Data & AI · Analytics
Analytics where every metric has one agreed definition
We design the semantic model between your curated data and your reports: metric definitions, dimensions, security rules and refresh behaviour. Board reports and team dashboards then read the same governed numbers, and your analysts keep and extend the model.
The problem
Where this usually breaks down
Same metric, different numbers
Finance, sales and operations each calculate revenue or churn slightly differently. Meetings start by reconciling figures instead of deciding anything, and trust in every report erodes a little each time.
Logic buried in each report
Calculations are copied into individual reports and workbooks. A definition change means finding and editing every copy, and the ones that are missed keep showing the old answer.
Security rebuilt per dashboard
Who may see which region or customer is enforced report by report, if at all. One report published without the right filter exposes data its audience was never meant to see.
What we do
What we set up, and how each part works
We work in your repositories and through your change process, so everything we set up can be reviewed, run and changed by your engineers after we leave.
A semantic model over curated data
We model facts, dimensions and relationships once, over the curated or serving zone, as a Power BI semantic model, a Looker model or metrics defined in dbt. Reports connect to the model, not to raw tables.
Metric definitions with named owners
Each metric has a written definition, a calculation in the model, an owner in the business and a change process. Revenue, margin or retention then mean the same thing in the board pack and in a team dashboard.
Security defined once in the model
Row-level and object-level security live in the semantic model and map to your identity groups, so every report on the model inherits the same rules. An access review checks one model instead of every report.
Reports designed for their audience
Board reports lead with a small set of governed metrics and their trend. Team dashboards go deeper into operational detail. Both follow one layout and naming standard, so readers know where to look.
Refresh and query performance planned
We choose import, direct query or cached aggregations per table, schedule refreshes after the upstream loads finish, and test the model against realistic query patterns before reports depend on it.
Self-service within clear boundaries
Analysts build their own reports on endorsed shared models. Separate workspaces and a promotion step keep personal experiments apart from the reports the business relies on.
What you keep
What stays with your team when we finish
Everything is committed to your repositories or written into your documentation, under your access controls. Nothing depends on us to keep running.
The semantic model as code
The model definition in source control, with its measures, relationships and security roles reviewed and deployed through a pipeline.
A metric catalogue
Every governed metric with its definition, calculation, owner, source tables and the reports that use it.
Report templates and a layout standard
Templates for board and team reports, with the naming, layout and accessibility rules that every new report follows.
Runbooks for change and access
How to propose a metric change, add a report, grant access through the model and investigate a figure that looks wrong.
How we work
How an engagement runs, phase by phase
Each phase ends with something your team reviews and signs off before the next one starts. We work inside your change control, not around it.
01
Agree the metrics that matter
We work with finance and the business owners to list the metrics behind real decisions, record how each is calculated today, and settle one definition and one owner for each.
02
Design the semantic model
We design facts, dimensions, relationships, security roles and refresh behaviour over the curated data, and test the model against the questions your reports must answer.
03
Rebuild the priority reports
The reports that matter most are rebuilt on the model and compared with the current versions. Where the numbers differ, we trace why and record which definition stands.
04
Open it up and hand over
Analysts get access to build on the model, with the promotion path in place. Your team owns the model, the metric catalogue and the change process from then on.
Technology
The tools this work runs on
Each group says whether a tool appears in our published work, so you can tell what we have shown from what we offer.
Offered, not in published work
- Power BI
- Looker
- Data Studio
- Microsoft Fabric
- dbt
- BigQuery
- Snowflake
- Azure Databricks
Questions
What teams ask before we start
We have hundreds of reports. Do we rebuild them all?
No. We start with the reports behind real decisions and move those onto the semantic model first. Many of the rest turn out to be duplicates or unused, and usage data shows which can retire rather than move.
Power BI or Looker: which should we use?
Usually the one your organisation already licenses and knows. Power BI suits estates built around Microsoft Fabric, and Looker suits teams centred on BigQuery. The semantic model and the metric definitions matter more than the tool on top.
Who decides when a metric definition changes?
The metric owner named in the catalogue, usually someone in finance or the business function concerned. The change is proposed, reviewed and deployed through the model’s pipeline, and the catalogue records what changed and why.
Can our analysts still build their own reports?
Yes, and that is the point. They build on the shared model instead of copying data into their own files. Anything the business relies on moves through a promotion step, and personal work stays in personal workspaces.
Related services
What teams often need alongside this
Data engineering and pipelines
Contracted ingestion, tested transformations in dbt or Apache Spark, and lineage at every stage, run from version control through one deployment path.
Data platform architecture
A lakehouse or warehouse organised into landing, curated and serving zones, with data contracts, access policy and lineage your team owns.
AI solutions on governed data
Retrieval-based AI applications on your governed data, with an evaluation set that gates each release and monitoring that feeds back into it.
Next step
Plan your semantic model with us
Tell us which metrics your leadership debates and which tools your analysts use. We will come back with where to start and which definitions to settle first.