CDPs: thank you, next?

It has been six months since sharing 2019 CDP Thoughts and Projections so I thought it might be time for a CDP check-in.  I’ve now spent the last year and a half looking at Customer Data Platforms (CDPs) and working with Enterprises around their needs and processes for customer data strategy and instrumentation.

I’ve known many of the companies in the CDP space from their previous incarnations as tag managers, SDKs, analytics and marketing optimization companies so it has been very interesting to watch the market unfold over the past couple years. As with any new market hype cycle, especially one for enterprise software, it’s very challenging. Whose budget has the CDP is a question not easily answered in many orgs. And in e-commerce and omni-channel retail where I’ve been working, there’s plenty to keep engineers busy and multiple issues, often glaring, with data/CDP readiness.

Still, my excitement about the core technology solution of CDP has not waned. Having the customer ID become the atomic unit of data collection is an absolutely transformational evolution in marketing. It unlocks the ability truly understand the incremental value of almost every marketing activity and find the right mix of improving customer lifetime value and performance-based marketing attribution. The services that power customer experience, price intelligence, inventory/product mix, supply-chain – everything related to business success – will be made more intelligent with data stitched to an OWNED first-party customer ID.

I’m so F’in Grateful

Businesses, of all sizes and in all verticals have different and unique ways customers consume them. No two sets of data and analytics are the same for any business. What I’ve learned over the last year and a half is that purpose-built services and applications need to be (and will be) created on top of CDPs.

ML/AI based applications and services will deliver intelligence and automation to solve complex, data-intense use-cases. Many of these will be specific to verticals and business models. For every business there are no shortage of use cases around customer growth and retention.

Most CDPs claim they have some level of technical ability to solve for use cases. Many do have segmentation capabilities, some have advanced methods of aggregation, some visualization, and some even have some basic machine learning and out-of-the-box data models. Still at the end of the day, the core technology of CDP is tradecraft; organized plumbing and table building.

Point solutions backed with deep-learning will be the next evolution. For some enterprises these solutions/applications may be built by internal teams and or consultants. This is current state of the market. However, in the next few years I expect we’ll see a new wave of vendors take data from the CDP and deliver configurable systems with embedded AI that sows the seeds for true integrated marketing and intelligent marketing automation.

Here are a few spaces I expect and am excited to see future-forward applications and services built on top of CDPs:

Conversion Rate Optimization

 I am going to admit to bias here. More than half my career has been focused on CRO. I guess that’s why it is amazing to me that despite the rise of Marketing SaaS, free Analytics and AdWords that e-commerce conversion rates are still 3%. That’s the same rate as when retailers were concerned people wouldn’t use their credit cards online! WTF?! How have conversion rates stayed so shitty with BILLIONS invested in marketing software? End rant.

Retailers more than ever need help optimizing conversions. This matters even more to omni-channel. BOPIS (Buy Online Pick-up In-Store) will make up more than 50% of Home Depot’s “digital” conversions. In our modern world, last mile can be anywhere.

Of course, the success of DTC is also predicated on conversion rate. Double your conversion rate and you halve your CPA or as the kids call it now, CAC. Plainly speaking current CRO tools mostly suck still. Much of the software is a generation or two old. A/B testing is fine but neural networks for conversion rate optimization are coming.

 Lead Scoring

 As just mentioned, most people don’t ever end up buying. So, it’s most important to understand the likelihood of people to convert so you can pay extra special attention to those with higher likelihoods than lower. As I used to say in my Landing Page Optimization days, convert the converted!

With a few exceptions, lead forms and the CX around them today are still basically what lead forms and their CX was 20 years ago. Exceptions are early efforts in “conversational commerce” which extends not just to chatbot interfaces but also nascent voice technology.

Of course, advances in machine learning are going to power this area. It’s an area ripe for deep-learning and AI. And if we view the internet properly as the greatest direct marketing platform ever built, we understand the value for business and consumers that will be created here by intelligent and dynamic data gathering. Understanding what people want is more than a lead-score. It’s a universal technology.

 Media Spend Attribution

 One of the powers of unlocking ID from vendor systems and the associated data silos are that these joins can for the first time create a clear picture of the customer journey. There are vendors focused on “journey orchestration” which is a terrible term…but more on that in the next section.

The most important unlock for many will be a clearer picture of customer response or lack thereof to your media. Most of all your paid media.

There is no mistaking the power of Search & Social Media to create brand and product awareness as well as drive sales. In fact, after years of focus on driving clicks, driving sales has now become the focus of Facebook and Google, and has always been the focus of Amazon. This matters because it changes their algorithms.

It matters more because Customer Acquisition Cost is for most the primary metric for DTC everything. Now with the capability to stitch together the journey WITH the outcome we have new powers. The most important of these powers for many will be understanding how different media channels effect Lifetime Customer Value. Looking beyond attribution as a customer acquisition metric but looking at media spend through the lens of LTV will redefine bid strategies and lead to orders of magnitude increased spend efficiency. The very reason why the duopoly wants to make your products be their products.

Targeted Communications

 Ads, email, website, targeting, and re-targeting have become a core part of the marketer’s tool belt. The unfortunate part – the systems that power these tools are not very intelligent. Very little go-to-market (GTM) strategy is power by advanced segmentation. What I mean by advanced is a real-time, predictive, closed loop and automated segment creation. First party-systems for targeting should be instrumented with cohorts that respond and create different experiences based on the predicted lifetime value of that customer.

Key advances here are the abilities to collect the data to fully understand the incremental benefits of targeting (this seems to be a theme with unified ID) as well as ML to improve those benefits over time. In addition, intelligence can be applied to look-alike modeling and advanced clustering algorithms like DBSCAN and Gaussian Mixture Modeling. Fun and games await!

Data Democracy

This is where I think the innovation is going to happen. There is going to be a next wave of ML/AI applications and services in all the categories above. It will be ushered in because of the newfound ability of CDPs and the Cloud to free customer data from the silos it has existed in for over a decade and bring it together in a unified view of the customer and thus a unified view of marketing and experience.

The important point is thanks to AWS, GCP & Azure, we now have the data, the databases and the data services that allow for the development of advanced data model solutions specific to the use cases of your business and most importantly, the way customers interact with it.

In the accelerated data landscape we live in, targeting and optimization platforms are now a decade or more old. Most companies (of all sizes) still have these antiquated systems deployed and are using them to make decisions. Or worse, they are deployed and not used to make decisions because no one in the organization trusts the data – while still paying for this software! There was nothing wrong with these tools. I loved and used and help build some of them. But they are old and not made for a world where ML/AI is going to have an everyday presence in all consumer marketing. That world in coming.

The most fascinating thing about all this to me as mentioned earlier, is that despite twenty years of advances in marketing technology the same problems from day 1 of marketing still exist. The most valuable use cases for marketers will always be the most valuable use cases. And we still can’t move the ball from only 3% of people who visit an e-commerce site buying. Now imagine what happens when we double or triple this in the coming years because of ML/AI.