“Nobody trusts the numbers in the report.”
Two reports give three answers, so the leadership team goes back to spreadsheets. Almost always a modelling problem, not a Power BI problem.
Semantic model rebuild
The Microsoft stack, end to end
Power BI
Reports, dashboards, semantic models
Microsoft Fabric
Lakehouse, pipelines, OneLake
Azure SQL, Synapse, Storage
SharePoint, Excel, Teams
CRM and ERP data
Microsoft, Power BI, Microsoft Fabric, Azure and Dynamics 365 are trademarks of the Microsoft group of companies.
Sound familiar?
Four ways Power BI goes wrong
None of them are really about Power BI. They are about what sits underneath it.
“The refresh takes four hours and keeps failing.”
Usually a model doing work at refresh time that should have been done upstream, or query folding quietly broken in Power Query.
Performance rescue
“We bought Fabric and don't know what to do with it.”
Capacity is running, the bill is arriving, and nothing is landing in OneLake. You need the first pipeline built properly, as a pattern to copy.
Fabric setup
“One person built it all and they've left.”
An undocumented model nobody can safely change. We reverse-engineer it, document the lineage and hand it back maintainable.
Handover & documentation
What good looks like
One page. Every question answered.
Headline numbers first, the trend behind them, then the detail — and a panel showing whether the data can be trusted today.
Decision-first layout
The four numbers a GM is accountable for, above the fold, every time.
Consistent time intelligence
Year-to-date means the same thing on every page, because it is defined once.
Health in the open
Last refresh, row counts, failures. If the data is stale, the report says so.
Under the hood
The part nobody shows you
Dashboards get the applause. This is what decides whether yours is still fast and correct in two years.
Revenue Rolling 12M.dax DAX
// Rolling 12-month revenue that respects whatever
// filters the user has applied on the page.
Revenue Rolling 12M =
VAR LastVisibleDate =
MAX ( 'Date'[Date] )
VAR Window =
DATESINPERIOD ( 'Date'[Date], LastVisibleDate, -12, MONTH )
VAR Result =
CALCULATE ( [Revenue], Window )
RETURN
IF ( NOT ISBLANK ( [Revenue] ), Result ) FactOrder.pq Power Query M
// Incremental load that folds back to the source.
// RangeStart/RangeEnd let Fabric partition the table
// so a refresh touches days, not years.
let
Source = Sql.Database ( ServerName, "Sales" ),
Orders = Source{[Schema="dbo", Item="FactOrder"]}[Data],
Filtered = Table.SelectRows (
Orders,
each [ModifiedUtc] >= RangeStart and [ModifiedUtc] < RangeEnd
),
Typed = Table.TransformColumnTypes (
Filtered,
{{"OrderDate", type date}, {"NetAmount", Currency.Type}}
)
in
TypedMicrosoft Fabric
Built once, properly, as a pattern
Fabric rewards a clean structure and punishes an improvised one — in both correctness and capacity cost. This is the shape we build to.
Services
How we can help
Report & dashboard build
From a blank canvas to something an executive actually opens on Monday morning. Built on a proper star schema so it stays fast as the data grows.
- Requirements from the questions you need answered
- Star schema + DAX measure layer
- Mobile and print layouts
- Row-level security where needed
Report rescue & tidy-up
Inherited something slow, wrong, or held together with hope. We find why, fix the model underneath, and leave the numbers reconciled.
- Performance profiling with DAX Studio
- Model and measure refactor
- Reconciliation back to source
- Before / after benchmark
Migrating old reports
SSRS, Tableau, Qlik, Cognos or a decade of Excel. We do not re-draw them one for one — we work out which ones anyone still reads, then rebuild those properly.
- Usage audit before we migrate anything
- Rationalise 200 reports down to the 30 that matter
- Rebuild on one shared semantic model
- Parallel run until the numbers agree
Microsoft Fabric setup
Lakehouse, pipelines, medallion structure and governance — built once, properly, as a pattern your team can extend without us.
- OneLake + medallion architecture
- Data Factory pipelines
- Workspace and capacity design
- Cost monitoring from day one
Semantic model design
The layer that decides whether self-service works or quietly produces wrong answers. Named the way your business actually speaks.
- Star schema design
- Calculation groups
- Time intelligence done once, correctly
- Field descriptions and synonyms
Health check & audit
A fixed-price week that tells you what you have, what it costs, what is broken and what to do about it. Yours to act on with or without us.
- Tenant and workspace inventory
- Model and refresh diagnostics
- Licence and capacity review
- Prioritised remediation plan
Training & enablement
Your team builds the next twenty reports, not us. Pitched at whoever is actually going to maintain this after we leave.
- Author and analyst tracks
- Your data, not demo data
- Written standards and templates
- Review sessions for early work
How we engage
Small, fixed, then bigger
You should never be asked to sign an open-ended hourly rate against a system nobody has looked at yet.
- 01
Conversation
Free. Half an hour on what you are trying to see and why you cannot see it. Often ends with us telling you it is smaller than you feared.
- 02
Health check
Fixed price, fixed scope, one week. You get a written assessment and a prioritised plan — yours whether or not you continue with us.
- 03
Build
Fixed price, scoped from what the health check found. Delivered in fortnightly increments you can see and react to, not one big reveal.
- 04
Handover or care
Documentation, standards and training so your team owns it. Or a monthly retainer if you would rather we kept the lights on.
We already move the data
Most BI consultancies start at the semantic model and hope the data arrives clean. Migration and lineage is our day job — we can fix the problem at source instead of writing DAX to work around it.
The model is the deliverable
Pretty dashboards are the easy half. We are judged on whether the numbers reconcile, the refresh stays under control and the next person can maintain it.
You can see the work
Every engagement produces artefacts you keep: a documented model, a lineage map, a benchmark, a runbook. No black boxes.
Not sure whether you need a rebuild or a tidy-up? Start with a conversation — it is free, and we will tell you honestly if the job is smaller than you think.
Book a Power BI consultationGet moving
Talk Power BI with us
Tell us what you are trying to see, what you have now, and roughly when you need it. We will come back within one business day.
- [email protected]
- Coverage
- Nationwide across New Zealand, remote-first delivery
- Response time
- We reply to every enquiry within one business day