By: Brandon Ray, Principal Solutions Engineer, Oracle Marketing Cloud | Dan DeZutter, VP – Solution Engineering, Oracle Marketing Cloud
As organizations grapple with growing customer expectations and rapidly expanding datasets, Customer Data Platforms have emerged to help provide a link between disparate touchpoints and channels. The CDP Institute defines a Customer Data Platform as “packaged software that creates a persistent, unified customer database that is accessible to other systems”1. Although this definition helps to clarify what a CDP is, it fails to explain some of the additional use cases that a CDP can enable or how organizations derive tangible value from a CDP.
Implementing a CDP is a complex process and companies often struggle to realize near-term value after the implementation. Organizations too often focus on a “big bang” implementation approach as opposed to utilizing an iterative process to derive value quickly. While CDPs can absolutely deliver significant value in the long-term, there are many ways to achieve quick wins right after go-live. Below we’ve detailed 5 steps to help you maximize the value of your CDP as quickly as possible and some additional actions to derive even more value.
The first and most important step to quickly get value from a CDP is to clearly define your use cases. All organizations want to drive more revenue or increase profits, but your initial implementation use cases must be specific, granular and clearly defined in order to create measurable results and impact. Use cases can vary widely but some common examples include more complete customer insights to drive personalization, more sophisticated segmentation based on combined profile, behavioral and transactional data points, or utilizing artificial intelligence models to define next best actions for customers. Every use case should be linked to an anticipated, measurable metric to track return on investment.
Data is the key to unlocking value with your CDP. As you define your use cases as discussed above, it’s important to understand where you will need data integration, including what sources you want to pull data from and what downstream activation channels you’d like to utilize. As an example, if you want to use an AI model to display a personalized offer on your website for visitors, you’ll likely need data from your CRM system for your customer profile, your website analytics tool for website behavior, your marketing automation platforms for previous promotion history and your POS for prior transaction history. Typically, CDP’s will also ingest third-party demographic / psychographic information to enhance zero- and first-party data. For execution, you’ll need an API integration from your website into the CDP for the personalization and potentially an integration into your content management system or web personalization layer to display the content. You’ll need to weigh your ROI defined in step 1 against the complexity and effort of the integrations needed in this step.
Now that you’ve defined the sources that you’ll pull data from to ingest into your CDP, you need to determine the data transfer method and what specific data you will ingest. It might not be necessary to ingest every single data record or attribute from a source system to accomplish your defined use cases, and this is a common mistake that organizations encounter as they try to ingest excess data that will not be useful for their goals.
In terms of data transfer methods, CDP’s typically can ingest data through batch, API or streaming. You’ll need to determine the applicable integration method for each dataset based on the availability of the data and the frequency in which each data point needs to be refreshed (real-time, near real-time or some batch cadence).
In the example above for personalization, you’ll likely need to ingest PII data from your CRM and POS systems, specific category/product browse behavior from your website or mobile app (historical and current session), previous email sends/opens/clicks from your email platform, direct mail promotion history, and order header/detail information from your POS system(s). The current session behaviors would likely utilize a real-time streaming method of data transfer, while some of the other sources like CRM profile data or direct mail history may utilize near real-time API or batch processes as they likely aren’t refreshed as frequently. You must be deliberate about what data is necessary and how data will be utilized for your use cases to ensure you are not wasting time, money, or energy on unnecessary integration work.
One of the primary benefits of a CDP is creating a consolidated view of your customers across all your systems. To create this unified view, you’ll first need to decide how to match multiple data records together and what data sources are the most reliable.
When matching your records, it’s important to understand the logic behind why two records are the same person. Depending on your data, you may want to match on email address, phone number, physical address or even internal identifiers like CRM ID or Loyalty ID. You may even want to create matching logic that uses a combination of identifiers for more accurate results. Utilizing these types of identifiers will help you match records together from the same or different source systems.
Next, you must create the single customer view (often called a “golden record”). In the example above, you may have customer data in your CRM system and your POS system. If addresses, emails, or phone numbers are different, then which is the most reliable data source and what system has the highest quality data? When creating your golden record, you can set matching criteria to favor data from systems with more reliable data or favor records that are more recently updated. Furthermore, you may choose to favor specific fields from different source systems (e.g., your CRM may have a more reliable phone number than your POS because the customer has called, but your POS likely has a more reliable address than your CRM because the customer has items shipped to their address).
