In my first month as a product evangelist at Hightouch, I talked to many organizations to understand why they chose Composable CDPs over traditional packaged CDPs. These organizations have realized that their existing data warehouse can be their CDP! I noticed a common set of reasons in my conversations, so I’ve compiled my “Top 10” list of why organizations should use Composable (warehouse-first) CDPs to turn their data warehouse into their CDP instead of purchasing packaged CDPs.
#1 - Access to all customer data
Organizations want to deliver holistic and personalized customer experiences. However, providing personalized customer experiences is only possible if you can access all customer data. If you attempt to create personalized experiences with only subsets of customer data, you can inadvertently create worse customer experiences. For most organizations, the data warehouse is the closest thing they have to a single source of truth for customer data.
Packaged CDPs became popular many years ago because marketers couldn’t get access to the customer data and audiences they needed. Back then, the data warehouse was simply a place to back up your data and was only accessible to a few IT folks. If marketers wanted customer data or audiences from the warehouse, they would have to request it and wait days or weeks to get it. This frustration prompted marketers to invest in 3rd party packaged CDPs to circumvent their internal IT teams.
But nowadays, data warehouses have evolved and become faster, more accessible, and more powerful. Composable CDPs act as marketer-friendly front-ends to the data warehouse, providing the best of both worlds: access to all customer data in a timely manner. The organizations I spoke with want to build audiences, run campaigns, build journeys, and perform other typical CDP use cases by accessing as much customer data as possible, which means accessing customer data directly from their data warehouse. Performing these functions directly from the warehouse allows you to tap into the richest customer data set available at the organization. Through the data warehouse, you can access customer data from the website, mobile app, stores, customer support, product inventories, transactions, etc. Access to this customer data provides many enrichment opportunities to help marketers personalize campaigns and customer journeys.
Additionally, using the data warehouse as the primary source of truth means that marketing campaigns run on the same data that BI reporting leverages. If you add a packaged CDP into the mix, you can have measurement/analysis conducted from the packaged CDP that doesn’t match the data in the warehouse. This type of analysis mismatch can erode trust in the performance of marketing efforts.
#2 - Lower overall CDP cost
When I met with customers who said they wanted to perform CDP functions on all customer data, I asked them, “Why don’t you just send all your customer data to a packaged CDP?” After a chuckle, all of the organizations I posed this question to said the cost of sending all customer data to a packaged CDP would be outrageously expensive. When most packaged CDPs or marketing cloud suites were initially conceived, cloud data warehouses like Snowflake, Databricks, and BigQuery didn’t exist. However, cloud data warehouses have become increasingly affordable over the past few years, especially relative to packaged CDPs. Adding rows of data and additional computing can add up, but these costs pale compared to sending data to a packaged CDP.
Imagine sending all data from your website, mobile app, customer support, store, financial database, etc., into a packaged CDP. Each data point would be treated as an “event” (or “row”) for which the packaged CDP vendor charges you. In my conversations, I have heard the cost difference between sending data to a packaged CDP and a cloud data warehouse could be as high as 10x! As a result of these cost implications, organizations have to pick and choose which customer data is and is not worth sending to the packaged CDP to save money. As noted above, once you don’t have all of your customer data, the overall value of your CDP diminishes greatly.
The business model and pricing of packaged CDP vendors are built on the premise that they will store your data in their proprietary data store. Most of the CDP functionality offered by packaged CDPs (identity resolution, audience creation, marketing activation, etc.) only works if customer data is housed in their proprietary data store. This creates challenges for packaged CDPs going forward, even if they consider moving to a composable model:
- Will they be able to re-design their entire product to work from an external data warehouse they don’t own versus their proprietary data store?
- How will they transition legacy customers to the new composable model?
- How will they adjust their business model and pricing for a world without storing data and simply reading it from the cloud data warehouse?
These are not trivial things for packaged CDP vendors to solve, and many organizations considering new investments in CDPs prefer to leap-frog to the newer Composable CDP model instead of investing in packaged CDPs that seem dated and may have a long road ahead of them to modernize.
So, if organizations cannot send all customer data to a packaged CDP, they often send customer data to the packaged CDP and their cloud data warehouse. The organizations I spoke to lamented the cost of sending customer data to multiple systems. Since data is stored in various places, these organizations pay numerous times for the same customer events and data. For example, I spoke to one organization that was paying to send all website/mobile app data into BigQuery, and then from BigQuery, a subset of customer data was sent to the packaged CDP. Given the organization's size, the cost of collecting and storing this data two times was enormous! No additional value was added to the data; they were simply paying a surcharge because the packaged CDP vendor required them to duplicate their customer data into the vendor’s proprietary system for it to function.
