If the ‘90s and 2000s were the age of information, since the 2010s we’ve been in the age of the customer.
They expect differentiated and seamless omnichannel experiences.
CRMs won’t get us there (and neither will Social CRMs).
What is a Customer Data Platform?
Not unlike a CRM, a CDP collects customer data.
But the similarities end there, with two crucial differences:
All users: CRMs focus on known or potential customers, CDPs track all users, anonymous or identified.
All data: instead of structured data collected manually or semi-manually, CDPs ingest data produced by any channel: online or offline, structured or unstructured.
With CDPs, we capture identity data (e.g. email, name, phone number, social accounts), qualitative data (e.g. motivation, opinion), as well as behavioral or quantitative data (e.g. e-commerce transactions, web analytics).
More importantly, CDPs handle identity resolution allowing you to create a unified profile of your customers and audience across all channels.
The anonymous or semi-anonymous interactions a user may have via your website and WeChat official account may still be attributable to their profile once they are identified via a conversion event (e.g. e-commerce transaction, subscription, login).
What’s more, we can use probabilistic rather than deterministic approaches for identification (e.g. phone number or WeChat account binding).
That single customer view allows us to derive actionable insights, answering questions like:
What was the product this customer bought before their current purchase?
Which segments/target groups does this customer belong to?
Is this customer likely to churn?
What is their (purchase) intent, and timing?
What is the value and predicted future value of this customer?
Where do they prefer to interact and create moments?
What are their preferences and where are they in the customer journey?
Why you should care…
CDPs make a lot of things possible: personalization, marketing attribution, abandoned cart recovery and prevention, customer look-alike analysis, behavioral customer segmentation, customer lifetime value analysis, social listening, next-best-action analysis, scoring…
These needs are spread across many teams: C-suite, technology, marketing, finance, customer support, sales…
If you’re letting these teams solve their problems on their own, you’re missing out on 3 fronts:
Cost: each team will buy, build and maintain tools that are answering facets of this whole problem, with a good deal of overlap.
Knowledge: having silos of expertise in each team prevents you from creating (and retaining) experts and creating cross-organizational knowledge on how to tackle these challenges.
Opportunities: some things are simply not possible without centralized tools and systems; in-app next-best-action marketing could require customer propensity analysis, location data and customer journey analytics among other things. These data points are collected from multiple channels and by different teams.
You’ll be stuck reinventing the wheel, scratching the surface of what is possible with your resources and data.
Since 2015, we’ve probably launched digital products for about 100 million users across our clients in China and abroad, from Starbucks and Walmart to Burberry and Nike.
I won’t get in the build v.s. buy discussion just yet, but regardless of your approach, I would keep the following 3 things in mind:
Own your data: make sure the data is made available to your teams in a format that is usable. API access is nice, but nothing beats running queries directly against a database.
Don’t fall for the buzzwords: if you’re being sold on “AI” or “big data”, run. You have a lot of ground to cover before you can or need to invest in machine learning. Save for a single one of our clients (Apple), we’ve never had to use technologies like Cassandra or Hadoop. You probably have “fat data”, not “big data”, problems.
Keep OPEX in mind: if every single change to your scoring strategy or additional data source costs you weeks or months of development, you’re doing it wrong (or you’ve hired the wrong partner).
For reference, our own CDP solution stores most things in PostgreSQL:
Business and operations teams are able to build dashboards with Metabase, Power BI or Tableau.
Marketing teams can create segments or scoring with a simple SQL query and tools like dbt.
Sales or customer support teams can easily integrate full customer profiles in their tools (creating a query based component in Salesforce Lightning is pretty trivial).
Keep it simple and tackle the boring but essential problems first.
In a follow up post, I’ll explain why we prefer Customer Data Pipelines to Customer Data Platforms and dive into how we build and leverage them with our clients.