If your primary business is B2C or B2B, then make sure your CDP can do matching and merging of profiles or accounts respectively. Every organization will have their own requirements or expectations for unifying and merging data records, so it’s important to be thoughtful about how, when, and why data is combined to ensure the results meet your use case requirements.
Now you’ll be able to execute against your first few use cases and harness the power of your CDP, plus you’ll be able to measure their overall impact against your KPIs. Throughout these steps, you’ll likely have identified additional use cases to implement – it is important to document these during the process and then prioritize them for the next round of execution. In doing so, you can continue to build on your initial success and momentum with additional wins as you expand your data set and execute new campaigns across additional channels. Campaign results should also be fed back into the CDP to create a persistent feedback loop, so campaigns become more intelligent and relevant as you drive towards your ideal outcomes.
In addition to the above steps to derive value quickly, it’s important to continuing setting new goals to utilize other functionality within your CDP. There are a number of other features your CDP should provide that can open up additional use cases and value – we’ve outlined a few of those below:
Data consolidation is one major use case, but it’s really only the first step in harnessing the extended capabilities of a CDP. CDPs should provide some additional intelligence and create new value beyond just data consolidation, and calculated fields are one example of that in action.
Analytics or data stewards may have standard definitions for calculated fields (sometimes called “Intelligent Attributes”) such as lifetime value, cross-channel engagement metrics or other important KPI’s that your organization utilizes for marketing or analytics. These should be created in the CDP so they can be centrally utilized across your organization for decisioning, not only for marketing but also for extended use cases in sales or service teams for example. You may also be able to create new calculated fields were previously impossible when your data was fragmented across systems. Typically, these calculated fields make it very easy to understand customer trends at a glance – for example, a marketer can easily segment a group of customers based on their calculated total lifetime value or the recency of their digital engagement, or a sales or service member can see if an individual is a high value or low value prospect based on calculated fields that combine profile, behavioral and transactional data points. While these calculated fields may seem simple, they can deliver quick wins for your teams by providing meaningful insights with data that was previously inaccessible.
Calculated fields only scratch the surface of the intelligence that a CDP can provide. By consolidating relevant data in one location, artificial intelligence models can access an expanded number of data points to help automate outbound customer efforts and lead your team towards data-driven results. Artificial intelligence models help to reduce the guesswork associated with right time, right channel and right content messaging.
Some examples of AI models include predictive models like next Best offer or next best action, recommenders like channel or campaign recommenders, or value-based analytical models like customer lifetime value. Additionally, many CDP’s offer the ability to create and train a model within the platform. If you have a custom model that you’ve developed outside the platform, those can also be ingested and utilized for your specific use cases, leveraging the expanded data set that has been consolidated in the CDP. The ultimate goal is to power your campaigns and outreach with data-driven decision making and help ensure that personalization happens at scale for all of your target customers.
Often, companies are limited in their segmentation strategy (either creating strategic segments or modifying strategic segments regularly) because they don’t have a single location where the data is consolidated, or there is an over-reliance on technical teams to query data or create audiences because their data is not in marketer-friendly environment. Without a CDP, segments are often based on singular sources of truth, such as an email marketing system that only has limited visibility into customer profiles and email-related behaviors. Now that you have a wealth of data within the CDP, you can enhance your segmentation strategy with a greatly expanded set of data points to be highly personalized and deliberate in your outreach. This could include segmenting customers based off consolidated profile information, real-time behavioral activity or consolidated transactional data. You can also co-mingle first-party data with second- or third-party data to ensure segments are as relevant as possible.
Typically, CDPs are utilized for high-level segmentation strategy across all channels because you have the most data regarding your customers. Once segments are sent to downstream execution channels, you could weave in additional criteria or filters based on logic defined within that specific channel, all while retaining the broader cross-channel strategies to ensure the experience customers have with your brand is consistent and highly relevant.
We have seen a number of organizations completely change their approach to data, personalization and overall customer experience thanks to their Customer Data Platform. By taking a deliberate implementation approach focused on measurable use cases, teams can start to see positive ROI quickly and iterate on success rather than waiting for everything to be solved at once. Hopefully these initial 5 steps and subsequent expanded use cases were helpful thought-starters as you begin your own CDP journey.
1. CDP Institute: https://www.cdpinstitute.org/learning-center/what-is-a-cdp