Skip to content

Business intelligence on the Microsoft stack

A reporting practice built for scrutiny.

Numbers that reconcile to the ledger. Pipelines from more than 200 source systems. Governance your security team will sign. Delivered by a dedicated practice, not a rotating bench.

Reference architecture
  1. 01

    Sources

    ERP, finance, CRM, HR, operations, files

  2. 02

    Ingestion

    Incremental pipelines, gateways, monitoring and alerting

  3. 03

    Storage

    Lakehouse or warehouse, layered from raw to curated

  4. 04

    The semantic model

    One definition per measure. Security applied once, inherited everywhere.

  5. 05

    Consumption

    Reports, paginated output, Excel, embedded, Copilot

Most reporting failures are decided in layers 01 to 04.

Most reporting budgets are spent on 05.

source systems on our connectivity index
200+
software and data projects delivered
125
years of engineering, as a Roth&Co company
20+

When clients call us

Rarely for a first dashboard.

Our clients usually have reporting in place already, along with a data team and opinions. They come to us when the estate has to carry more weight than it was built for.

  1. 01

    Several ERPs, one set of numbers

    Consolidated reporting across entities and acquisitions that run on different systems and different charts of accounts.

  2. 02

    An estate that has outgrown itself

    Reports that take a minute to render, refreshes that fail overnight, a capacity running at its ceiling.

  3. 03

    A move to Fabric, or off a legacy tool

    Migration from SSRS, Tableau, Qlik or Cognos, with the numbers reconciled report by report before the old system is switched off.

  4. 04

    Self-service without the sprawl

    Hundreds of workspaces and no agreed source. Certified models, a workspace strategy and a center of excellence that outlasts us.

  5. 05

    Analytics inside your own product

    Power BI Embedded for your customers, with tenant isolation and your interface around it.

The practice

Six disciplines, and what you hold at the end of each.

DisciplineWhat we doWhat you hold
Data engineeringWhat we doIncremental pipelines from ERP and operational systems into a lakehouse or warehouse. Data Factory, Dataflows Gen2, on-premises gateways.What you holdMonitored pipelines that alert on failure
Semantic modelingWhat we doOne definition per measure, agreed with the people who own the number, then built as a star schema. Storage chosen table by table, so a report stays fast as the history grows behind it.What you holdA documented model that ties to the general ledger
PerformanceWhat we doIncremental refresh, aggregations, query and DAX tuning, capacity sizing against real workload.What you holdBefore-and-after timings for every report we touch
Security and governanceWhat we doRow-level and object-level security, sensitivity labels, certified models, workspace and tenant design.What you holdAn access matrix your auditors can read
LifecycleWhat we doDevelopment, test and production stages with deployment pipelines, source control and release checks.What you holdA release process your own team can run
Reporting and enablementWhat we doRole-based reports, paginated output, mobile layouts, embedded analytics, then training for authors and consumers.What you holdReports people open, and staff who can extend them

Connectivity index

215 systems. One model.

If it holds your data, Power BI can read it. What differs is the route, and the route decides refresh speed, cost and what breaks at upgrade time. Each system below is marked with the route we would normally take.

Native
A connector Microsoft lists for Power BI. Fastest to stand up.
Database
We read the SQL Server, Oracle or other database underneath the application.
Driver
The vendor's ODBC driver or OData feed, through Power BI's generic connectors.
API
A pipeline we build against the vendor's API, landing the data where Power BI reads it.

  • Native9
  • Database8
  • Driver2
  • Dynamics 365 Business CentralNative
  • Dynamics 365 FinanceNative
  • Dynamics 365 Supply ChainNative
  • Dynamics NAVNative
  • Dynamics GPDatabase
  • Dynamics AXDatabase
  • Dynamics SLDatabase
  • SAP S/4HANANative
  • SAP ECCNative
  • SAP BW and BW/4HANANative
  • SAP HANANative
  • SAP Business OneDatabase
  • SAP Business ByDesignDriver
  • Oracle Fusion Cloud ERPNative
  • Oracle NetSuiteDriver
  • Oracle E-Business SuiteDatabase
  • JD Edwards EnterpriseOneDatabase
  • JD Edwards WorldDatabase
  • PeopleSoftDatabase

The route shown is the usual one. The right one for you depends on version, hosting and licence, and we confirm it in the first technical session. Native status follows Microsoft's published Power Query connector list. Product names belong to their owners.

