15 January 2025

CDP: build or buy

In summary

• As first party data becomes the backbone of a company’s customer intelligence and targeting abilities, marketers are grappling with the challenges of managing and accessing this goldmine of information.
• The convergence of marketing and IT means that teams need to work better together and define what success looks like.
• Knowing where to start can be a real challenge and coupled with the incoming consent and privacy reforms, scalability is a major concern going forward.

What is a customer data platform (CDP)?

In the conventional sense a customer data platform (CDP) is a ‘packaged software’ that offers end to end data collection, storage and real-time activation capabilities, as defined by the CDP institute.

The architecture of a CDP is a debated topic as traditional CDP’s now compete with data lakes, data warehouses and cloud based applications, to facilitate the collection of customer data from multiple sources and build this into a single view of the customer. The term composable CDP is increasingly common and refers to an architecture where customer profiles are built in a company’s enterprise data warehouse, rather than a separate CDP database.

Should you build or buy a CDP?

Positioned as the centrepiece of customer data intelligence, CDP’s often sit alongside legacy systems traditionally controlled by IT teams and can have a number of duplicate functionalities. For this reason the final decision around a CDP strategy should be made in conjunction with marketing and IT to design the most effective solution.

The build or buy approach to acquiring a CDP is a convoluted discussion with individual departments presenting different concerns and priorities. Below we look at some of the factors that should be considered in harnessing customer data for marketing and analytical purposes, and how to choose the right approach.

Determining the project sponsors, owners, and outcomes

CDP’s have typically been positioned as a tool for marketing teams to create bespoke audience segments, so ease of use and access to relevant data are important considerations. Marketers need agility in unpredictable environments, however a CDP can be a pricey option with volume and feature based cost models making it a complex acquisition, alongside the implementation.

Building your own CDP is not a straightforward project either, so having a partner with the right skills and technical abilities needs to be considered. Today engineers with these skillsets are easier to come by and most mid to large scale organisations would have some of these teams in place, already managing and querying data from disparate systems. 

Any bottleneck between IT and marketing also needs to be addressed. How these teams can work together to access and activate customer data will play an important part of building your own stack. The more recent advancement of reverse ETL tools has meant extracting information out of data warehouses and sending it to downstream systems has bridged this divide, helping to automate process and removing the friction experienced some years ago.

Building a time to value approach

Before considering the implementation of a CDP it is important for companies to understand what data they have and if this data is clean, accurate and useful. This starts with constructing a detailed data architecture diagram to understand where data sits and what the data is currently used for, as well as who the data controller is. As CDP’s are fundamentally a marketers tool, understanding how different data silos add value to a customer profile, and if this data can be used is a good starting point.

Depending on the size of an organisation and the project buy-in, it may make more sense to start with a simple use-case to prove business value, rather than tackling a large-scale deployment. Taking a time to value approach using what you currently have, can help scale a POC to staged business outcomes, delivering demonstrable results, without a large initial cost outlay. It should also be noted that a full scale CDP deployment can take up to two-years depending on the size and data complexity of your organisation.

The demand for flexibility in a changing landscape

A packaged CDP has the advantage of coming with pre-built data schemas and identity resolution options to unify disparate data sources which may take longer to build in-house. However these pre-built frameworks can struggle to adapt to bespoke and complex data structures. Much of the identity resolution in a packaged CDP is dependent on deterministic data matching. These platforms may struggle to incorporate probabilistic matching which is likely to become more of a necessity as privacy and consent changes reduce the scalability of deterministic personal identifiable information (PII).

Saying that, large-scale CDP vendors are quick to adapt and consistently bring out new features and functionalities, however these are done on a mass-scale and may not be relevant to your business, particularly outside of the e-comm world. In a future where change is inevitable, having the flexibility to adapt and customise data structures relating to your business needs, is an attractive option in keeping abreast of ever changing challenges and requirements.

Preparing for tightening data compliance policies

Hosting data within your own data warehouse increases the control you have over how that data is used and processed. The privacy reforms that are expected this year apply stringent controls on the use, consent and management of customer data with further cautions about sharing data with third-party providers. (see OAIC Tracking Pixels and Privacy Obligations for more information.)

Whilst on-shore data storage and processing is now expected, it is only enforced within financial and health industries. Due to this many of the large tech companies now have local on-shore data centres in Australia. However, the new privacy reforms may issue more precise controls over this, including informing customers when their data leaves local shores even for temporary processing purposes.

The introduction of strict consent frameworks next year will mean the volume of consented data will be severely depleted. It’s expected that over 60% of Australian adults will opt-out of their personal data being used in most cases. For many organisations this is going to have a considerable impact on scalability and how customer data can be used. A one size fits all approach, in particular in terms of costs, may be counter productive when these changes take place, as volumes decrease and customised algorithms play a more centralised role.

Louder’s recommendation

Louder has been helping a number of our customers take a more cautious approach by using what they have to build use-cases and measure impact before a full scale deployment of an expensive product suite. Whilst it’s not a one-size fits all approach, below is an example of the initial steps we follow:

1. Identify the project owners, sponsors and contributors.
2. Ideate and develop a list of use-cases with them.
3. Build out an architecture diagram to identify what data is available, where it sits and if it can be used.
4. Identify effort vs outcome to build a delivery strategy and ascertain what’s possible. This includes looking at data quality and organisation to ensure delivery is compatible across teams.
5. Begin data aggregation into a single source.
6. Identify and build audience segments to activate across your main digital channels.
7. Continue to develop and refine audience strategies and outcomes.

Proving value on a time and materials basis

One of our clients is using a large portion of the Adobe Marketing Cloud but omni-channel personalisation was still a huge challenge. A further challenge was that known customer identities were limited and much of the audience engagement and interest was cookie-based.

They needed a way to demonstrate value without committing to a large commercial license. 

Louder architected and developed a solution to consolidate first-party data from disparate systems into the Google Cloud Platform. The client used this first-party data to define intent based first-party audiences through BigQuery. These audience segments are activated across Google Ads, DV360, Meta, and Adobe Campaign through connectors built by Louder. The results achieved a 47% higher CTR than baseline figures. We continue to expand use-cases across more channels whilst assessing the scalability of different solutions.

Key takeaways

• Many companies have experienced great success investing in packaged CDP solutions but for the most part these are large organisations with a huge volume of known customer identities.
• The output of the CDP investment is predominantly online within an e-commerce-type environment.
• As a company it is important to understand your data, ensure that you have scalability and know that you can achieve return on investment through a couple of proven use-cases. If you tick these boxes then a CDP could be the right investment in the long term for your business.
• Alternatively you may want to approach a data-driven communication strategy by consolidating what you already have in place. Hosting your data on your own cloud infrastructure allows more control amid significant regulatory shifts within our industry, and enables a steady approach to testing and changing requirements.

Get in touch

Louder looks forward to opening up this discussion further in 2025 to help you achieve proven outcomes with your data and audience strategies. Get in touch today.