Marketing analytics is custom to every business. So is your model.
The people in your marketing team who know your platforms, campaigns and conventions define the metrics and dimensions. No SQL, no engineering tickets.
What it does
Metrics, dimensions and business logic, defined once and used everywhere — dashboards, BI tools, the reporting API and AI assistants all reading the same governed definitions.
Simple on the surface: building a metric, combining data sources or creating a calculation takes minutes. Underneath, there's no ceiling. That combination is unusual — tools non-engineers can use normally hit a wall the moment the question gets specific.
Why the model has to be yours
Every marketing business is custom.
Your markets, your brands, your funnel stages, your definition of an active customer. None of it is shared with anyone else. A general-purpose BI tool gives you somewhere to put the numbers, not a way to express your business.
The usual setup is three disconnected layers.
A collection tool dumping into a warehouse, views built on top, reporting built on the views. Every layer is owned by a different team or tool, and every layer is a hardening step. A new question means going back through all three.
So the reporting stops changing while the marketing keeps changing.
When a change takes three weeks, nobody iterates — they specify. People ask for what they can justify waiting for, then live with it for a year. You don't just get slow reporting, you get reporting nobody asks questions of.
The nuance lives with the marketing team.
Naming conventions, the exceptions, why that campaign was tagged wrong for three weeks in March. A data engineer can't encode that — not for lack of skill, but because they weren't in the room. The model built by the person with the context is a better model, not just a faster one.
What's inside
Metrics with multiple definitions
Real marketing data arrives at different grains: impression-level from one platform, daily aggregate from another, weekly from a partner. Most tools make you pick a grain and lose whatever doesn't fit.
A single metric here can carry several definitions and resolve the right one for the dataset in play.
Custom dimensions built from your naming conventions
Rules that read campaign, ad set and ad names and extract the structure inside them: campaign type, audience, funnel stage, market. Dimensions that span every platform, not one at a time.
See Taxonomy Manager →Custom inputs
Targets, buy rates, forecast factors. The numbers that live in someone's spreadsheet today, brought into the governed model alongside platform data.
Metrics of limitless depth
Metrics that roll up other metrics, calculations built on those rollups. Cost-per figures spanning ten platforms, defined once.
Combine sources into new ones
Weave multiple data sources into a shape that doesn't exist in any of them — new tables built from combinations of your existing ones, without waiting on an engineer to build the join.
Custom transformations
For the shaping that needs to happen before the model sees it, via Pipeline Manager.
Edge cases aren't edge cases for us. The semantic layer has been purpose-built for marketing since 2014, extended in response to what demanding advertisers and agencies have thrown at it — and it's the same governed model your AI assistants query over MCP today.
See it in action
Two short walkthroughs of the semantic layer in use.
A full walkthrough: building metrics from raw columns across platforms through to a shared dashboard everyone reads from.
A new marketing analytics framework.
Your analysts in control, every team and tool connected to trusted data.