Blog - Series The Steps to CDP Success

The CDP Implementation Process: A Look Under the Hood

In this CDP implementation guide, we review how to implement a customer data platform and set yourself up for success.

As strong believers that the best customer experiences are data-driven, we believe in the power of customer data platforms (CDPs) to help your organization work faster and smarter while delivering exceptional value to customers. In our Steps to CDP Success series, we’ve gone through what businesses need to know to evaluate if they need a CDP and the quick wins and ROI companies can earn from a CDP in just 30 days!

However, the process of selecting and implementing a CDP is not a quick-fix it’s actually a digital transformation project. To take advantage of everything a CDP has to offer, businesses need to consider more than just the product capabilities, but also take a look at their people and processes. Even the best data won’t create value if you don’t know how to work with it. Let’s examine the different phases of a CDP implementation project in order to show how a CDP can transform your entire approach to data.  

If you're looking for a range, a normal CDP implementation can run from three to nine months

While something in the middle of that range can be a good estimation of how long it will take, this time really depends on the specific customers, purpose, and goals. That’s why it’s important to have a thorough discovery process to probe different ways a company can unleash its data.

At Acquia, we recommend quick implementations where the customer works on a more agile mindset, extracting some value from the platform as soon as possible and iterating post-launch. That gives everyone the advantage of familiarizing themselves with the platform and learning about their data from the start. Once we know more about the product, the possibilities and their data itself, we can add new features and transformations.

An implementation like this would look like a few weeks of work setting up the basics to get the data onboarded in the CDP (more on that next) so you can see that value quickly. As you reach more mature phases in your CDP journey, you can start implementing things like machine learning models and more advanced approaches to segmentation and targeting.  

Before You Start Your CDP Implementation 

A CDP implementation isn't just about the product, but also about the end user. Your implementation team should make all details clear to the organization from day one. A few weeks of implementation is usually the discovery phase of back and forth questions with clients. Making sure good communication is established from the start will speed up the whole process. 

We recommend:

  • Have your data ready, and know how you are going to export it and its format
  • Know who on will be involved in the implementation
  • Have your goals clear. What do you expect from the CDP? (This is often a crawl, walk, run approach. Set simple goals to start out and grow to more advanced targeting efforts over time)

Phases of a CDP Implementation 

For the TL;DR (too long, didn’t read) folks, a CDP implementation has a few well defined phases where both the vendor and customer go through discovery and implementation phases. Discovery, as already anticipated, will work in a two way direction. In one direction, the implementation team will use this to find the information we need, but in the other direction for the customer to extract the maximum value of their data, and use it in ways to uncover new opportunities. 

Simplifying how this works, here are the general steps in an implementation: 

  1. Discovery and Education 
    1. Connect data and customer teams and write out requirements
  2. Development / implementation
    1. Map the data, connecting it, writing scripts to do any transformation 
    2. Feed in data
    3. Write transformations
    4. Configure and test identity resolution
    5. Data cleanup
    6. Setup inbound and outbound connections (esp, webtag, sftp)
  3. Test
  4. Launch and iterate

A high level detail on the process would look something like this image.

Data Strategy 

"Strategy is not the consequence of planning, but the opposite: its starting point." — Henry Mintzberg

Every data transformation project starts with data strategy. On this first step, we need to figure out what data the customer has, what vision they have for that data, what they want to do, and  what different technologies they may need.

On the other side, once CDP has been implemented, there's a data enablement phase where customers can run reports, personalization, and execute and implement campaigns, direct mail, and content creation. During this process the customer will have independence MCC and CSM supporting them.

1. Planning Phase

An initial planning phase is not just about the product or even the data. Customers need to understand what features and capabilities a CDP has, so they can extract their maximum value. That means there needs to be some evangelization or teaching effort from your professional services team on how you can use and leverage the platform. 

In essence, the more familiar the customer is with their own data the less time from the implementation point of view will be needed in this phase. The duration of this phase will depend on:

  • Customer maturity,
  • Existence of data strategy,
  • Customer knowledge of their data,
  • Complexity of the use cases,
  • Data quality
  • Known data model,
  • Resource availability,
  • Cross department or global collaboration

Some documentation will be produced in this phase, so implementation teams can work with the customer to mainly understand the data and how it matches with the platform itself.