While common sense would tell you this approach is absurd, organizations are often unaware that this antiquated model is no longer needed. I am shocked that many organizations are unfamiliar with how Composable CDPs work. Once they understand they can work directly off their cloud data warehouse (BigQuery in the above example), they realize the benefits immediately. In addition to many of the issues outlined above, the sheer cost of data duplication is obnoxious and unnecessary.
#3 - Better time to value
Packaged CDP implementations are notorious for being expensive and time-consuming. Sending multiple customer data sources into a packaged CDP requires creating many data models, schemas, and data collection techniques. I spoke to several organizations that spent more than a year trying to aggregate customer data and get it to conform to the data schema of the packaged CDP vendor.
Most organizations I spoke with mentioned that the customer data they needed already existed in their data warehouse. However, the packaged CDP vendor required them to send similar data in parallel or find ways to backfill data from the data warehouse to the packaged CDP. Regardless of the approach, the requirement of packaged CDPs to send or duplicate customer data in the packaged CDP meant they spent their first year implementing instead of deriving value from CDP use cases.
In addition to the lost opportunity cost of a lengthy packaged CDP implementation, many organizations mentioned they paid as much or more money to outside or packaged vendor consultants than they did for the packaged CDP itself! The excessive costs make sense since a year’s worth of consulting isn’t cheap. I repeatedly heard that packaged CDPs had many unique nuances/limitations that made them reliant on consultants specializing in that packaged CDP.
Because Composable CDPs leverage data in the data warehouse, implementation is much more straightforward. Most implementation work involves choosing which data warehouse tables/columns you want marketers to access. Composable CDPs typically don’t require a lot of incremental data collection or reformatting since you don’t need to adhere to a packaged vendor’s schema. With Composable CDPs, you can get up and running in days or weeks instead of months or years! In extreme cases, I heard from organizations that set up Hightouch’s CDP while they were implementing a packaged CDP as a stop-gap solution until their packaged CDP vendor implementation was completed. Additionally, because minimal implementation is required, any investments you make in outside consulting can be applied to adding value through CDP use cases instead of non-value implementation work.
The net result of Composable CDPs is that they provide significantly better time to value. Without as much need for a lengthy implementation or consulting, organizations can begin deploying CDP use cases and iterate quickly to achieve short-term value.
#4 - Modular vs. monolithic
Having been around for over a decade, packaged CDPs come with a lot of features and functions. In many cases, organizations don’t need all the functionality provided, but because package CDPs are often monolithic, you pay for the entire package regardless of what you use. A funny phrase I heard was, “I feel like we have purchased a Ferrari but are only using it to drive to the grocery store!” While all of the features of packaged CDPs may demo well, most organizations I spoke with weren’t ready for all the functionality and used less than 40% of what they were paying for.
Conversely, Composable CDPs break their features out into distinct modules, so you only pay for what you use. This modular architecture allows organizations to pick and choose what CDP functionality they need and only pay for that while still providing a runway of functionality if required in the future. Feature composability allows organizations to work with the vendors they want instead of being forced to use one packaged CDP vendor for all CDP functions. For example, one organization I spoke to had invested much time and money in identity resolution. When they talked to a packaged CDP vendor, they were told that the identity graph they had spent years building would have to be replaced by the packaged CDP identity graph. However, with a composable CDP, they could continue to use their identity resolution vendor.
CDP feature modularity is important because it’s unrealistic to think that any vendor can be the best at every CDP function. While there are certainly advantages to reducing how many vendors you work with, if there are areas in which your organization wants to go deeper, they should have the option to do so.
#5 - Better warehouse functionality
As described above, to be effective, packaged CDPs need to be the single source of truth for customer data. And since packaged CDPs require all customer data to be in their proprietary data store, organizations must choose whether the packaged CDP or their cloud data warehouse will be the primary store of customer data (or pay for both). The need for one source of truth puts packaged CDPs in the unenviable spot of competing with cloud data warehouses. In addition to the increased cost of using packaged CDPs instead of cloud data warehouses, organizations must consider which source of truth is the best long-term investment.
Over the past few years, cloud data warehouses have surged in popularity. Cloud data warehouses like Snowflake, Databricks, and BigQuery have become ubiquitous, amassing thousands of customers. These customers bring new feature requests, which begets more advanced warehousing functionality. The pace of innovation in the cloud data warehouse space has been staggering. When many customers flock to a new technology, all other vendors and consultants tend to follow. The net result is that cloud data warehouses:
- Achieve massive economies of scale and lower costs
- Have more ways to ingest data
- Are less proprietary
- Have many 3rd party vendor integrations
- Can be used for other functions besides CDP
- Have many people trained as experts that are available as consultants or full-time hires
Organizations I have spoken to have concluded that no packaged CDP vendor will achieve the scale or innovation pace of the cloud data warehouses. These organizations want to maximize their investments in data warehousing, have the flexibility to integrate with as many vendors as possible, and have a rich talent pool of people to help them. For this reason, they are choosing to avoid single-purpose, proprietary packaged CDPs that will have difficulty keeping up with cloud data warehouses.
