Dashboards that decision-makers trust.

Overview: federal BI is a trust problem

A federal dashboard is not a dashboard. It is an evidentiary artifact that shows up in Congressional testimony binders, Inspector General reports, Government Accountability Office studies, budget justifications, and performance-review materials that shape next year's appropriations. The people who stare at it do not care about the chart type — they care whether the number is defensible, whether it reconciles to the authoritative system of record, whether the methodology can withstand scrutiny from hostile staff, and whether the GS-15 presenting it will be blindsided next Tuesday.

Commercial BI builds rarely operate under that pressure. Precision Federal builds BI that does. Every metric has a documented definition in Git. Every dashboard traces back to source columns via lineage. Every user sees only what they are authorized to see. Every visualization meets Section 508 / WCAG 2.1 AA. Every refresh is logged for audit. Our SAMHSA production work is where this discipline was forged.

Our technical stack

• Enterprise BI: Tableau Server (on GovCloud or Azure Gov), Tableau Cloud (FedRAMP Moderate), Microsoft Power BI (GCC, GCC High, DoD), Power BI Premium, Looker (enterprise and embedded), Qlik Sense Enterprise.
• Open-source BI: Apache Superset, Metabase, Apache Zeppelin, Grafana (for ops-style dashboards), Redash.
• Semantic layers: LookML, dbt metrics, Cube, Malloy, Power BI datasets, Tableau published data sources, AtScale.
• Embedded BI: Tableau embedded analytics (JWT, connected apps), Power BI Embedded (Azure AD tokens), Looker embed (signed URLs), Sigma, Mode.
• Visualization frameworks: D3.js, Observable Plot, Vega-Lite, Plotly, deck.gl (for geospatial), Apache ECharts — when off-the-shelf BI cannot meet the mission need.
• Accessibility tooling: NVDA, JAWS, VoiceOver for screen reader testing; axe-core and Pa11y for automated audits; WAVE for manual review.
• Warehouse connectivity: Snowflake, Redshift, BigQuery, Synapse, Databricks SQL, Trino, Postgres, Oracle, SQL Server. See data warehousing.

Federal use cases

• SAMHSA behavioral health dashboards (confirmed past performance): production reporting over treatment admission and survey data with governance and Section 508 conformance.
• Army readiness dashboards (pursuing): unit readiness, platform availability, personnel status with mobile-friendly views for commanders.
• Navy fleet availability and logistics (pursuing): ship status, depot throughput, sparing position at fleet and type-commander level.
• Air Force sortie generation and base operations (pursuing): daily operational dashboards for wing commanders.
• FBI investigative dashboards (pursuing): case metrics, workload management, regional heat maps with strict need-to-know filtering.
• HHS / CDC public health dashboards: outbreak surveillance, treatment capacity, public-facing dashboards meeting Plain Writing Act standards.
• Treasury budget execution and obligations: OMB-aligned dashboards for CFO Council consumption.
• GSA federal buying and spend analytics: agency-facing dashboards over GSA Advantage, SAM.gov, and FPDS data.
• VA veteran service delivery: claims throughput, wait-time, and outcome dashboards.
• DHS operations dashboards: border throughput, case processing, enforcement activity with CUI-appropriate views.

Reference architectures

Architecture 1: Tableau Server on AWS GovCloud with Snowflake backend

Tableau Server deployed in a multi-node cluster on EC2 GovCloud behind an internal ALB. Active Directory / Okta Federal federation. Snowflake Government as the primary data source via FIPS TLS. Tableau extracts tuned for refresh schedule and storage cost. Published data sources as the semantic layer with row-level security tied to AD groups. Tableau Prep for data shaping when dbt does not reach. CloudWatch and a Splunk forwarder shipping Tableau logs to the agency SIEM. Section 508 conformance testing with NVDA and JAWS in every release cycle. All traffic inside a VPC with PrivateLink to Snowflake.

Architecture 2: Power BI on Azure Government with Synapse and Fabric

Power BI Premium capacity in Azure Gov or GCC High tenant depending on workload classification. Synapse or Microsoft Fabric as the warehouse layer. Power BI datasets as the semantic layer with DirectQuery to Fabric and aggregation tables for performance. Row-level security via Entra ID Gov groups. Power BI embedded in an internal Next.js portal via Private Link. Automated dataset refresh with failure alerts routed to Teams in GCC High. Sentinel ingesting Power BI audit logs. Section 508 conformance baked into report templates. Best fit for M365 GCC High agencies.

Architecture 3: Looker + dbt semantic-layer-first

Looker hosted in the agency's chosen FedRAMP-authorized compute path. dbt as the pipeline layer, LookML as the semantic layer, with every metric defined once and referenced everywhere. Embedded Looker in agency apps via signed URLs. Scheduled Looks for routine reporting; Looker API for programmatic consumption. Tight integration with Git for LookML change control. Strong fit when dashboards serve both human decision-makers and API consumers.

