A Power BI health check is a scored review of your reporting estate against a written standard. Mine is £950 fixed and runs 48 checks across six areas: data model and measures, logic and traceability, design and accessibility, performance and refresh, security and access, and adoption and usage. Every check comes back as Healthy, Watch or Critical. You get a scored report card within five working days, the issues ranked by impact, a roadmap you can hand to any developer, and a 60-minute walkthrough call.
That's the short answer. The rest of this post unpacks it, because "health check" is a phrase doing a lot of unexamined work in this industry, and if you're comparing quotes you deserve to know what should be behind it.
Why "health check" needs pinning down
Plenty of consultancies will sell you a health check that amounts to a day of someone poking around your workspace followed by a document of observations. That has some value, but it has a problem: it's one person's taste. If the reviewer thinks your model is fine, it's fine, and there's nothing for you to push back against.
A scored review against a written standard is a different product. Every check has a stated standard, every result is one of three states, and any single line can be challenged. If I mark your refresh setup Critical, the report says what the standard is and what I found. You can disagree with the standard. What you can't get is a vague thumbs up that turns out to mean nothing six months later.
So when you're comparing health checks, the first question isn't the price. It's "can I see the list of what you check, and against what standard?"
What does a Power BI health check cover?
Six areas, eight scored checks in each. Here's what each area is asking, and what a Critical finding in it looks like in practice. All of this is published on the service page.
| Area | The question it answers | What Critical looks like |
|---|---|---|
| Data model and measures | Is the model sound, and do measures give the same answer everywhere they appear? | The same measure returns different numbers on different pages, and nobody can say which is right |
| Logic and traceability | Can a number be traced from the visual back to source? | A board pack figure that nobody can explain when an auditor asks where it came from |
| Design and accessibility | Do the reports surface what changed and what's at risk, or bury it? | A wall of numbers nobody reads, with meaning carried by colours some readers can't tell apart |
| Performance and refresh | Do the visuals load, and does the refresh hold? | A refresh that's been failing quietly, so decision makers are reading stale numbers |
| Security and access | Who can actually see what? | Row-level security assumed to work but never tested, or a workspace half the company can edit |
| Adoption and usage | Is anyone actually opening the reports? | A dashboard that took weeks to build and gets three views a month |
That last area surprises people. Usage gets scored because a technically perfect report nobody opens is still a failure, and the usage data usually tells you why long before anyone says it out loud. If two of your reports disagree with each other, that has its own consequences beyond any score.
How is a Power BI health check scored?
Each check lands in one of three states: Healthy (2 points, meets the standard), Watch (1 point, works today but it's a known weak point) or Critical (0 points, needs action). Three states rather than a mark out of five, deliberately. Three states are quicker to score, less subjective to argue about, and far easier to defend. The detail comes from having 48 checks, not from a wider scale.
Those points add up to an area score, then one estate score shown as a percentage. The percentage sits in a band: 80% and above is Strong, 60% to 79% is Solid, 40% to 59% is Fragile, and under 40% means fix the foundations before trusting what sits on them.
Two rules stop the scoring from lying to you. First, every Critical finding is named on the front page whatever the totals say. An untested security setup doesn't get to average itself into a comfortable 72%. Second, no state is carried by colour alone: every indicator has its label and an icon shape, so nothing depends on telling red from green. I'm red-blind myself, so this one isn't a nice-to-have.
What do you actually receive at the end?
Four things, inside five working days:
- A scored report card. Six areas, 48 checks, each with its state and the evidence behind it.
- The issues ranked by impact. Not everything that's imperfect, the things that actually matter, in order.
- A practical roadmap. Written so you can hand it to a developer, mine or anyone else's, and they'd know what to do. It's deliberately not written so that only I can act on it.
- A 60-minute walkthrough call. We go through the card together, and you can challenge any line.
The scope is published too: one workspace, up to 15 reports, up to 3 semantic models. If your estate is bigger than that, we scope it on a call rather than pretending one size fits.
How much does a Power BI health check cost?
