Your data made it into the warehouse. That's just the first step.

Getting data into one place is a solved problem for most FMCG and retail organizations. What happens to its meaning once it gets there is a different matter entirely.

You know this moment. The data has been ingested. The pipelines are running. The dashboards are populated. From the outside, it looks like the hard work is done. And then someone asks a straightforward question: which campaigns drove incremental revenue last quarter, or how a promotion performed net of cannibalization, and the system that looked complete produces an answer nobody can fully defend.

Not because the data is wrong. Not because the engineering is poor. But because the interpretation of that data, what a campaign is, what incremental means, what counts as a promotion, was never formally established. It lives in the assumptions of the analyst who built the report, in a calculation embedded in a dashboard that nobody has reviewed in eighteen months, in a spreadsheet that Finance uses to sense-check numbers that the warehouse was supposed to replace.

This is the interpretation gap. And it sits directly above the data access problem that most organizations have already solved.

Ingestion isn't interpretation

Marketing and commercial data is fundamentally unlike the enterprise data most IT teams are built to manage. ERP systems, CRM platforms, HR databases: these are internally governed, schema-stable, and built around clear ownership. Commercial data is the opposite. It's external, multi-source, and semantically inconsistent by design.

Retailer POS data arrives at a different granularity than media platform data. Brand tracker metrics use survey methodologies that differ by provider and by market. Promotion calendars don't align with media spend cycles. Financial closes happen monthly while retail data refreshes weekly. Each source was built by a vendor for its own purpose, with its own logic, its own hierarchies, and its own definition of the metrics it exports.

Standard data engineering handles the movement of this data well. It was never designed to handle the meaning. Translating a retailer's SKU taxonomy into a consistent product hierarchy across markets, aligning promotion mechanics across different trade calendars, mapping media spend to Finance-auditable cost centres: these are interpretation problems, not ingestion problems. And they don't get solved by getting data into a warehouse. They get solved, or more commonly, deferred in the layer above it.

According to the ScanmarQED 2026 Industry Report, only 10% of organizations have a fully harmonized commercial data platform. The majority have centralized their data without centralizing its meaning. One warehouse, still producing multiple interpretations.

Where interpretation goes when it has nowhere to live

In the absence of a governed place for commercial interpretation to live, it migrates. It moves into dashboard logic, where a calculation built for one purpose gets reused for another without anyone updating the underlying assumption. It moves into analyst conventions, where the person who built the original report made a reasonable call about an edge case, and that call became the de facto standard without anyone formally agreeing. It moves into spreadsheets, where Finance maintains its own version of the numbers because it can't reconcile the warehouse output to the general ledger.

Each of those migration points creates its own version of the truth. None of them are visible to the people relying on the output. And none of them are governed, which means they change without notice, differ across teams, and accumulate over time into a body of commercial logic that no single person in the organization fully understands.

The result isn't a data problem in the conventional sense. The pipelines are intact. The dashboards are live. But the organization is making commercial decisions on interpretations that were never formally sanctioned, and in many cases, never even consciously made.

Only 53% of organizations combine data sources at all, and many of those do so manually. Where integration exists, the semantic layer that makes it interpretable is typically the piece that was never built.

The cost of unsanctioned assumptions

When interpretation is distributed rather than governed, the costs are predictable. Planning cycles slow down because teams can't agree on the numbers underpinning the plan. Analytical work gets repeated because different functions are starting from different baselines. Leadership confidence erodes because the outputs of expensive analytics investments can't be consistently explained or defended.

But it also limits what organizations can responsibly do with their data. Forecasting requires that the definition of what's being forecast doesn't shift between cycles. AI initiatives require data that is interpretable and auditable, not just available. The interpretation gap doesn't just affect today's reporting. It puts a ceiling on every analytical ambition that depends on the data beneath it.

Making interpretation a managed capability

The organizations that close this gap make a deliberate architectural decision. They stop treating interpretation as something that happens implicitly, buried in code and analyst convention, and start treating it as a managed capability. That means establishing a commercial semantic layer: a governed, shared set of definitions, hierarchies, and calculation logic that sits above the source systems and below the analytics outputs.

This layer doesn't require rebuilding what's already in place. It works alongside existing infrastructure, formalizing the interpretation that was previously scattered and making it visible, auditable, and maintainable as the business evolves. New data sources extend the model. Definition changes are governed rather than ad hoc. And the outputs that flow from it can be defended across Marketing, Finance, and IT because everyone is working from the same agreed logic.

It's a shift from data access to data readiness. And it's the step that most organizations have yet to take.

PulseQED is built around this principle. Its commercial semantic layer centralizes interpretation, governs definitions, and ensures that the meaning of commercial data is as consistent and auditable as the data itself. The result is analytics that organizations can act on, not just report from.

--------------------------------------------------------------------

For a deeper look at why the interpretation gap persists and what a governed commercial truth layer looks like in practice, the PulseQED white paper sets out the full picture: From DIY to Trusted Commercial Truth: Building a Data Foundation that Scales.

Written for stakeholders across Marketing, Finance, and IT. Download the white paper →