Key takeaway: Power BI for finance teams takes over the assembly work: it pulls figures from your bookkeeping system, refreshes them on a schedule and shows P&L, cash and margins on one page. It does not correct your bookkeeping, does not file your annual accounts and does not update in real time. It sits next to Excel rather than replacing it: repeating reports move to Power BI, ad hoc modelling stays in Excel. A first working dashboard takes 2 to 4 weeks.
Most finance teams do not have a reporting problem. They have an assembly problem. The figures exist, but they live in the bookkeeping system, in a bank export, in a webshop and in three Excel files that only one person fully understands. Power BI for finance teams is the tool that does that assembly for you, every morning, without anyone copying a single cell. This article is about what it can actually do with your finance data, what it cannot do (which almost nobody writes down), what you need before you start, and how long it takes.
What is Power BI and what does it do for a finance team?
Power BI is Microsoft’s reporting tool. It connects to your bookkeeping system, refreshes the figures on a schedule and shows them as one interactive report. It sits on top of your ledger, it does not replace it.
Power BI is Microsoft’s reporting tool. It connects to your source systems, keeps a copy of the data in a model, and shows that model as interactive reports. For a finance team the useful part is the connection plus the refresh: once the link to your bookkeeping system is built, the numbers update themselves on a schedule and everyone looks at the same version.
It sits next to your bookkeeping system, it does not replace it. Your ledger stays the ledger. Power BI is the layer on top that answers management questions your accounting package answers slowly or not at all, such as margin per customer over the last 24 months, or cash coming in over the next 6 weeks.
What can Power BI actually do with your finance data?
Six recurring jobs: the monthly reporting pack, budget versus actual, cash and receivables, a group overview across entities and currencies, margin analysis and distribution to different audiences from one source.
In practice a finance dashboard takes over six jobs. These are the ones that pay for themselves fastest, roughly in the order most teams build them.
| Job | What it replaces | What you get |
|---|---|---|
| Monthly reporting pack | Copying the trial balance into an Excel template | P&L, balance sheet and cash flow refreshed automatically, same format every month |
| Budget versus actual | A variance sheet built by hand after closing | Variance per account with a click through to the underlying invoices |
| Cash and working capital | An ageing list exported once a month | Receivables by age and by customer, DSO, overdue amounts, payables and DPO on the same page, updated daily |
| Group overview | Four workbooks glued together | Several entities and currencies in one view, per entity and consolidated |
| Margin analysis | A one-off analysis nobody repeats | Margin per product, project, channel or customer, always current, with a drill down from the consolidated figure to the underlying ledger entries |
| Distribution | Mailing a PDF to management, the bank and the parent company | One source, a different view per audience through row-level security, no version discussion |
The common thread: Power BI is strongest where the same work is repeated every month. A one-off analysis is usually faster in Excel. Anything you rebuild twelve times a year belongs in a dashboard.
Not sure which of the six applies to you?
In half an hour we can tell you which report would save your team the most time.
Book a short call →What can Power BI not do?
It does not clean up your bookkeeping, does not let you type figures back in, does not produce statutory accounts and is not real time: 8 scheduled refreshes a day on Pro, up to 48 on Premium Per User or Fabric.
This is the section vendors leave out. Knowing the limits up front is what keeps a dashboard project from disappointing, so here is the honest version.
| What people expect | What actually happens |
|---|---|
| It cleans up my bookkeeping | It makes the mess visible faster. Costs booked to the wrong account stay on the wrong account, only now on a screen everyone sees |
| I can type or correct figures in the dashboard | Power BI reads, it does not write. Entering data needs an extra component such as Power Apps or a paid write-back visual |
| It produces my annual accounts | Statutory accounts and the XBRL filing stay outside Power BI. This is management reporting, not statutory reporting |
| It is real time | Refreshes are scheduled. On a Power BI Pro licence you get 8 scheduled refreshes per day, on Premium Per User or Fabric capacity up to 48 (Microsoft documentation) |
| It does the consolidation for me | It applies the elimination rules you define. Someone from finance still has to decide how intercompany, currency translation and minority interests are handled |
| It explains the variance | It shows the variance and where it sits. The reason a margin dropped is still a human sentence under the chart |
Two of these deserve a number. A refresh on shared capacity has to finish within two hours, five hours on Premium capacity, which is why a model built straight on a slow API eventually needs a data warehouse. And the refresh limit is per day, not per hour, so “this morning’s figures” is realistic, “this minute’s figures” is not.
Power BI or Excel: which one does what?
Neither replaces the other. Power BI handles the repeating, shared reporting on large volumes with rights per user. Excel keeps ad hoc modelling, scenario work and manual reconciliations. Almost every finance team we work with runs both, on purpose.