Not listed? Ask us about it

Governance and security

Designed to pass review, not to avoid it.

Your security, audit and IT teams will ask how this is controlled. These are the answers we build in from the first week, and document.

  • Access

    Who sees which rows

    Row-level and object-level security defined in the model and tied to your directory groups.

  • Lineage

    Where every number came from

    Each measure traced from the report back to source tables, with the business definition beside it.

  • Change

    How a change reaches production

    Separate development, test and production stages, with source control and a recorded release.

  • Classification

    What may leave the building

    Sensitivity labels that follow the data into exports, and certified models marked as the source to trust.

  • Operations

    What happens when a refresh fails

    Monitoring and alerting on every pipeline, so a stale dashboard is caught by us before a meeting catches it.

  • Continuity

    What you own if we leave

    The model, the pipelines, the documentation and the conventions, in your tenant, with your team trained on them.

We sign HIPAA business associate agreements, and your data stays in your own tenant, in the region you choose. Our engineering team works under an ISO 27001 certified information security management system.

Selected work

The fix is usually one layer below where it hurts.

  • Retail · NetSuite · Power BI

    >90%

    Refresh time cut from hours to minutes

    The ERP's API returned a thousand rows per request. We replaced it with saved-search exports feeding the model directly. Data now refreshes every 15 minutes and the servers behind the old process were retired.

    Read the engagement
  • Retail pricing · Data platform

    1B+

    Price points collected across 6,000 sites

    A competitive pricing platform covering 55 countries. Collection, matching and normalization run continuously, so pricing teams act on a competitor's move the same day.

    Read the engagement
  • Specialty chemicals · Deacom · 1,000+ distillation jobs

    97

    Crude materials, each one now automatically costed from the ERP data

    Crude goes through the still several times, and each pass yields several grades plus waste, so finance needed to see what every run produced and which crude it came off. The ERP's job records carry the output lines and nothing that points back to the material that went in. The costing figures were assembled by hand every month. We rebuilt that link for each item and put 1,000+ closed jobs and 3,000+ output lines, waste included, behind one costing report.

“At Roth&Co, a question used to take a day, because someone had to tie the numbers together first. Now the answer is there when the question comes up.”
Zacharia WaxlerCo-Founder, RothTech

Engagement

Four ways to work with the practice.

They differ in one thing: how much is already decided when we start.

  • Fixed scope

    A defined build, at a stated price

    You already know what you want built. You receive the architecture, the sequence, the timeline and the total cost before you commit to the build, then four fixed stages: assessment, modeling, build, rollout. The engagement ends when the scope is delivered.

    Suits: a migration, a rebuild, a rescue.

  • Agreed outcome

    You set the destination, we work out the route

    You know what the reporting has to let you decide, but not what your systems will give up until somebody reads them. We estimate the whole as a range, price the first stage only, and cost the rest once discovery has turned the range into a plan. The engagement ends when the outcome is reached.

    Suits: a first serious platform, or an estate that has grown past its original design.

  • Dedicated team

    Named engineers, working to your backlog

    You direct the work week by week, the same way you direct your own team. Senior engineers join your priorities, your branching rules and your release process, for as long as the roadmap needs them. It ends when you say it does.

    Suits: a standing roadmap, and an internal team to run it with.

  • Managed service

    We keep it running, you use it

    The platform is built, and somebody has to look after it every day. We watch the refreshes, tune what slows down, manage capacity and add reports as the business asks for them. There is no end date and no project to run.

    Suits: lean internal teams.

  • Artificial IntelligenceAI grounded in your data and workflows: automation, prediction, and governed access to the models your team already uses.
  • Cloud DevelopmentCloud infrastructure that scales with the business and costs less to run than the servers it replaces.

Next step

Request a technical briefing.

A working session with a practice lead, not a sales call. Bring your architecture diagram and the report that causes the most arguments.

  1. 01We review your sources and current estate with you.
  2. 02You hear plainly what the data can support today and what needs work first.
  3. 03You receive the architecture, the order of work and a price you can commit to, in writing.

Or speak with Chuck Gartenhaus, our Founder & CEO, directly: 718-975-5500 · hello@rothtech.io

We reply within one business day.