Beyond the tech: Mastering customer data with a modern approach

Build a foundational “Customer 101” dataset and unleash its potential through enrichments and a data-driven learning cycle.

The modern customer data strategy

When many organizations contemplate the path forward for their martech stack, they quickly embrace a modern data stack as the foundation. They are anchored on a Cloud Data Warehouse and use tools and technologies that can scale and have interoperability within a composable architecture.

The benefits of a modern data stack are justified and broad, such as speed to market, agility and lower costs from limited operations. However, using a modern data stack for your customer data won’t realize the value you are expecting if you’re not taking a modern approach to using your customer data.

‘Customer 360’ is misguided 

Everyone says it and seems to want it. Too many martech platforms and organizational initiatives have “360” in their name. For too long, we’ve been made to think that personalization is the primary way to drive growth and that having a 360-degree view of your customers is the best way to enable it. 

Customer 360 is a nice aspiration, but it is an irresponsible pursuit. It just costs too much time and budget to accomplish — not to mention the opportunity cost of missed growth opportunities that could have come from applying effort in other ways. 

If you have the right customer data strategy, you’ll never come close to a Customer 360 because you will get distracted as you follow the data. There’s simply too much to learn. The best customer data strategy is based on a learning agenda. Identifying clear learning objectives that align with achieving measurable business outcomes will guide the way. When planned properly, this approach should get in the way of achieving the dream of a 360-degree customer view, and the results you drive will leave you wondering why you had that dream in the first place. 

We recently saw this approach work for a travel company with limited capacity within their salesforce to convert all the leads sourced with their upper funnel tactics. Instead of working toward a more complete profile of prior guests and historical interactions, they focused on understanding the propensity to book. They re-engineered their lead scoring model with machine learning (ML) based on the basics — customer profile, web interactions and conversion data.

This approach replaced their traditional, overly complex approach to lead scoring within their marketing automation platform. It allowed them to invest their sales efforts better, leading to operational efficiencies and more bookings.


Build your Customer 101 first

A college 101-level course is an introductory course that focuses on the basics. Your approach to building out a customer data store should be similar — solve for the basic data needs:

Use the right data

Forget about having all the data about your customers from every interaction with your brand across all channels. The right data includes:

  • The core profile characteristics of your customers.
  • Interactions from the most common touchpoints.
  • The behavior most associated with your most important metrics that drive growth (i.e., a new lead, a repeat purchase) or metrics that prevent growth (i.e., churn). 

Data management best practices

Establish core capabilities as part of your foundation. Implement sound practices on day one for cataloging your data, data governance, data lineage and data privacy.

Create great data

Using the right data will only be effective if that data is great. Ensuring your data environment applies upfront and ongoing data quality practices is paramount if your data is going to be consistent, usable, and, most importantly, trusted.


Identity resolution is the linchpin

A key assumption of any customer data store is that you are solving for a single customer view (SCV). SCV can not be created without applying identity resolution to connect and deduplicate customer data sources. The level of sophistication required can vary greatly depending on the nature of your data:

  • How and where customers identify themselves to your organization. 
  • Variety of identifiers that present themselves across your data sources — terrestrial vs digital, email address only, etc. 
  • Presence of key PII gaps within data sources.
  • Level of standardization and cleanliness of data.

Identity resolution (IDR) can be confusing, and the market providers are just as varied.

IDR 101

No matter the approach you take to solving for identity, the solution must handle the basics and be able to do the following:

  • Construct and persist an identity graph.
  • Adjust and expand the identity graph and matching rules as you add new sources and identifiers.
  • Generate and assign a persistent ID to uniquely identify a person/user/customer.

There are multiple approaches to solving IDR. Examples include leveraging an ID Spine provider, an MDM platform, a packaged CDP or even rolling your own ID graph as part of a custom solution. 

Enrich your Customer 101

Instead of working toward a Customer 360 view and integrating all your customer data, enhance your Customer 101 datasets by finding creative ways to use it through enrichment approaches. The best customer data strategy is grounded on a learning agenda that uses enriched data. 

Start by identifying the ways you can enrich your core customer data assets, such as:

  • Developing predictive models using machine learning to generate new segments. Examples include purchase propensity, churn propensity, next purchase and channel propensity.
  • Analyzing your data and generating new customer segments (i.e., high-value customers).
  • Building lookalike models based on high-value customers to extend reach to a universe of prospective customers.
  • Leveraging second-party data by sharing with partners.
  • Overlaying your customer data with third-party data sources such as demographics, psychographics, firmographics and weather.
  • Using publicly available data sets such as climate and environment, census, economics, geospatial, transportation and travel.

