Definition
- Plain-language meaning
- Formula and filters
- Grain: per order, per customer or per day
- Known exclusions
Business Intelligence
Business intelligence implementation services
When finance, sales and operations bring different numbers to the same meeting, the problem is not the charts. Agnotic Technologies implements business intelligence with one governed definition per metric and security built into the model, so every report agrees.
Semantic layerMetric registerRow-level securityReport migration
Trusted by global innovators
Business intelligence
What do business intelligence implementation services cover?
Business intelligence implementation is setting up reporting so every team gets the same number for the same question. Agnotic Technologies builds a governed semantic layer with one definition per metric, row-level security, the right deployment model and a migration of legacy reports, on the BI tool you have chosen.
Dashboard screens and chart design are covered by our dashboard development service, and the pipelines and warehouse underneath by data engineering. This page is the layer between them, where numbers are defined and protected.
We work in Power BI, Tableau, Looker and other tools. We do not resell licences or rank vendors.
Decision table
Which BI deployment model fits your organization?
Most organizations mix two of these. The choice decides who owns definitions and where security is enforced.
| Deployment model | Who owns metric definitions | Security model | Licensing driver | Typical failure |
|---|---|---|---|---|
| Central governed | A BI team builds and publishes every report | Roles managed by the BI team in the semantic model | Viewer licences for most users, a few authors | The backlog grows and teams build shadow spreadsheets |
| Governed self-service | The BI team owns certified metrics; analysts build reports on top | Row-level security in the shared model, workspace roles for authors | More author licences across departments | Analysts redefine metrics in their own reports and numbers drift |
| Embedded analytics in your product | Your product team, per customer-facing metric | Tenant isolation enforced on every query, often through embed tokens | Capacity or usage-based pricing for embedding | One tenant sees another's data because a filter was applied only in the front end |
Metric register
What goes in a metric register?
Every certified metric gets an entry before it appears on a report.
Ownership
- Business owner who approves changes
- Technical owner in the BI team
- Change log with dates
Lineage
- Source tables and pipelines
- Refresh schedule and last refresh time
- Upstream data quality checks
Access
- Who can see it
- Row-level rules that apply
- Columns hidden from some roles
Security
How should row-level security work in BI?
- Rules live in the semantic model or the warehouse, never only in report filters a user can remove
- Roles map to groups in your identity provider, not to named users
- In Power BI, RLS applies to Viewer access, so workspace Admin, Member and Contributor roles stay with a small group
- Object-level security hides whole columns, such as salary or diagnosis fields, from roles that do not need them
- For health data, views follow the HIPAA minimum necessary standard so each role sees only what its job requires
- Every rule has a test user per role, checked after each model change
Report migration
How do we migrate legacy reports?
01
Inventory
Example: Hundreds of reports on an old reporting server and in shared spreadsheets
Approach: Collect usage logs and owners. Reports nobody opens are retired, not rebuilt.
02
Define once
Example: Revenue calculated three ways in finance, sales and operations
Approach: Agree one definition per metric with its owner and record it in the metric register.
03
Rebuild on the semantic layer
Example: A monthly pack assembled by hand from exports
Approach: Rebuild it on shared measures so it refreshes on schedule.
04
Reconcile
Example: The new dashboard shows a different total from the old report
Approach: Compare old and new numbers for the same periods and explain every difference before sign-off.
05
Switch off
Example: Old and new reports both in use with no end date
Approach: Set a date, redirect links and archive the old reports once owners sign off.
Analytics consulting
What does an analytics assessment cover?
Before building anything, we find out what people need to decide and whether the data can support it.
Source inventory
Systems, owners, refresh frequency and known data quality problems.
Metric conflicts
Every metric defined differently across teams, with the reports that use each version.
Use-case backlog
The decisions people want to make, ranked by value and by how ready the data is.
Tool and model fit
Whether your BI tool and deployment model fit, judged on security model, embedding, governance, skills and licence structure, without ranking vendors.
Roadmap
The order for the semantic layer, migrations and new reports, with the first release defined.
Who do we implement business intelligence for?
Finance and operations leaders
Monthly packs and KPIs that come from one model instead of spreadsheets.
SaaS companies embedding analytics
Customer-facing reports with tenant isolation enforced on every query.
Healthcare organizations
Reporting on PHI with role-based views built around minimum necessary access.
Companies after a merger
Two reporting stacks and two sets of definitions brought into one.
Teams on an old reporting server
Legacy reports moved to a current BI tool with reconciled numbers.
References
Platform documentation we build to
Power BITableauLooker45 CFR 164.502(b)Identity provider groups
Facts and sources
Facts behind governed BI
Vendor documentation and one regulation that shape how we build semantic layers and security.
Power BI row-level security restricts data only for users with Viewer permissions; it does not apply to workspace Admin, Member or Contributor roles.
Source: Microsoft Learn: row-level security with Power BILookML is the language Looker uses to create semantic data models that describe dimensions, aggregates, calculations and relationships, and Looker writes its SQL queries from that model.
Source: Google Cloud: What is LookML?Tableau supports several row-level security methods inside and outside Tableau, the simplest being a user filter that maps users to values manually.
Source: Tableau Help: overview of row-level security optionsWhen using or disclosing protected health information, a HIPAA covered entity or business associate must make reasonable efforts to limit it to the minimum necessary for the intended purpose.
Source: 45 CFR 164.502(b) (eCFR)
Scope and cost
What drives the scope of a BI implementation?
We do not publish prices. These factors decide the size of the work.
Metrics and sources
How many certified metrics and source systems the first release covers.
Reports to migrate
The count left after unused reports are retired.
Security rules
Row-level and column-level rules, and how many roles they cover.
Deployment model
Central, self-service, embedded, or a mix.
BI tool
Working within your current tool, or moving to a new one.
Data readiness
Whether the warehouse already holds clean, modelled data.
Frequently Asked Questions
It depends on how many metrics, sources and reports the first release covers and how ready the warehouse is. The assessment produces a roadmap with the first release defined, so you see working reports early.
We do not rank vendors. We compare the tools you are considering on security model, embedding needs, governance features, your team's skills and licence structure, and most teams keep the tool they already have.
It is a shared model that holds metric definitions, table relationships and security rules. Every report reads from it, so a change to a definition reaches every report at once.
Yes. We build the semantic layer on your current warehouse and migrate existing reports onto it, reconciling the numbers before the old reports are switched off.
Analytics consulting decides what to measure and why, through the assessment. BI implementation builds the governed model, security and reports that deliver those measures. We usually do both in one engagement.
This service sets up your BI so your team can run it. For healthcare organizations that want monthly analytics delivered as a service, we offer insights as a service.
AI tools may draft DAX or SQL measures, but each measure is reconciled against known totals and approved by its business owner before it is certified.