For example, I spoke to one organization that asked if Hightouch had a way to connect to Vendor X so they could get its data into the CDP. They had asked a packaged CDP vendor and were told they could add a data ingestion connector to the product roadmap but couldn’t commit to a timeline. I mentioned to this organization that there was already a native connection from Vendor X into Snowflake, Databricks, and BigQuery, so there was no need to have a data ingestion connector to Hightouch (since so many organizations are used to dealing with proprietary vendors, it took a while for this concept to sink in, but eventually it clicked for them!). These examples of innovation and critical mass underscore the value of going with the flow of where the industry is headed with cloud data warehouses instead of using packaged CDPs.
#6 - Data schema flexibility
Each organization is unique. Some are B2B, some are B2C, some are B2B2C, and almost all of them have different ways they want to capture and store customer data. Because each organization is unique, they often have custom data models/schemas. For example, if you sell pet products, you might have households with pets, and each pet may have its own profile that stores things like grooming appointments, toys, or food preferences. A pet data schema will differ significantly from a streaming media service that stores customer data related to movies, TV shows, genres, etc. The fundamental idea here is that the Composable CDP conforms to your data schema (with infinite flexibility), not the other way around where you need to wedge your data into the opinionated objects within a packaged system.
Unfortunately, many packaged CDP vendors force customers to conform to their proprietary data schema (often consisting of just users and events) because that’s how their back-end warehouse processes and activates customer data. It would be extremely expensive for packaged CDP vendors to provide additional flexibility in their schemas. As a result, you may have to purchase a different packaged CDP product if you are in B2B vs. B2C, even if your organization might have both types of business models. Many organizations I spoke with mentioned “putting a square peg in a round hole” when making their data fit into their packaged CDP vendor. Schema limitations create many limitations regarding CDP audience creation and activation.
Some examples of entities that are easily represented in a Composable CDP, yet quite difficult (or impossible!) to in a packaged CDP, across different industries:
- Financial Services - Banking or investing accounts related to a customer
- Hospitality - Reservations or bookings related to a customer
- Travel - Passengers, bookings, flights related to a customer
- B2B SaaS - Teams, workspaces, or accounts associated with an individual user
Composable CDPs allow you to leverage your data warehouse's data schema and adapt to any level of complexity or granularity. You can even leverage other data schema investments you have made (e.g., dbt), where you can manage changes and maintain version control. This approach allows your organization to inherit any future schema changes without duplicating them in your packaged CDP.
#7 - Better data privacy & security
Over the past decade, customer privacy has been a hot topic. Privacy regulations like GDPR, HIPAA, and CCPA have forced organizations to get consent to store customer data and respond to customer data deletion requests. When organizations store customer data in their data warehouse and a packaged CDP, ensuring that privacy consent and deletion requests are performed globally is more challenging. Most organizations I spoke with had initiatives to reduce how many places customer data was stored. Minimizing customer data storage is often mandated by internal privacy officers who want to limit potential fine liabilities. Because Composable CDPs don’t store data, they are much more privacy-compliant and have thrived in privacy-conscious regions and industries (e.g., healthcare, finance, etc.).
In addition, storing customer data in the cloud data warehouse and a packaged CDP creates an additional security risk. Everyone is familiar with customer data breaches over the past few years. While no customer data warehouse is immune from attack, minimizing the number of places customer data is stored is prudent. Because Composable CDPs store less customer data, they are more secure, better for compliance, and more aligned with where enterprise organizations want to go. For obvious reasons, most organizations are already spending significant effort ensuring their cloud data warehouse is privacy-compliant and secure. However, if customer data is also stored in a packaged CDP, these privacy and security efforts must be duplicated in another environment.
#8 - Unlimited data retention
In addition to the above packaged CDP data privacy regulation risks, organizations leveraging packaged CDPs can also be limited to specific data ownership and retention policies. Because customer data is stored in their platform instead of your data warehouse, packaged CDP vendors may force you or financially incentivize you to remove historical customer data to limit their potential liability. While this may help the packaged CDP vendor limit its liability, your organization cannot leverage all of its customer data. For example, you may want to build an audience of customers who have ordered at least twice yearly over the past five years. Building this type of audience can be problematic if you only retain twenty-five months of data.
Since Composable CDPs use your data warehouse, you can indefinitely use your customer data. Whether you deprecate historical data is your choice, not the vendor’s. In addition, it is relatively cheap to retain historical customer data in cloud warehouses versus the expensive data retention charges you may incur from packaged CDP vendors.
