What on earth is a CDP?
What on earth is a customer data platform (CDP)? If you are confused, you're not alone. The market also seems to be confused as engineers, marketing execs, CTOs, and CEOs all seem to not only be unsure about what characterizes a CDP but also skeptical of its promises. Forrester’s definition of a CDP is a platform that allows companies to collect, process, and consolidate customer data while offering functionalities for the activation and execution of this data. Sounds like it is a technology platform to turn data into better customer experiences.
Using the above definition, there are a number of steps required to turn the raw data into a better customer experience:
- The collection of the data. First we need to access and move all the data we need into one place. The customers are interacting with our brand over a large number of touchpoints.
- The process and consolidation of the data. Raw data is just noise until we process, clean and consolidate it into a form that the company can work with. On top of that, we need to ensure the quality and the continuity of this data.
- The activation and execution of this data. What does this mean? Mainly that it’s not enough to just collect and store the data, we need to perform actions that will translate this data into informed business actions. To do that we will need to further process the data, load it into different systems and make sure that it is in a form that can be consumed by downstream systems.
Every one of the above steps has unique challenges. Collect data? Well, there are public companies out there with products that are only about collecting data. Process and consolidate? Check the Apache Foundation and try to count the collective effort that has been put into building data processing platforms. Activation and execution? Just count the number of different applications for each available business function out there, now multiply this by the number of possible business scenarios for each company out there.
It is clear that it’s impossible for a company to excel in every aspect of a CDP. This explains why there are so many companies offering a CDP without a definite leader in the market. The complexity of the digital-first world we live in requires specialization, you need to focus on one thing and make sure you do it well.
There is a good reason when it comes to data infrastructure, that we use the term "stack." Most successful implementations of data infrastructure are a stack of different technologies stitched together. Each technology specializes in one activity and does it extremely well. At the end, the whole stack is the right solution to maximize the value you can get from your data.
How CDPs should be built
Here is how we think a CDP infrastructure should be built using the technologies available and considering the challenges today.
Collect customer data
This is the first step and there’s no way you can overlook or bypass it. When it comes to data infrastructure, there’s a universal rule: garbage in - garbage out. No matter how innovative your platform might be, if you don’t collect the right data and ensure its quality, you will never succeed in extracting value from it.
Specialization is important here, building a data collection infrastructure that can scale up and down and offer good delivery semantics is a hard engineering problem. It doesn’t make sense to pay for a whole CDP if you are starting this journey, instead invest in a robust and resilient data collection infrastructure. Building the pipeline is the first step and the foundation for building the rest of the stack.
Process and consolidate the data
Data warehouses have evolved a lot in the past decade. By introducing Amazon Redshift, AWS started a revolution in the space. Today, we have a great selection of amazing technologies that can scale almost indefinitely, need minimum maintenance and have a very predictable cost. Amazon Redshift, Snowflake Google BigQuery, and Microsoft Azure keep innovating at an amazing pace turning a very specialized technology of the past, into a commodity that any company of any size can access today.
The data warehouse is where your data lives. Doing the processing and the consolidation of your data here not only is possible today but it actually can offer a very expressive environment where complex data tasks can be performed, something that is impossible and very costly to perform directly on the event stream.
Using the data warehouse also democratizes your data. SQL is a dialect that everyone inside the company can use, from by engineers to business users. This allows the whole company to interact with the customer data that has been collected.
Consolidating your customer data into your data warehouse has a massive benefit of being able to enrich your customer data stream with the rest of the data you are already collecting. Support Tickets? Marketing data? Lead scores? Touches from Salesforce?
Your organization has a lot more data than you think and linking this to your customer event stream can be the difference between dealing with a noisy data channel and figuring out your next successful marketing campaign.
Data activation and execution
The best way to approach data activation is by finding a specialized vendor. There are many journey orchestration platforms out there that can turn your data into journeys that your customers will love. What is important is to understand that journeys are not universal, each company and each market has different needs.
That is why we see vendors specializing in verticals, i.e., ecommerce and B2B. There’s no reason for your company to compromise on specialization, instead choose something that is built specifically for your market and use case.
You need the whole stack
Working with data is a complex task and the promise of one product that can deliver everything around it is an empty one. A generic CDP solution, promising end-to-end capability is going to leave you dissatisfied in the end. It might look like a good idea to procure one product instead of building a whole stack, but you will soon figure out that you need to use the right tools for the job.
Today, companies that assemble best-of-breed data infrastructure stacks, win.