The question comes up in every first meeting, and the honest answer is that they do different jobs. The split below is the one that holds up in practice.
| What you are doing | Power BI | Excel |
|---|---|---|
| Building the same report every month | Refreshes itself from the source system, same format every time | An export and a copy-paste round, every month again |
| Sharing outside the finance team | One version, rights per person, no files by email | Copies drift apart the moment they are sent |
| Volume | Millions of ledger lines without complaint | Slows down and eventually refuses |
| Ad hoc modelling and scenarios | Awkward, a new question often means changing the model | Faster, and this stays Excel’s home ground |
| Manual reconciliations and corrections | Reads only, you cannot type in it | Type wherever you like, which is exactly the point |
| Audit trail | One definition, applied the same way every month | A formula anyone can overwrite without it showing |
There is also a middle road most teams miss: you can connect Excel to the Power BI model and pull a pivot table straight out of it. The controller who prefers Excel keeps working in Excel, but on the same governed figures as the dashboard, so there is still only one version of revenue. That combination usually settles the discussion faster than trying to move everyone off spreadsheets.
“The dashboards that fail are the ones that start with a wish list of charts. The ones that work start with three questions the management team actually asks every month. We once replaced a 14-tab Excel pack with four pages, and the finance manager got two days a month back. Nothing in that project was technically difficult. The hard part was agreeing on what revenue means.”
What does this look like in practice?
Three anonymised examples: a monthly pack for a German parent that moved from working day 8 to working day 3, a product group with a negative margin once advertising was allocated, and four BVs consolidated onto one page.
Three examples from our own work, anonymised, with the numbers that mattered.
- A Dutch subsidiary of a German group. The monthly pack for the parent took two full days in Excel. The same pack now refreshes every morning from the bookkeeping system, in the parent’s format and currency. The closing report goes out on working day 3 instead of working day 8.
- An e-commerce company on Exact Online. Margin was known per product but not per channel, because shipping, returns and advertising sat in different systems. Putting them in one model showed one product group with a negative margin once advertising was allocated. That group was dropped within the quarter.
- A services group with four BVs. The board looked at four separate workbooks and a fifth that summed them. One model with intercompany eliminated gives a group P&L on one page, with a click through to each entity. Month-end went from a reconciliation exercise to a review.
None of these needed new software beyond Power BI itself. All three needed someone to write down the definitions first.
What do you need before you can start?
A source system with an API, a consistently used chart of accounts, written KPI definitions, a Power BI Pro licence per user and an owner in the finance team with 2 to 4 hours a week during the build.
A dashboard project stalls on preparation far more often than on technology. Five things have to be in place.
- A source system with an API or a scheduled export. Exact Online, AFAS, Twinfield, Business Central, NetSuite and SnelStart all qualify. If your figures only exist in a manually maintained Excel file, that file becomes the weakest link in the chain.
- A chart of accounts that is used consistently. Not perfect, but consistent. Mapping to a reporting structure is normal work; guessing what a general account contains is not.
- Written definitions. Revenue, gross margin, EBITDA, an active customer. Half a dashboard project is a definition discussion, and it is cheaper to have it before the model is built.
- Licences. One Power BI Pro licence for every person who views or builds, or a Fabric capacity if you want viewers without individual licences.
- An owner inside finance. Reserve 2 to 4 hours a week during the build for questions, checks and approval. Without that person the project slips, whoever builds it.
One more thing, easy to overlook: your closing calendar sets the ceiling. A dashboard cannot show reliable figures on working day 3 if the books are only closed on working day 12. If that is your situation, fix the closing process first and the dashboard second.
What does a good finance data model look like?
Ask for five things: a star schema around the general ledger, a separate layout table for the P&L order and its subtotals, time intelligence defined once in DAX, row-level security per entity or cost centre, and incremental refresh once the history gets heavy.
You do not have to build this yourself, but you should know what to ask for, because “it works now” and “it still works in two years” are different questions. These five choices are what separate the two.
| What to ask for | Why it matters for finance |
|---|---|
| A star schema | One fact table with the general ledger transactions, separate dimension tables for period, entity, cost centre and the chart of accounts. Without it every new question becomes a rebuild instead of a filter |
| A separate layout table for the P&L | Your ledger does not return accounts in statement order, and gross margin and EBITDA are not simple sums of the rows above them. A disconnected layout table holds the order, the subtotals and the sign conventions, so the P&L reads like a P&L |
| Time intelligence written once | Year to date, same month last year, rolling twelve months: defined once in DAX and reused everywhere, not rebuilt per chart. This is what keeps two pages from disagreeing |
| Row-level security | A branch manager or entity director opens the same report and sees only their own figures. One report instead of five filtered copies that quietly fall out of step |
| Incremental refresh, when it is warranted | Only the recent period is reloaded and closed years stay put. Microsoft position this for tables that grow large and change often (Microsoft documentation). Most SME models never get there, so treat it as the fix when refresh time starts creeping toward the window, not as a box to tick on day one |
If your builder cannot explain the first two in plain language, that is a warning sign. A dashboard that looks right in month one and cannot answer a new question in month six is almost always a modelling problem, not a Power BI problem.
How long does it take to build a finance dashboard?
A first working dashboard takes 2 to 4 weeks, a complete reporting set 6 to 10 weeks. After go live, budget 2 to 4 hours a month for maintenance.
