Online, offline, point of sale, attitudinal, media, advertising — bringing together all of these disparate data sources to create a cohesive, accessible whole to produce actionable insights requires dealing with a tremendous amount of ambiguity. What data source to use? Which records to filter out? Which dimensions should be included for segmentation and drilldowns?
When multiple independent actors make these decisions without centralized guidance, logic will start to wobble.
In any large organization, translating data into action requires buy-in from many parties. That makes getting all of your decision makers on the same page equally as important as clean data capture. It doesn't take much variation between teams before debate is closed off and fundamental differences in worldview start to develop.
Without a common foundation and a unified worldview, data driven decision making quickly devolves into the default power dynamics of your organization.
In a worst case scenario, foundational elements simply don’t exist, or the foundational elements themselves are disputed ground.
How do large organizations and brands avoid logic wobble in their data?
Start Speaking the Same Language
Metric and element naming conventions should never be an afterthought. Common terms like "activities," "conversion" and "users" often have different meanings across teams. Begin the process with definition management and naming convention standardization.
Terms used to define data entities need to have distinct, explicit meanings that combine in unambiguous ways to create more complex entities.
For example, if you have seven different "conversion" metrics, each full name must differentiate itself from the rest (e.g., "registration conversion," "purchase conversion," etc.). Business concepts rarely translate directly into the data environment, so after settling on high level definitions in business terms, I suggest building out sub-definitions that note the technical compromises required context by context.
You may not be able to achieve literal parity, but a rigorous definitions management process allows you to counter "apples vs. oranges" assertions with a "gala apples vs. Fuji apples" response.
Make Your Data Worldview Attractive and Accessible
Your common foundation has to be the easiest way to get at your organization’s data. If your access problem isn’t solved, people will continue to get their data elsewhere, and the logic wobble cycle will continue.
This is both a technical and an organizational problem. You need the right data model powering a robustly featured reporting front end that has been adequately socialized.
Deeper investigations will require walking back up the data lifecycle. I recommend creating and managing intermediary data structures between your source data and the high level aggregates that power your reporting solution. As you build out these levels of data access within your data lifecycle, it is essential to understand who is capable of interacting at what levels of depth.
A data scientist can handle source data access, but there are other people who can only handle pictures. If you grant full access to everyone who asks, you’ll end up in a world of confusion. As analysts explore at depth, make it a requirement that each analysis relate itself back to the common foundation. Different decisions may be appropriate across contexts, but explicitly relating those decisions to your foundation reduces their potential to create confusion.
When onboarding new employees, start them on the reporting front end and move them back through the data lifecycle step-by-step. Echoing the analytical process above, require that they understand how each step in the data lifecycle relates to the next. First, they must be able to recreate the reporting front end’s numbers from the reporting data mart. Then, make them responsible for recreating the reporting numbers from intermediary data structures.
Ride the Pendulum Between Discovery and Chaos
Deeper investigations drive change. In a healthy data lifecycle, a pendulum swings back and forth between enabling discovery and reigning in chaos. To avoid worldview drift, plan to periodically integrate successfully vetted analyses back into the common foundation. If you follow this process, you will take the logic that previously lived all over your organization — spread among teams, living in personal code, shared via email and wobbling slightly in each independent application — and make it documented, formalized and accessible.
Only then will your organization be able to realize the true actionable power of all your cross channel marketing data.
About the Author
Dave De Noia lives in the balance of chaos and order inherent to working with data. Dave worked previously with Microsoft and Redfin and now serves as the discipline lead in data management at Pointmarc, where he helps some of the world’s largest brands get value from their data. Naturally functioning as a bridge between business and technical teams, Dave’s professional passion lies at the intersection of data and people.
- Extracting Insight from Unstructured Data
- Box Cops to Bad IPO Timing, It's Time to Unbox
- Are You Too Old to Work in Tech? IT's Midlife Crisis
- Big Data is Getting Smaller and Smarter
- Who Are the 100 Fastest Growing Software Companies?
- Chaos Reigns at Content Management Vendors
- B2B Marketers: Think More Like Brand Marketers