An organization managing multiple marketing programs was struggling to reconcile performance across its channels.
The question seemed straightforward: if the business knew how many sales it had made, why couldn’t it see how many came from search, social, and other marketing investments?
Every advertising platform had its own version of the results. Those numbers didn’t add up to a consistent picture, and substantial investment in television and radio was even harder to account for.
What appeared to be a reporting problem was actually a much broader measurement challenge.
Challenging the question to find a better answer
A dashboard could bring the numbers into one place, but it couldn’t make competing attribution claims agree—or establish the contribution of channels without a direct, observable connection to a sale.
We worked with the organization to rethink what a useful answer would look like.
Rather than promise perfect attribution, we outlined an approach combining complementary forms of measurement: marketing mix modeling to understand broader channel contribution, controlled experimentation to assess incremental impact, and more sophisticated attribution where the available data could support it.
Each method would help answer a different part of the question. Together, they offered a path toward a more credible understanding of how marketing investment contributed to business results.
But the organization didn’t yet have the information infrastructure needed to support that ambition.
Building the capabilities the vision depended on
Marketing information was scattered across platforms and programs. Reporting required manual work, definitions weren’t always consistent, and source information wasn’t being systematically preserved for deeper analysis.
Before introducing more sophisticated measurement, we needed to make the underlying information dependable.
We designed and built a centralized data environment, connecting marketing and analytics sources through automated pipelines and establishing consistent, reusable data models across programs.
Importantly, the architecture preserves granular source information rather than limiting the organization to the summarized metrics needed for today’s dashboards. That creates the flexibility to investigate discrepancies, revisit assumptions, and pursue deeper analysis.
During validation, we also identified duplicate purchase events that were overstating reported revenue. We accounted for the discrepancy in the reporting models, preventing the known duplication from carrying through into the new reporting environment.
Starting with what makes the bigger picture possible
The resulting foundation automates much of the collection and preparation of marketing data, establishing a more dependable basis for reporting across the organization.
The broader measurement vision remains future work. Marketing mix modeling, experimentation, and attribution will each require additional design and validation before they can support investment decisions.
But the foundation is now in place to pursue those questions with better information.
The immediate deliverable was a reliable data environment. The larger purpose was to make a more meaningful understanding of marketing effectiveness possible.