As we can see in the image below, this phase includes:

  1. Extracting and profiling the data
    1. Based on samples provided for the data discovery
  2. Modeling
    1. PS team designs the necessary changes so the data gets transformed into a universal data model (UDM)
  3. Mapping
    1. Getting the data into the CDP

2. Implementation Phase


AIF and Data Warehouse

In this phase the mappings and light transformations will happen. For example, Acquia’s  Integration Framework (AIF) is a proprietary extract, transform, load (ETL) process.

This AIF framework defines the input, but also output connectors, which will guide how data will come in and out of the CDP.

As an example, let’s say we have some data on a csv format that gets uploaded daily in the sftp. Data could look something like the next table:





Internal note





[email protected]

Some notes






Some notes about Olivia





[email protected]

Hac notes


The csv would be stored in a file that would be named something like: Retail_Customer_20160918.csv, where the prefix will always stay the same while the date obviously will be different each day.

The data can come from different sources and in different ways, and all data does not necessarily have to be mapped or used in the CDP. What is important here is to understand the process and what we want the CDP to achieve.

It is also in this phase that you’ll be able to do some transformations. For instance, imagine that you receive a Gender field with values {1,2,0}. This is where you could do some transformations, say, Map the value and write it into something else more readable. For example {'1':'Female','2':'Male','0':'Unknown'}.

Mapping data to the platform

Identity Resolution and Business Intelligence (BI)

The real key, the real value of a CDP is identity resolution. This is where deduplication will occur as well, based on certain rules and conditions. CDP here will find what users on different platforms are indeed the same ones, or what we know as well as deduplication.

Acquia CDP has more that 200 data columns which our team has identified and recognised as standard columns across different industries. Data columns include things like total revenue, lifetime transaction value, and average transaction value. This makes integrating data much easier, because it doesn't require any customization.

Two hundred is a lot, but for the more we could have there could never be a point where some customers may not need some specific data to be hooked into the platform. That’s where BI (business intelligence) needs to kick in next. Basically in this phase the team will work with the customer to find which custom data needs to be pulled into the CDP.

Lastly, this identity resolution is not something that it just happens once. The internet is a living environment, and all information that the users generate needs to be fed back into the system constantly. The opposite after all would be like not having any CDP at all, because at some point the data would go stale and no longer be valid as user interests change. 

Machine Learning

While doing the set up, cleaning up the data, and establishing the identity resolution will unleash value from your CDP, adding machine learning capabilities to really take your data to the next level. 

And indeed, it’s probably the fanciest, most eye-catchy part, where we’ll start playing with nothing less than machine learning capabilities of the platform.

Some ways to use machine learning in a CDP include:

  • Predictions, like likelihood to buy, likelihood to pay full price, potential lifetime value, …
  • Finding personas and segments, like products that tend to sell together, seasonal activity or observed behaviours (on and offline),
  • Personalisation, where we’ll be able to identify opportunities like best next product, upselling, cross-selling or optimal send time.

3. Production Phase

Last, but of course not least, it’s time to get your CDP to production. Here, there will be some reconciliation, a kind of UAT process where the customer will check the data and results, potentially there will be some further adjustments and once everything is ready for prime time, the CDP will reach its final production phase. 

In a nutshell, what will happen here is:

  • Provision production environment,
  • Configuring production to match development
  • Load all data from all sources,
  • Reconcile all data between original sources and CDP and resolve discrepancies,
  • Add and train users,
  • Build segments, reports, campaigns,
  • Complete integrations,
  • Setup ingestion schedule.

Data, Data, Data

To finish up, we want to reiterate the power of Acquia CDP. On average, current Acquia CDP users have 30 million customers stored with an average 75 Million transactions and more than 800 Million events triggered. Some customers are even running up to billions of transactions!  

The Future

The potential of a CDP is always growing, and organization’s can scale and change their approach as their goals evolve. With so much data to uncover and unleash, the future possibilities are endless. 


This post has been made possible because of the invaluable input and collaborations from:

  • Sevgi Çavuşyan,
  • Nishant Dixit,
  • Meagen Williams 
  • Ben Graney Green

Featured Resources

View More Resources