NCompas builds self-service BI platforms, governed semantic layers, predictive analytics, and AI-native conversational analytics on Power BI, Tableau, Looker, Databricks, and dbt — so every business decision is backed by the right number, not the nearest one.
Why NCompas
The highest-performing analytics platforms share one trait — every number has a certified definition, a business owner, and a clear decision it informs. That's what we build.
Most analytics projects start with building dashboards. We start with the semantic layer — defining every business metric once, certifying it, and then building dashboards on top. Every metric is consistent across every tool from day one.
Self-service BI fails when the data isn't trusted and the tool isn't trained. We design the governance model, certification workflow, and analyst training programme that makes self-service a cultural reality — not just a software deployment.
We map every analytical deliverable to a business decision before we build it. Who owns this KPI? What decision does it inform? What SLA does it need? Analytics without a business owner becomes shelfware within six months.
We integrate LLM-powered natural language querying, AI narrative generation, and automated anomaly explanation from the start — not as a bolted-on feature after the platform is built. Your analysts ask questions in English; the platform surfaces answers.
Row-level security, data masking, metric certification workflows, version control for reports, and automated usage monitoring. Enterprise analytics governance that doesn't slow down the analysts — it builds confidence in the numbers.
Dashboards that load in under 2 seconds at 10,000 concurrent users. We design query layers, caching strategies, aggregation tables, and materialization patterns so your analytics platform performs when the business needs it most — Monday morning at 9am.
Analytics maturity spectrum
Most organisations are stuck at descriptive. We architect the full analytics stack — from trusted dashboards to AI-native autonomous insight surfacing — at the pace your business and data maturity support.
Statistical models and ML that forecast demand, predict churn, estimate lifetime value, and surface early warning signals before business impact. Turns hindsight into foresight at scale.
Dashboards, reports, and KPI scorecards that tell you what happened in your business — accurately, consistently, and fast. The foundation of trust in your data. Most organisations think they have this; few have it done right.
Root-cause analysis, drill-down exploration, and anomaly detection that explain why a metric moved. Reduces the time your analysts spend chasing numbers from days to minutes.
Statistical models and ML that forecast demand, predict churn, estimate lifetime value, and surface early warning signals before business impact. Turns hindsight into foresight at scale.
Optimisation models and decision engines that don't just predict outcomes — they recommend the best action to take. Pricing optimisation, inventory allocation, resource scheduling. The highest-value analytics layer.
AI-driven analytics that surfaces patterns invisible to human analysts — unsupervised anomaly detection, natural language querying, generative AI report generation, and autonomous decision agents. The frontier of modern analytics.
Six core capabilities
From the semantic layer that makes every metric consistent, to the AI layer that lets anyone ask a question in English and get a reliable answer.
A single source of truth for business metrics — dbt Semantic Layer, Cube.dev, or LookML — so "revenue" means the same thing in every dashboard, every team, every tool. The foundation of trusted analytics that most organisations skip and then regret.
Power BI, Tableau, and Looker deployments that business users actually use — governed semantic layers, certified metric definitions, and training that makes your analysts self-sufficient. We don't build dashboards that only we can maintain.
A single source of truth for business metrics — dbt Semantic Layer, Cube.dev, or LookML — so "revenue" means the same thing in every dashboard, every team, every tool. The foundation of trusted analytics that most organisations skip and then regret.
Statistical models, ML algorithms, and forecasting systems built in Python, R, and SQL — deployed as scheduled models in Databricks or dbt, surfaced in your BI tool, and re-trained automatically as new data arrives. Predictive analytics as a product, not a one-off.
Analytics that update in seconds — operational dashboards, live KPI boards, SLA monitors, and fraud alerts powered by streaming data from Kafka or Event Hubs feeding into Databricks SQL, ClickHouse, or Apache Druid.
Analytics baked directly into your product or application — usage dashboards for customers, embedded reports in SaaS platforms, and product telemetry that drives feature decisions. Built with Sigma, Cube.dev, or Superset for multi-tenant security.
Natural language interfaces, AI-generated narrative summaries, and autonomous anomaly surfacing — so analysts ask questions in English and the platform returns insights, not just charts. Built on LLMs integrated with your certified semantic layer.
Platforms & tools we deploy
How we work
Six phases that take you from metric chaos to a governed, self-service analytics platform — with a production-ready dashboard delivered every two-week sprint.
Baseline your current analytics estate — tools in use, data quality, metric consistency, self-service adoption, and decision-making patterns. Identify where analytics is generating value and where it's creating noise.
Define every business metric with ownership, definition, calculation logic, and SLA. This dictionary becomes the input to the semantic layer — and the foundation of trusted analytics across every team.
Build the certified semantic layer in dbt Semantic Layer or Cube.dev — connecting your lakehouse or warehouse to every BI tool with consistent, governed metric definitions. No more metric disagreements.
Iterative sprint-based delivery of dashboards, reports, and self-service workspaces. Every deliverable reviewed with the business owner before sign-off. Analysts trained to explore and extend independently.
Predictive models, forecasting engines, and statistical analysis deployed as automated, monitored workloads — surfaced in dashboards and BI tools so insights reach decision-makers, not just data scientists.
Usage monitoring from day one — which reports are used, which are ignored, which need improvement. Monthly metric reviews, quarterly platform upgrades, and analyst coaching to keep adoption high.
Industry results
Four industries, four analytics transformations. Every metric is from a production deployment — not a projected business case.
The Challenge
14 regional managers each maintaining their own Excel reports. "Revenue" calculated differently in every one. Monthly business review preparation taking 3 full days of the analytics team. Zero forecasting capability — everything reactive.
What We Built
Power BI with a certified dbt Semantic Layer defining a single "revenue" metric, 22 KPIs certified at exec level, and self-service exploration enabled for regional managers. Demand forecasting models deployed in Databricks Serving with Power BI integration.
14 regional managers each maintaining their own Excel reports. "Revenue" calculated differently in every one. Monthly business review preparation taking 3 full days of the analytics team. Zero forecasting capability — everything reactive.
Risk analytics running in SAS on-premises — 6-hour overnight batch, results available at 9am. No intraday risk visibility. Regulatory reporting taking 18 days per quarter. Quant team building models in isolation from BI.
Clinical analytics team receiving 40+ ad-hoc data requests per week. No self-service capability — every question required a data analyst. Population health dashboards updated monthly. Zero predictive capability for readmission risk.
Product analytics split across Mixpanel, Amplitude, Segment, and a custom events DB — no unified view. Churn analysis impossible to run accurately. Customer success team had no visibility into usage health scores.
of analytics projects fail to deliver measurable business value — the root causes are almost always missing semantic layer, poor data quality, and no adoption programme (not wrong tool choice).
is the average time a business analyst spends per week re-reconciling the same metric across different reports. A certified semantic layer eliminates this entirely.
of BI dashboards are never viewed after the first week of launch — built without a defined business owner, decision use case, or adoption plan. Analytics governance prevents this.
the year 75% of business analytics queries will be initiated via natural language — conversational BI is not a future feature, it's the next 24 months of analytics roadmap (Gartner).
Insights from real-world engineering, cloud, and AI leaders - helping you make better decisions, faster.
We're putting the finishing touches on this. Check back soon for in-depth insights.
Start with a free Analytics Maturity Assessment — we'll audit your current BI stack, identify metric inconsistencies and self-service barriers, and show you what a governed semantic layer would look like for your data estate.