Thinking About Building or Buying a CDP?
For Most Enterprises, It Probably Makes Sense To Do A Bit Of Both
The tech requirements underpinning 1:1 customer engagement are myriad and diverse; CDPs represent a faster path to this promised land, but within the scope of “CDP” some aspects make sense for businesses to build and own as part of their competitive data strategy, while others make sense to license.
In the quest for 1:1 engagement across the customer journey, many brands are rallying around the customer data platform, or CDP. And with so many options, many businesses are struggling to understand which of the CDP requirements they should build and which makes more sense to buy.
First of all, you may be wondering why, as a sales leader for a CDP business, I would advocate for any business to build a CDP. Being in this position affords a unique position to have conversations with other business leaders on what they’re looking to accomplish within their teams and business functions. I like to think of this position as an opportunity to ally with our clients against a common enemy in the status quo.
Over the course of dozens of conversations during the past year, two realizations have come to light: first, there’s still limited cohesion on what a CDP should do; second, most businesses should build some form of the things that they believe a CDP should do.
What should a CDP ideally do?
Following the spectrum from data collection to customer interaction, businesses may expect a CDP to:
- Integrate data across all disparate sources
- Unify and manage customer records
- Perform identity resolution
- Provide a segmentation interface
- Stream data in real-time
- Trigger messages
- Orchestrate customer journeys
- Personalize messages
- Experiment, test and optimize campaigns
- Display key reports and insights
- Conduct predictive optimization
- Provide a campaign design interface
- Deliver messages and manage opt-in / deliverability
It’s quite a bit. The truth is, it’s nearly impossible to build a platform that does all of these things, and does them well. After all, the big marketing clouds have purported to do all of these things well for over half a decade and just announced that they’re building CDPs (see Adobe and Salesforce).
What part of this should businesses own?
I orient the build vs. buy question to what businesses ideally want to own in the long run as part of their proprietary technology strategy. There is no cost-based or time-to-value-based argument around building a technology this wide that already exists in the market.
So, what aspects of the CDP vision would a business ideally want to own as part of their competitive advantage?
The enterprise should own its single customer view. Licensed technology platforms can help with integrating data across different sources, de-duplicating customer records and with hosting this data. Most category leaders are years ahead in building out their own, owned first-party data ecosystem, and while it may be temporarily necessary to have platforms hold your single customer view, ultimately an enterprise should build and maintain this within their databases.
Additionally, some data science components make sense for a business to develop as a competitive advantage. While machine learning applications and infrastructure may make sense to license, insights driven by data science and the algorithms to turn those insights into predictive inference can represent a significant competitive advantage for a business.
What should a licensed CDP ideally deliver?
Although there are players in the CDP category who can assist with the above “single customer view” goal, these businesses are more closely aligned to a new sub-category of marketing tech called Customer Data Infrastructure (one of the larger players in the CDI space, Segment, coined this term a year ago). These platforms aren’t competitive with CDPs and are often used in tandem with a CDP. These investments are best evaluated on a build vs. buy basis, when considering building out a single customer view.
This question could actually be reframed as “what does it make sense NOT to build?” Or, in terms of the earlier defining consideration around build vs. buy “what does it make sense NOT to own as part of a proprietary technology strategy?”
Many of the above customer experience orchestration requirements don’t make sense for any individual business to build. There’s nothing proprietary about the technology that allows you to better segment your customers or build and orchestrate customer journeys. These features are largely commoditized across the marketing tech industry. Further (from experience) these are the areas businesses have a ton of corner cases, creating a Magna Carta of feature requests against each of these capability sets. From speaking with clients who have worked at businesses who chose to build this tech internally, I can tell you, it’s a nightmare. Fortunately, many of your vendor partners have been through enough RFPs to have seen the dusty corners of business requirements and have either built against them, consciously de-prioritized them and / or can help you find workarounds.
So why a “CDP” in addition to (or instead of) a “marketing cloud?”
If it makes sense for the business to own their single customer view and then license orchestration capabilities, why not license a marketing cloud? Isn’t this just the kind of solution that cloud companies like Adobe, Oracle, and Salesforce already provide?
There are really two answers to this question:
- In the long run, the “CDP” will probably become the “marketing cloud.”
I’ll expand on each of these points a bit more:
- A good CDP will integrate directly with businesses data infrastructure. This means no cumbersome integration process or 6–9-month migrations. The data in the CDP will exist at parity with downstream data and will refresh as authoritative data sources refresh.
- Further, a good CDP will be agnostic to the systems it plugs into on both ends. If you have Google, Amazon, Snowflake, or an on-prem database (or multiple databases) and want to sync data out into 15 different marketing technology applications, a good CDP will do that off the shelf and integrate in a month.
- Finally, a good CDP will give you the cross-channel orchestration, analytics and data write-back capabilities such that you conduct much of your sophisticated customer targeting in your CDP and not in your “marketing cloud,” which leads us to….
- You may have heard, but Salesforce and Adobe are both building CDPs. The go-to-market motion of these cloud juggernauts is generally to market well ahead of beta releases and furlongs ahead of general availability, and I don’t expect this to be any different. This means we’ll likely see a product generally available in market from “the clouds” in 2021. With these “CDP” products providing a data activation layer to existing “marketing cloud” features like journey orchestration, it’s safe to say that a CDP product with marketing orchestration features will be indistinguishable from a marketing cloud with data activation enhancements.
- In our experience, clients value having their orchestration and analytics capabilities in the same system that processes all of their data. This allows for much deeper insights and much more sophisticated campaign development. Ultimately, it makes sense for businesses to view the “CDP” as their central orchestration system and the “marketing cloud” as one of their message / experience delivery engines
- Given the terms “CDP” and “marketing cloud” don’t actually mean anything, really, and businesses are collecting far more and investing far more in data, it stands to reason that teams will want a system that can handle all of their data driving their customer experience
Every business and team will have their own goals and sets of requirements for deploying CDP technology. In that process of aligning marketing and technology stakeholders on a single vision for “CDP,” often not without disagreement, it’s worth considering where potential competitive advantages exist. In my view, the part to build involves single-customer-view ownership and data science assets; the part to buy is a marketing activation layer with user-friendly features (built for the corner cases) built on top of flexible data infrastructure.