Round table recap: Customer data platforms

A recap of the Acro Commerce round table discussion on Customer Data Platforms (CDPs). Attendees got the chance to learn more about CDPs as well as get detailed answers and agnostic advice from our subject matter experts. Read on to learn more.

The basic definition of a CDP:

What is a customer data platform (CDP)?

• A packaged software that creates a persistent, unified customer database that is accessible to other systems.
• Gives marketing teams a complete 360° profile of your exact customer that is dynamically updated and statistically unified.

How does a CDP work?

• A CDP provides a way for all your systems to deposit data into one centralized place where it can be stored, analyzed, updated and enriched for your teams to use.

Planning for the implementation of a CDP is best done in a 3 pronged approach:

1. Data architecture and mapping: How do I get all my different types of data to play together, and how do I plan for that?
2. Administration and application of the CDP: What do we want to get out of the CDP? What do we need the CDP to do? How do I administer the CDP from a marketing perspective?
3. Fine-tuning and execution of the marketing automation element: How do I show one set of products on a homepage to one customer and a different set of products to another? How do I flex the CDP insights into my marketing email? How do I upsell and cross-sell on my ecommerce platform leveraging the data from the CDP?

Sometimes, companies like to jump too quickly to the 3rd step and want to immediately use the CDP to market to the customer and personalize their ecommerce. But to do any of that effectively, your data architecture needs to be considered and thoughtfully planned out and implemented. Once you bring your cleaned data into the CDP, and then the marketing automation of the CDP can be executed. The key to finding success quickly with a CDP will depend on how well your data has been cleaned and organized and how well you have planned what you want the outcome to be from the CDP.

It is also important to note that there is a time for a CDP to be introduced into a business. You have to make sure that you have enough data to warrant using a CDP and that you have the time and company energy set aside to organize that data. This can be done by either working with a consultant or a team or having an employee or team in-house to get things ready to go shop for a CDP. The next step is organizing and implementing the data architecture to bring the CDP online. Once that is done, the CDP is ready to use as a marketing tool.

Questions to consider before buying a CDP:

Does the data match across the systems that I want to put into the CDP? If not, can I make the data work, or do I know what I will scrub out?

Before getting ready to buy a CDP, companies should be prepared to go through an in-depth internal data architecture mapping exercise where you a) define your segments and data points in each system that is going to contribute data to the CDP and then b) map out those data points and segments, figuring out how you want to use that data and what you want to gain from the CDP.

If you go into a project to implement a CDP, having already bought your license, but haven’t taken the time to map out the architecture or how you will use the data, you will probably be paying for that license for 12 to 18 months before you can put the CDP into play.

Do I have the manpower to clean, organize, and use the data?

Do we have a business analyst on staff? Do we have software programmers that can clean the data for us? Do we have marketing experts that can use the insights from the CDP to build a better strategy?

It is important to note that most companies looking to invest in a CDP will have some, if not at least one person in each of the abovementioned roles. But if not, a solid option would also be to look into hiring an ecommerce consulting agency to help you with planning, strategy and manpower.

Specific questions from the round table

Question: What is required of a company to be organized and in a place where introducing a CDP can be effective?

Answer: At a very base level, a company must have enough data from different systems that they want to amalgamate. A CDP is used to connect, store and amalgamate multiple sources of data, such as website analytics, analytics from ads, social platforms, marketing platforms, CRMs, ERPs and ecommerce platforms. Then, once the CDP has that data, it gets cleaned and unified into customer profiles which are then put out to other platforms like digital marketing platforms or are used in the personalization on your website.

For example, in a customer service portal, you could pull up the CDP profile on a client and get up to date, accurate information on that customer's combined history from all the data sources to be able to give the best customer service or serve up personalized product suggestions for upselling and cross-selling opportunities.

Question: Is there a general format for data in a CDP? Or do you go out and find the one you want to use and then do you architectural mapping based on that CDP, or is it universal?

Answer: There are aspects of most CDPs that are universal, as in customers are similar and most systems have similar generic data points like client name, address and phone number for example. Where it can get difficult is when there are multiple variants for the same data across different systems. The key is figuring out what data points should map across your systems.

For instance: if you have multiple systems that are collecting a data point for “region”, it is important to figure out which system takes precedence over the others. Depending on what you pick for a CDP, some of that may be set up automatically with integration support out of the box. But, you can get away from “out of the box” pretty quickly when adding in your business’s unique systems and applications. The more systems you need to put into the CDP, the more important and intense the data mapping exercise becomes.

Question: From what you described, you put everything into the CDP and then there are different layers, like a raw data layer, the cleaning and stitching and whatnot, and then you have some intelligence? Or do you do all of that cleaning before, and that scrubbed data then goes into the CDP?

Answer: It depends on the CDP, but usually, you try to do as much cleaning as you can before you put the data in so that you’re not putting in any junk. You can then scrub the data again once it is in the CDP once you have combined data from all your sources. As an example, there will be a data cleanup and normalization process in the integration phase from each system into the CDP. Then once the data is in the CDP, you will merge all profiles and remove duplicates.

The process for getting your different types of data to correlate to each other can be technically complex, but as part of the mapping exercise, these connections will become clearer and then the integrations will become smoother with the help of developers. The idea is to put as much validated, clean data as possible into the CDP and then determine what you use afterwards by turning reporting on or off for data. That way the data is always there, in case there is a use for it. Data can always become useful in the future so it is a good practice to have it warehoused.

Question: If the architecture and cleaning are done in-house, then we send the data to the CDP, it seems like quite a lot of the work is done before, so what is the magic of the CDP?

Answer: A CDP is 2 things:

1. The data layer underneath, which can be as simple as complex as your database setup needs to be. If you have multiple systems, the CDP acts as a central repository for all their data.

2. The interface for managing and combining the data and putting it all into unified customer profiles. The “magic” comes from machine learning and algorithms that are specific to each CDP that make data link up.

The point of a CDP is that you have this unified customer data available to aid any type of decision: Marketing, sales, products, ordering, inventory; any business-related function that depends on an accurate, unified, and dynamic customer view.

A CDP will do the intelligence for you after the data is in the platform. It brings together multiple pieces that can potentially categorize a user into predefined segments, industries, sectors or other identifiable groups, and then it can serve up different experiences for that user based on what you have set up as the variables for that group.

Some CDPs also tout a machine learning capability that will search your data and come up with predictions for appropriate segments, industries and verticals without your intervention. When talking to CDP companies individually, they will focus on the sexy analysis and marketing automation features of the platform, but the key to getting the most out of the CDP is making sure that the data going in is nice and clean.

Question: How is a CDP different than a CRM?

Answer: A CRM is a functional tool that tracks your sales cycle and a CDP is much more on the marketing automation side.

When a CRM has a marketing automation portion, it is generally only using the data from one source, the CRM and your sales cycle. But a CDP will use data from all the connected systems to glean a better data set and provide better, personalized marketing automation for your website.

Do you have more questions about customer data platforms? We want to hear them!

The decision to bring on a CDP is not one to be made lightly. We want you to have as much information as you need about what exactly needs to go into that decision. Our team of experts can help with anything from your data architecture mapping exercise to actually meeting with CDP vendors with you and offering agnostic advice during the purchasing process. If there are any questions you have that may not have been touched on in our discussion, please do not hesitate to reach out to us at any time. Our ecommerce software consultants are always here to help with advice and suggestions.