#9 - AI readiness
In the past year, AI has taken marketing by storm! The introduction of agentic AI is starting to bring decades-old promises of personalization to life. Of course, AI requires enormous amounts of trusted, clean data to work. If you have the right tools and data, you can use agentic marketing to analyze customer behaviors, suggest product cross-sells, drive experiments, choose the best marketing messages, etc. Hightouch’s AI Decisioning product is an excellent example of leveraging data and AI to personalize customer communications and learn customer preferences.
As mentioned, the data warehouse is typically most organizations' best customer information source. The data warehouse is where your organization spends the most time ingesting, cleaning, and verifying customer data. The data science team builds its models from the data warehouse and outputs its results there. The business intelligence team also builds its executive dashboards and reports from the data warehouse. So, doesn’t it make sense that the data warehouse would be where AI agents look for the data they need to help marketers? The better your marketing data, the better your marketing AI will be.
But what if you have a packaged CDP that may not have the cleanest customer data, is not the source of truth for the data science and BI teams, and only has a subset of customer data? If you were to apply AI models against this data, the results may be sub-optimal. By making the data warehouse your organization’s CDP, you benefit from another dividend - having the ideal dataset for the upcoming agentic AI revolution.
#10 - Avoid vendor lock-in & data ownership issues
As discussed above, when you purchase a packaged CDP, you are using their proprietary data store, most of their CDP features, and are subject to their data retention policies. Since most packaged CDPs are SaaS products, you rent space on their servers, send them your customer data, and build customer audiences within their product. While you technically “own” the customer data you send to the packaged CDP vendor, this type of ownership differs from a cloud data warehouse. For example, packaged CDP vendors will require you to use their proprietary data collection methods (e.g., SDKs), which makes it difficult to leave the vendor without re-tagging websites and mobile apps (which means you will be with them for a long time since no one likes re-tagging!). If you build hundreds of customer audiences, you cannot easily migrate those audiences to another CDP if you decide to make a change. If the packaged CDP is part of a marketing suite, it may be the case that audiences only work with other products in that vendor’s suite of products.
As mentioned above, data privacy regulations force organizations to obtain data tracking consent and rely less on 3rd-party cookies. Therefore, many organizations seek ways to build their own 1st-party customer/prospect identity graph. Identity resolution is one of the core tenets of CDPs, but when you use a packaged CDP, they own and manage your 1st-party identity graph. Composable CDPs help you build and manage your 1st party identity graph directly within your warehouse, which makes sense given its importance to all your marketing initiatives.
Customer data is increasingly becoming a strategic corporate asset in today's digital world. Organizations' effectiveness in leveraging customer data can be the difference between winning and losing to the competition. So, if data is one of the biggest differentiators, it makes sense for organizations to want complete ownership of their customer data. The organizations I spoke to wanted to take ownership of their data collection and take a more vendor-agnostic approach. They don’t want to be locked into any one CDP vendor in case a better vendor becomes available. Avoiding vendor lock-in is one of the advantages of the Composable model. While there will always be some vendor switching costs, packaged CDP vendors have much more lock-in and switching costs than Composable CDPs that sit atop your data warehouse.
Closing Thoughts
As you can see, I have been pretty busy learning what is happening in the CDP landscape! While packaged CDPs had their rationale and reasons in the past, the list above illustrates why organizations are using Composable CDPs to transform their data warehouse into their CDP and are rethinking their investments in packaged CDPs.
Another interesting observation is that almost every organization I spoke to that had purchased a packaged CDP mentioned that the reality of what they could do with the packaged CDP never matched the vision that the packaged CDP vendor sold. Several organizations described the grandiose vision they saw during the sales process—how all data was perfect, all customer profiles were clean, and audiences flowed freely between vendors. However, several years into their multi-year contract, the CDP they had didn’t look anything like it was described during sales demos.
Unfortunately, many of these organizations didn’t perform a proof of concept (POC) before signing a multi-year contract, mainly because the months-long implementation of packaged CDPs made a POC impossible.
The ability to conduct a POC highlights one last advantage of Composable CDPs - how easy it is to try them out and prove they will provide value since they don’t require lengthy implementations. So, even if you have invested a lot in a packaged CDP, if it isn’t working for you, don’t hesitate to pause and re-assess. You can always try to accomplish one or two essential use cases with a Composable CDP parallel to your packaged CDP implementation.
One of the reasons I was excited to join Hightouch was to help educate organizations about the industry's direction and why Composable CDPs are changing the entire CDP landscape. The idea of organizations extracting even more value from their investments in data warehouses and turning them into CDPs is exciting. The above learnings have confirmed my rationale and excitement for joining Hightouch and entering the Composable CDP world.
I have a presentation that covers all of the above. If you would like me to host a free informational session with your organization where we can dive deeper and tackle questions, you can use this link to schedule it. Thanks!
