As organizations creep toward a common standard of customer data collection through marketing automation, CRMs, and a host of attribution and other pinpoint marketing technologies, the bar is raised for staying ahead of the curve, and marketing looks to IT to solve for it.
Marketing and IT have two choices from there:
- Spend millions of dollars and one to two years building an in-house solution via a data warehouse, business intelligence (BI)/ETL, or
- Install a Customer Data Platform (CDP) for 10% of the cost, with a three-month implementation.
The explosion of the CDP, this new category within enterprise marketing, exists because we have an overabundance of disparate data and a severe deficit in insight into it. CMOs struggle to piece together marketing systems and data, with 52% of marketers pinning integration as their biggest obstacle to success with MarTech.
This challenge of remaining competitive around data is a good thing, driving innovation in a time of marketing revolution. But “being innovative” isn’t the point, is it? Yes, the “winners” lay the groundwork for a modern marketing data architecture. But what defines success in any new marketing venture isn’t a “we did it first” mentality.
It is proven impact to revenue.
That makes data capture, analysis — and even prediction — a business-critical capability. This applies to all businesses, but particularly to the enterprise marketing organization, whose quantity of data and scale of operation is so massive that manual parsing, analytics, and insight simply isn’t an option anymore.
Data Warehouses, Data Lakes, and the Customer Data Platform
Think about all of the data that sits across the customer journey. From the beginning of that journey to the post-customer end of it, it’s siloed. Not just a little siloed — a LOT. Enterprise Marketing organizations have over 100 systems at play, according to a 2017 study. Mid-market companies regularly use between 20-30.
The technology you’re using today simply wasn’t created to weave it together; at least not at the scale required for meaningful full-picture insight.
The CRM or the typical system of record is heavy on later stages of the customer journey.
Marketing automation, the system of engagement, sits earlier in the journey but stops short before the journey is complete.
We know we need to get a more comprehensive picture of our customers — the channels, programs, attribution, and engagement working within our organizations to drive pipeline and revenue growth.
We’ve seen the CDP ecosystem explode from that need, with one of the first questions an enterprise exploring a CDP asks being: “Should we build it ourselves?”
The answer lies within a list of considerations around obtaining that comprehensive customer picture. In short, the “build it” option doesn’t actually result in a CDP. What you get is a cobbled-together data warehouse, BI/ETL system.
Ask a marketer about their marketing database, and they will point to their CRM or MAP. But without the ability to match customer interactions over different channels, and because they don’t have the machine intelligence to pick up on customer behavior and track activity outside of the system itself, CRMs and MAPs see just a fraction of the customer journey.
When initially looking into solving for enterprise customer data problems, terms like “customer warehouse,” “data lake,” and “customer data platform” can be seemingly used interchangeably. They are not the same thing, and this is critical to any purchase or build decisions you’ll need to make, so let’s break them down.
- A data warehouse is a place to centralize structured data and information for reporting. The data has already come in with a particular use in mind.
- A data lake is a collection of data as well, but it can be raw and unstructured.
- A CDP is a purpose-built data warehouse for customer data, unlike generic data warehouse solutions. It comes pre-packed with connectors to CRMs, MAPs, websites, and other customer systems, and includes the integrations, reports, and automation necessary for orchestration of your marketing data lake and a data warehouse. Finally, it adds functionality that ranges from data cleansing and segmentation to marketing attribution and deep marketing analytics.
The CDP is uniquely designed to meet the needs of modern, revenue-driven marketing organizations.
Why should I use a CDP and not just build my own data warehouse?
When IT is tasked with solving Marketing’s problem, it’s critical they understand the core business problems driving the need for change.
Let’s look at them.
- The need for significant improvement in marketing campaign effectiveness.
Let’s be honest: Marketing managers have been through the ringer. An onslaught of tools and shifting priorities have left them with more information, yes, but also without clear conviction around what to do with it, or how to measure its impact.
A CDP increases your visibility into how effective your campaigns and content have been across the entire customer journey, from unknown visitor to paid customer. With that information, a marketing manager can confidently report on overall campaign effectiveness, and not lose her sanity over it.
- The need for deep insights into marketing spend and ROI.
The disconnect between spend and ROI is glaring and painful. With a comprehensive view of performance that’s tied directly back to your budget, you can better understand the return on your marketing investments to improve quarter over quarter.
- The need for more efficient analytics processes and automation.
Most marketing teams miss the mark because they stop at the dashboard and either don’t or cannot take the actions necessary to change what the dashboard shows them.
A CDP not only automates reports, but can also automate many of the marketing actions between systems. From fixing data in Salesforce, to segmenting a Smart List in Marketo, to matching an ABM audience for a LinkedIn campaign, CDPs can write data back to other systems. This is not something a homegrown data warehouse can do.
“Insights with no action is an academic exercise and a waste of money.”
– Allison Snow, Senior Analyst, Forrester
Here’s what it takes to build and maintain an in-house system, if your IT teams are up for the challenge.
If the primary goal of your project is to be able to support deep marketing analytics, then we should begin by defining exactly which types of reports would be required. For this exercise, let’s assume that the focus is on showing these:
- Data Management
- Marketing Analytics & Attribution
- Return On Ad Spend
- Customer Engagement
- Funnel Performance
For attribution, the system needs the ability to handle multi-touch attribution models and optionally support deeper machine learning models down-stream (Chain-Based Attribution). The second goal is to be able to orchestrate and push segments back out to the systems of engagement. To simplify things, let’s assume there is a requirement to push segments of users into the marketing automation platform and CRM based on receiving intent signals.
Your build-out would look something like this:
The devils are in the details of each of these. Keep in mind that IT teams aren’t made up of marketers (for good reason). The ongoing analysis, maintenance, and fluid nature of AI is paramount to an effective CDP. Keeping up with constant API changes from your CRM, MAP, Web Analytics systems, and other vendors presents persistent challenges and demanding dedication.
In addition to the IT team, you’ll need a number of resources to bring it all together at various points:
- Marketing Ops team
- Data Analyst
- Database/SQL developer
- Marketing Demand team
- Business Analyst
- BI tool expert
- Marketing Leadership
Where does machine learning fit in? The holy grail of having all this data is to be able to get insights and understandings about the behavior of your customers and prospects, and act upon it quickly and efficiently. Machine Learning is the next evolution of marketing analytics, and promises to allow you to turn that data into activities and actions.
To support these algorithms, you need to have your data set up in a way in which the algorithms can learn from, and that supports meaningful outcomes.
If you are building your own platform and wish to do machine learning down the line, there are requirements over and above what I just listed that would expand the scope of the project and require additional resources that understand machine learning and data pipelining.
For a detailed step-by-step on the DIY route, take a look at our full CDP Build or Buy Guide.
Using a Customer Data Platform (CDP)
The alternative to building out your own platform is to use a third-party vendor to deploy an application with similar capabilities. A CDP is, at its core, a data warehouse with tools built around it to support everything from segmentation to deep data analysis.
There are different vendors in the CDP space and all have different capabilities. It is worth evaluating the vendors to make sure their capabilities align with your core business objectives.
The steps involved with using a pre-built, fully customized CDP are far more simple and faster to deploy, with the critical added component of a dedicated team of experts in marketing analytics, BI, AI, and business analysis to supplement the tool itself.
Here’s how an in-house build and a SaaS CDP compare:
A CDP positions the business for revenue acceleration. It is the data and analytics bridge between marketing and business, and between the CMO and the CFO/CEO and the board.
Data is power only if you can make sense of it. Marketing’s time has come to prove itself as the function toward that end.