Delivery methodology

1. Discovery: interview decision-makers and analysts, inventory existing reports, identify mission decisions the BI is supposed to enable.
2. Semantic model: metric definitions, dimensional model, row-security model. Documented in Git.
3. Prototype: first three dashboards on a thin slice of the semantic layer. User acceptance with actual decision-makers.
4. Accessibility audit: Section 508 / WCAG 2.1 AA at prototype stage, not at delivery.
5. Build & deploy: full dashboard suite, embedded where needed, permissions wired to agency SSO.
6. Performance validation: p95 load time < 3s, refresh success > 99%.
7. Training & handover: analyst training, author training, runbooks for refresh failures.
8. Iterate: monthly dashboard health review, usage analytics, content pruning.

Engagement models

• SBIR Phase I — $150K-$250K, BI-focused prototype.
• Fixed-price dashboard suite — 60-120 days, $75K-$300K.
• Migration project — legacy Cognos/MicroStrategy/BusinessObjects to modern BI, 6-18 months, $300K-$2M.
• T&M under prime on CIO-SP4, Alliant 2, GSA MAS IT.
• OTA prototype for DoD rapid delivery.
• Staff augmentation — embedded Tableau / Power BI / Looker developer.

Maturity model

• Level 1 — Ad-hoc: every analyst builds their own dashboard on their own query.
• Level 2 — Shared: central server, published data sources, versioning.
• Level 3 — Governed: semantic layer, row-level security, 508 conformance, lineage to source.
• Level 4 — Mission-integrated: embedded analytics in operational apps, mobile-first, API exposure.
• Level 5 — Continuously assured: automated accessibility testing, usage analytics driving pruning, continuous ATO evidence.

Deliverables catalog

• Dashboard source files (Tableau workbooks, Power BI PBIX, LookML repos).
• Semantic layer (Tableau data sources, Power BI datasets, LookML models).
• Row-level security model wired to agency SSO.
• Section 508 conformance report with remediation evidence.
• Data dictionary and metric definitions in Git.
• User guides and decision-maker playbooks.
• Author / analyst training materials.
• Refresh monitoring and alerting runbooks.
• NIST 800-53 control narratives (AC, AU, SC).
• Usage analytics dashboard for the BI platform itself.

Technology comparison

• Tableau: best exploratory experience, strong federal standardization, rich visual grammar, premium price.
• Power BI: best value in M365 agencies, native to Azure Gov, strong DAX modeling.
• Looker: best semantic-layer discipline, LookML in Git, strong embedded story.
• Qlik: associative engine, strong for self-service exploration, smaller federal footprint.
• Superset: open-source, free, solid dashboards, requires engineering investment.
• Grafana: ops-first, pairs naturally with Prometheus; not for analyst-facing BI.

Federal compliance mapping

• AC-3, AC-6, AC-16: row-level security, least-privilege, attribute-based filtering.
• AU-2, AU-12: dashboard view logs, export logs, failed authentication logs to SIEM.
• CM-6: dashboard templates as configuration baselines.
• IA-2, IA-5: MFA + agency SSO, no local accounts.
• SC-7, SC-8: private endpoints, TLS 1.3 end-to-end.
• Section 508 / 29 U.S.C. 794d: WCAG 2.1 AA conformance on every deliverable.
• Plain Writing Act: public-facing dashboards using 8th-grade-readable labels and tooltips.

Sample technical approach: migrating a 14-year Cognos estate to Power BI

A civilian agency runs 680 Cognos reports against a DB2 warehouse. Support is ending, users hate the UX, and accessibility is nonexistent. Approach: scrape Cognos package metadata and report XML to build a report inventory. Cluster by business owner, criticality, and structural similarity. Prioritize the top 40 reports representing 80% of usage. Stand up Power BI Premium on GCC High tenant. Build semantic datasets using DAX that reproduces the Cognos framework model. Rebuild prioritized reports as Power BI reports with Section 508 conformance. Dual-run for 90 days, reconciling key figures. Decommission the Cognos corresponding report once parity signed off by the business owner. Train analysts and citizen developers on Power BI. Retire Cognos in 14-18 months. Typical outcome: 60% usage concentrated on 80 reports, long tail decommissioned by non-use, 30-40% license cost reduction, dramatic accessibility improvement.

Past performance

Related capabilities, agencies, and insights

See data analytics, data warehousing, data governance, ETL / ELT, responsible AI. Agency pursuits: SAMHSA, Army, Navy, VA, DHS, Treasury. Vehicles: GSA MAS IT, CIO-SP4, SBIR. Insights: Section 508 for Tableau, Tableau vs Power BI for federal, Semantic layers for federal.