Mine is £950, fixed, no day rates. It's fixed because the scope is fixed: 48 checks on a bounded estate is a known quantity of work, and a fixed price means you know the cost before you know the findings. The market alternative is usually a day rate, where the review costs whatever it ends up costing.
There's a second reason the price is fixed and published: it makes the credit rules possible. If the health check turns into a bigger job within 90 days, the £950 comes off the Full Estate Health Check in full, and half of it comes off any build work booked in the same window. So the worst case is you paid £950 for a scored map of your estate, and the best case is the £950 was effectively free.
What if the problem is underneath Power BI?
A slow or untrustworthy report has two possible causes, and only one of them lives in Power BI. The report that crawls might be a badly written measure, or it might be a missing index on a table your refresh hammers every morning. Look at the reporting layer alone and the honest answer is "it could be either", which means paying to find out twice.
That's why the bigger tiers exist. The Full Estate Health Check is £2,450 fixed and extends the same scored review down to the data itself: database resilience, performance and contention, data quality, governance, lineage and source sprawl, six more areas delivered by a SQL Server specialist while I score the reporting half. Ninety six checks, one combined score out of 192, one roadmap, five working days. There's also a £4,950 tier where the three worst findings get fixed and the estate re-scored, so the improvement is evidenced rather than claimed.
Most estates start at £950. The upgrade path exists for when the findings point downwards.
Can you check things yourself first?
Yes, and you should. The free Report Trust Checklist runs twelve of the same checks as a two-page DIY score you can do this afternoon, no email gate tricks, no consultant required. It's here.
You'll get a long way on your own with the visible stuff: refresh reliability, whether reports agree with each other, whether anyone opens them. What's genuinely hard to self-score is your own model, for the same reason it's hard to proofread your own writing. You built it, so it makes sense to you. Whether it makes sense is a different question, and that's the part an outside review earns its fee on.
Common questions
How much does a Power BI health check cost in the UK?
Mine is £950, fixed, covering 48 checks across six areas with a scored report card, a ranked issues list, a roadmap and a walkthrough call. Much of the market prices reviews on day rates instead, so the total depends on how long the reviewer takes. If you're quoted a day rate, ask how many days and what the written deliverable is before comparing.
What is checked in a Power BI health check?
Six areas: data model and measures, logic and traceability, design and accessibility, performance and refresh, security and access, and adoption and usage. Eight checks in each area, every check scored Healthy, Watch or Critical against a written standard. The area most often missed by informal reviews is adoption, and it's frequently where the real problem lives.
How long does a Power BI health check take?
The scored report card is delivered within five working days of getting access, followed by a 60-minute walkthrough call. Your team's time commitment is small: access setup at the start and the call at the end.
What do I get at the end of a health check?
A scored report card across all 48 checks, the issues that matter ranked by impact, a practical roadmap written so any developer can act on it, and a walkthrough call. You keep everything, and there's no obligation to book any of the fixes with me.
Do I need the £950 check or the Full Estate version?
If your question is "can I trust my reports?", start at £950. If your question is "can I trust my reports and the databases they sit on?", the £2,450 Full Estate Health Check scores both halves in the same week with a specialist on each. If you start small and the findings point below the reporting layer, the £950 comes off the upgrade in full within 90 days.
Can I do a Power BI health check myself?
Partly. The free Report Trust Checklist covers twelve of the checks as a self-score, and it will tell you honestly whether you have a problem. What it can't do well is score your own model design, because the person who built a thing is the worst-placed person to judge whether it's understandable. That's the piece worth paying an outside reviewer for.
What happens after the health check?
You get the roadmap and you choose what to do with it. Some people fix things internally, some book the fixes with me, some do nothing for six months and re-score later. Half the £950 comes off any build work booked within 90 days, but the roadmap is written to be usable by whoever you like.
Every price on this page is published on the Health Check service page and excludes VAT where applicable. If you're not sure which tier fits, the free 20-minute intro call exists to answer exactly that, and "neither yet" is an answer I give regularly.
Prices and scope as published on this site, last checked 1 September 2026.