Introduction
Over the last three decades, marketers and customer experience experts learned the importance of employing data in data-driven customer decision making. With the right data, they realized, machines could assist them in running better programs. The result was more customers receiving relevant offers, and in turn leading to improved response rates and increases in customer satisfaction and retention.
This journey, however, wasn’t short on painful and costly lessons. Stories were common of virtually endless data warehouse projects seriously behind schedule and overbudget. In some of the most infamous cases, $10’s of millions were spent over years, with little to show for it. Why? Because from the onset the goals were misguided, and in many cases the wrong people drove the project.
What went wrong? Simply put, project sponsors set out with the wrong sequencing of goals – trying to solve for the ultimate data repository first and putting the most important aspect, who would use it and how, on the back burner. In other words, they set the priority on sourcing data, cleaning it, and structuring it, and put off concerns on which applications would leverage it. Build it, they posited, and they will come.
Challenges:
Sadly today, many embarking on CDP projects are falling into this same trap:
Select the best Customer Data Platform (CDP) first, build it to solve for nagging problems of fragmented data and cross-device identity. Later, help customer decisioning applications get connected to it.
The problems with this approach are:
- Without considering first which specific outcomes are crucial to success and working back to the data needed to support those, chances are extremely high the CDP won’t have the right data.
- History shows it could take years to agree on the right data, amass, cleanse, stitch, and organize it into a brand-new platform.
- Nearly every vendor calling themselves a CDP is now also claiming to solve for enterprise customer decisioning requirements. Yet selecting the same vendor for both means a direct dependency on this repository, where the CDP must be up and running before the business can run its first new customer engagement programs.
Twenty years ago, at Unica, we saw this exact same problem. The business was waiting for IT to complete the never-ending data warehouse project. Or worse, they took matters into their own hands and selected a tool like Epiphany that required all the data structured and uploaded into its marketing data model (essentially a CDP – just not called it at the time). Sound familiar?
Again at Unica, to tackle this problem, we designed a different solution and approach. We called the solution UDI (Universal Data Interconnect) which allowed marketers to map to existing data sources and run campaigns leveraging that data in place.
We advised frustrated clients to set goals such as improving promotional response rates and urged them not to wait for data warehouse projects to complete. The advice we gave them – focus on redesigning campaigns, use advanced analytics to improve lift, and connect only to data sources required for those redesigned campaigns. Essentially, let the new campaign rules drive the data source requirements. References reported running successful campaigns shortly after project inception. In just months they touted tangible economic benefits, bolstering their case to expand rollout.
CDPs are all the rage – what should I do?
First, the fact that CDPs are “all the rage” is part of the problem. Upon closer inspection it’s the CDP vendors generating the hype, and not the paying clients. Oddly missing are stories of resounding project success and massive ROI, and instead infamous stories of CDP projects failing to meet goals are piling up. In Gartner’s 2021 Cross-Functional Customer Data survey, just 14% of respondents that reported having a CDP also reported achieving a 360-degree view. [i] What we’re witnessing is the classic Gartner technology hype cycle, with CDPs now passing peak hype, and falling into the trough of disillusionment. [ii]
In my 2019 article, To CDP or NOT – 3 tips – then you decide, the advice was beware of the hype in a poorly defined market. Now, in 2022, vendors are trying to differentiate in a still nebulous market. Here are some of the CDP subcategories that have emerged since 2019 [iii]:
CDPs selected primarily by Marketing and Business buyers:
- Smart Hubs / Hub & Spoke CDP
- Real-Time CDP
- Marketing Cloud CDP (e.g., Adobe AEP, Salesforce CDP)
- Campaign & Delivery CDP
CDPs selected primarily by IT, Data, and Analytics buyers:
- Data Integration and Management CDP (focused on data collection and identity management)
- CDP toolkits (used by IT to build a CDP)
- Customer Analytics & Insights CDP
Certainly, the right answer isn’t to buy multiple CDPs. Yet that is exactly what’s happening. And for larger enterprises, some are buying as many as three, simply proving poor alignment between the business and IT. [iv] Having lived through those days, be assured, the result is not alignment on outcomes, rapid access to the right data, and improved customer experience.
