients have implemented CDPs to empower their Marketers to make the best use of their customer data and be able to activate their data across different channels and destinations. CDPs are usually seen as the one central point for a company’s customer data where everything is being exported upon.

After having implemented off the shelf CDPs acting as an integration hub, we have seen some of our clients been faced with significant decrease in match rates at some of their media destinations, being forced to postpone campaigns due to contractual audience limitations or been unable to unlock use cases due to data integration latency.

We recommend to our clients, to use the data integration capabilities of off the shelf CDPs, but to not exclusively rely on their capabilities. Not all the connectors of customer data platforms perform the same and they often tend to be a black box to the end user.

# Conclusions

At WiseAnalytics, we help companies make the most of their data. We are firm believer in using the right tool for the job and in taking a pragmatic approach. We see CDPs as great tools to empower marketer to leverage their companies’ first party data and “sweat” their data assets. However most CDP implementations severely hinder companies’ ability to make the best use of their data and can cause operational maintenance challenges.

We see the future of CDP lie within the composable space, allowing to better support regulations, while providing a faster pace of innovation through a best of breed approach.

CDPs should be part of a modern company landscape, the difficulty is however to make it fit within a company’s existing and future architecture while allowing marketers and the business to operate at the speed they need.