Use, prove and move

Too often, after adopting a new capability like a CDP or integrated customer data store, organizations look to recreate many of the customer experience strategies they have employed historically. They often lack evidence that their past strategies are the best approach.

The potential risk they see in negatively impacting legacy KPIs keeps them from trying new approaches. However, taking an immediate and careful approach to testing new strategies can quickly uncover new insights from your learning agenda.

Employing a continuous improvement process allows you to iterate on each learning objective quickly. As illustrated below, it starts by creating an enrichment, then orchestrating your program to leverage that enrichment in your testing and deriving insights from your test results and performance metrics.

Great data – Enrich, orchestrate, derive insights

Use enrichment strategies applied to your data

A typical first step is better understanding who your customers are — which are high value? Which are most engaged with your brand? To learn this, start by enriching your great data with something like a machine learning model.

Leverage the model scores to generate new segments. These segments should be deployed and made available to the areas where user communities consume consumer data:

  • As dimensions in your customer reporting and campaign performance dashboards.
  • As segments to be used wherever you build your audiences, like a CDP.
  • In product, as user attributes or dynamic values.

Prove (or disprove) your hypothesis

Be sure to set up your measurement and testing framework properly. This means creating a structured, iterative learning methodology that enables quantifiable measurement. Ensure that you:

  • Define — and align with — business metrics.
  • Document formulas for each metric for team visibility.
  • Organize metrics into distinct categories.
  • Utilize experiment design to enable actionable measurement outcomes.
  • Be sure to create a statistically significant sample size for your test.

Whether applying insights directly or testing first, it’s important to plan measurement to ensure impact can be quantified. 

Move on to the next learning objective

As you derive insights from your testing, make the required adjustments to your strategy based on those learnings:

  • Adjust the channel strategy to optimize the experiences you’re creating.
  • Tune your ML model with additional features.
  • Reduce waste in marketing efforts for non-responsive audiences.

Sometimes, your learning is complete and you move on to the next learning objective. Perhaps you learned something that requires you to explore that initial hypothesis further. Or it’s time to move on to the next learning objective in your learning agenda. Follow the data breadcrumbs. 

A typical progression could start with asking, “Who is our customer?” Subsequently, that question leads to a line of questioning that may look like the following:

  • Who are our high-value (HV) customers?
    • Who has the highest potential to become HV?
      • How do we reach more HV customers?
        • How do we acquire more HV customers?
      • Who is likely to make a subsequent purchase?
        • Which products are more likely to be purchased by HV customers?
          • Which product is each customer most likely to purchase?
  • What customers are most engaged?
    • How does their LTV compare to other cohorts?
      • Which products are they most likely to purchase?

Each question above requires another cycle of continuous improvement to build an enrichment, test it in-market, evaluate the results and move on to the next idea, as illustrated below:

High-value customers, next purchase propensity, engagement



Rethinking your customer data strategy

No team isn’t feeling pressure to get more from less. A modern approach to evolving your data asset requires returning to the basics and avoiding the distractions of 360-degree view nirvana. There is no harm in working toward creating a 360-degree view, but when this is core to your data strategy, you may miss out on other higher-value, more creative opportunities along the way. 

Taking a creative approach to using your data includes enrichments, but also looking for other ways to monetize your customer data, such as enabling sales, service and perhaps even other enterprise functions such as finance. A modern approach to your customer data means establishing it as a key capability within your organization rather than an ongoing project.


The post Beyond the tech: Mastering customer data with a modern approach appeared first on MarTech.


About the author

Craig Howard


Craig leads Actable’s solutions consulting team, developing customer data strategies and deploying technologies for Actable’s customers. Craig also oversees Actable’s industry partnerships with MarTech providers, including customer data, cloud database, and customer experience platforms. Craig has 20 years of experience delivering multi-channel enterprise marketing technology solutions. He has broad technology and industry experience, driving technical architecture solutions by providing technical expertise in strategy, design, and implementation. Prior to joining Actable, Craig led Merkle’s solution architecture team as Chief Solution Architect. He was responsible for design of marketing technology solutions for all practices as well as providing thought leadership, innovation, and architectural standards to enable solution deployments. Prior to joining Merkle in 2009, Craig served as Director of Business Intelligence Systems and Services at a startup focused on providing marketing management to small businesses. Prior to that, he served as Director, Marketing Technology at Epsilon/DoubleClick, acting as Chief Architect for an interactive database marketing solution, enabling advanced segmentation and reporting capabilities specific to online channels. Craig began his career with Accenture and Braun Consulting, serving as a Manager within the CRM and Database Marketing consulting groups. Craig holds a Bachelor of Science in Civil Engineering from the University of Illinois at Urbana-Champaign.