At the same time, the right answer isn’t to let the business (or IT) solely determine the selection. Although the business must have primary responsibility and control, it also must tightly collaborate with IT where both parties understand their roles and stick to them. Though unfortunately not common, brands that get this right, and take inventory of what data & systems they have and what roles each party should play, report better success and ROI. As such, follow these rules:
Do –
- Establish a strong partnership between the business & IT, align on use cases, outcomes, and how to measure success. Take inventory of existing capabilities and chart a roadmap together.
- Work back from the highest value use cases and desired outcomes and map out the data needed to support them.
- Make it a requirement to be able to iteratively add to the data repository, as new programs might demand new data sources. It won’t be instantaneous (think in terms of quarterly releases for production data source changes).
- Insist that the decisioning and execution capabilities and the CDP solution be evaluated on their own merits, and if in the end different vendors provide what’s best and can be integrated without herculean effort, select accordingly. Demand references that attest to their enterprise decisioning operational use, scale, and effectiveness.
- If evaluating (or already embarking on) a CDP project, simultaneously consider a re-vamped RTIM project. [v] If a CDP project is ongoing, let the RTIM’s data requirements feed into the CDP’s, not the reverse. And don’t wait for the CDP project to complete. Select an RTIM vendor that can map quickly to existing data and can provide tangible proof of fast time to value and ROI.
Don’t –
- Accept at face value that the CDP’s RTIM engine will be “good enough.” Rather, insist the vendor demonstrates unified inbound and outbound decisioning, real-time re-decisioning at scale, advanced analytics features, and capabilities to incorporate contextual streaming data.
- Don’t accept that having a single vendor will outweigh the benefits of having a best-of-breed real-time interaction management (RTIM) engine.
- Wait until teams agree on all the right data. That day won’t come. Instead, if a CDP has been selected, demand an agile approach for how to enhance the CDP over time. Ask the vendor of choice for RTIM to provide plans for running before and after the CDP project is done.
- Make the mistake thinking that a CDP Smart Hub can deliver scalable and maintainable RTIM decisioning. None can. Most rely on traditional segmentation and scripted / deterministic rule-based journey orchestration – all fraught with old problems of static segment definitions, deterministic offer assignment, and hard to maintain eligibility and engagement rules. A modern RTIM engine with a 1-1 personalization approach solves for all these traditional limitations.
Conclusion
A CDP project, aimed at rationalizing customer data, improving identification, providing segmentation, and streamlining access seems a worthy cause. Yet history teaches us that chasing a complete view of every single customer across all their devices and interaction points is an elusive goal. What’s more likely is a CDP project turns into a giant hole, sucking time and resources. And its expected benefits, like the light bent back on itself by a black hole’s vortex, may never emerge.
Instead, if a CDP project is ongoing, set it on careful rails, and manage scope. Meanwhile, evaluate RTIM capabilities and embark in parallel to address those shortcomings and gaps. Research ROI evidence from CDP and RTIM projects and compare. If resources to fund both projects compete, pit them against each other based on business cases and prioritize investments accordingly. And remember the lesson of sunk costs, and don’t be afraid to adjust project plans and budgets already in flight. Many who have placed bets on RTIM cite quick successes that propel massive long-term returns — some with 10x ROI and more than $500 million in incremental revenue. [vi] Don’t make the mistake of waiting and suffering huge opportunity costs.
[i] Market Guide for Customer Data Platforms, Gartner, March 2022
[ii] Hype Cycle for Digital Marketing, Gartner, July 2021
[iii] Customer Data Platform Industry Update, CDP Institute, January 2022
[iv] Ometria, https://ometria.com/blog/5-reasons-standalone-cdp-might-not-right-solution-retailers, January 2022
[v] What is RTIM, https://www.teradata.com/Glossary/What-is-RTIM, 2022
[vi] Forrester RTIM Wave, https://www.pega.com/forrester-rtim-2022, Q2 2022