Acceptance criteria before a single measure.

1. 01

   Clear root cause and next action

   Not a gallery of charts. A number you can defend, and what to do with it on Tuesday.

2. 02

   Maintainable models and logic

   Transformations and measures that another analyst can open without a guided tour.

3. 03

   Decision-ready Power BI

   Dashboards used in commercial decisions — desktop and mobile — with guided navigation.

4. 04

   Reconciled results, concise handover

   Outputs checked against source. Unresolved issues named. Documentation that travels.

1. 01

   BI troubleshooting and audit

   Reproduce the wrong number. Isolate. Repair the smallest useful cut.

   A productized audit: model, queries, measures, grain. Source vs model vs visual, named. Focused fix, change summary, unresolved issues written down — not a rebuild for its own sake.

2. 02

   Power BI dashboards and models

   Decision-ready reports on a model that can be maintained.

   Requirements, KPI definition, dimensional model, Power Query, measures with business names. Desktop and mobile. Validation against source. Training so the file stays.

3. 03

   SQL / ETL, analytics backend

   Transformations you can follow. Pipelines you can reconcile.

   SQL Server, SSIS, Power Query. File-based ingest from SAP, BW and Navision: content-detected, quality gates, weekly and annual grains that do not overwrite each other. Tests at the edge. Then a model a dashboard can trust.

4. 04

   KPI and BI solution blueprint

   Confirm the outcome before anyone writes a measure.

   Business outcome, inputs, constraints, grain, acceptance. The smallest useful model or report, assumptions on the page. A document the build can be checked against.

5. 05

   Credit-risk controls and recon

   Validate the book. Name the exception. Tie the grain.

   Independent scoring and dataset validation. Outliers and boundary cases in Python and SAS. AML classification at volume against a written rule. Accounting recon before the report. Exception queues with a criterion.

6. 06

   AI output evaluation

   A rubric before the run. Line by line. A person signs.

   Task specs, must and must-not, acceptance. Line-by-line review of generated analysis and code. Unsupported claims labelled unverified. Confirmed defects turned into regression tests. Codex drafts. A person signs the gate.

7. 07

   Pricing and profitability analytics

   Segment. Price. Prove the margin two ways.

   Retail-banking CVM: client data, taxonomies, classification rules, Python and SAS on large sets. Pricing and profitability models with methodology written for the business. Two-path recon of the margin. If it will not tie, it does not ship.

8. 08

   Cross-source data quality

   Consistency rules. Aberrant records named. Then reporting.

   Investment-risk and insurance cuts: consistency between sources, incomplete and aberrant records detected before they enter the book, parameterised investigation applications. A quality gate, not a cleanup after the dashboard exists.

9. 09

   Enterprise reporting estate

   A chapter 100+ people open. Every working day.

   MicroStrategy dossiers and Power BI Service on a model that can be kept. Recurring commercial KPIs, guided navigation, desktop and mobile. Training so the floor does not export to Excel. The client can run without me.

10. 10

    Reverse engineering and method package

    Someone left. The file stayed. Recover the logic.

    Map the model, the queries, the grain, the undocumented measure. Architecture decision record. Requirements recovered from the file, not from memory. Tests, change summary, training. Documents that survive being challenged.

Prefer to write it yourself? Open the desk.