For a single entity on one bookkeeping system, the pattern is fairly predictable.
| Phase | Time | What happens |
|---|---|---|
| Definitions and access | Week 1 | Questions, KPI definitions, connection to the source system |
| Model and first version | Week 2 to 3 | Data model, mapping, the first working report |
| Review and correction | Week 4 | Reconcile against the trial balance, fix, hand over |
| Full reporting set | Week 6 to 10 | Budget, cash, margin and a group view |
So a first working dashboard in 2 to 4 weeks, a complete reporting set in 6 to 10 weeks. After go live, budget 2 to 4 hours a month for maintenance: a new account, a changed budget, an extra entity.
Do you need a data warehouse or is a dashboard enough?
Start with the dashboard. Add a data warehouse when three or more systems have to be combined, when you need history the API no longer returns, or when refresh time becomes unreliable.
Start with the dashboard. A separate data warehouse is a layer between your systems and Power BI, and it costs time and money you do not always need to spend.
A dashboard straight on the source is enough when you have one or two source systems, need a few years of history, and your refresh finishes comfortably inside the window. Move to a data warehouse when three or more systems have to be combined, when you need history the API no longer returns, when API rate limits make the refresh slow or unreliable, or when several entities each bring their own administration. The trigger is almost always refresh time or source count, not company size.
What does Power BI cost for a finance team?
Licences are €12,10 per user per month for Pro and €20,80 for Premium Per User, excluding VAT. A build runs from about €500 to €1.500 for a simple dashboard up to €3.500 to €7.500 and more for a multi-entity model.
Short version, because this article is about what is possible rather than what it costs. Licences in the Netherlands are €12,10 per user per month for Power BI Pro and €20,80 for Premium Per User, both excluding VAT and billed annually, according to Microsoft’s own price list. On top of that comes the build. In the Dutch market a simple dashboard runs from about €500 to €1.500, a proper finance reporting set from €1.500 to €3.500, and a multi-entity model with a warehouse behind it from €3.500 to €7.500 and up.
Those ranges say nothing about your situation until someone has looked at your systems. The number that usually decides it is not the build cost, it is the days a month your team gets back.
How Oakhill helps you with Power BI
Oakhill is a finance firm that builds dashboards, not a software supplier. We define the reporting logic first and can keep running the bookkeeping underneath, so the source data stays clean.
We are a finance firm that builds dashboards, not a software supplier that happens to know accounting. That matters at the point where the reporting logic is defined, because that is where most dashboards go wrong. Oakhill delivers financial reporting and dashboarding for companies in the Netherlands, from a first monthly pack to a full group model, and we can keep running your bookkeeping underneath it so the source stays clean.
Frequently asked questions about Power BI for finance teams
What is Power BI used for in finance?
Mainly for recurring management reporting: the monthly pack, budget versus actual, cash and receivables, and margin analysis. It connects to the bookkeeping system, refreshes on a schedule and shows the result as one interactive report instead of several Excel files.
Can Power BI replace Excel?
For repeating reports, yes. For one-off analysis, ad hoc modelling and scenario work, Excel stays faster and most teams keep both. The practical split: anything you rebuild every month belongs in Power BI, anything you build once belongs in Excel.
Can Power BI connect to Exact Online, AFAS or Twinfield?
Yes. All three have APIs, and there are standard connectors as well as third party connectors for them. The connection is rarely the hard part; mapping the chart of accounts to a reporting structure takes more time than building the link.
How often does a Power BI dashboard update?
On a schedule you set. With a Power BI Pro licence you can schedule 8 refreshes per day. On Premium Per User or a Fabric capacity that goes up to 48 per day. Most finance teams settle on one refresh early in the morning, sometimes a second one after lunch.
Do I need Power BI Premium?
Usually not at the start. Pro is enough for a single entity and a normal reporting set. Premium Per User or a Fabric capacity becomes relevant when your model gets large, when you need more frequent refreshes, or when you want many viewers without giving each of them a licence.
Can Power BI consolidate several entities?
It can present a consolidated view and apply elimination rules, including currency translation. It does not decide those rules for you. Someone from finance defines how intercompany transactions, exchange differences and minority interests are treated, and the model executes that consistently every month.
Can you enter or change figures in Power BI?
Not by default. Power BI reads data and displays it. Adding budgets, forecasts or comments needs an extra component, such as a Power Apps form or a paid write-back visual, or you keep those inputs in a controlled Excel file that the model reads.
Is Power BI worth it for a small finance team?
Often yes, and sometimes more than for a large one, because a team of one or two people loses a bigger share of its month to assembling reports. The deciding factor is repetition, not headcount: if the same pack is built every month from the same sources, the payback is usually a few months.
How long does it take to build the first dashboard?
Two to four weeks for a first working version on one entity and one bookkeeping system, six to ten weeks for a complete reporting set including budget, cash and a group view. The main variable is how quickly definitions are agreed, not how quickly the model is built.
Can different people see different figures in the same report?
Yes, that is row-level security. You set a rule per user or per group, for example an entity, a branch or a cost centre, and everyone opens the same report but only sees the rows they are allowed to see. It saves you maintaining five filtered copies of one dashboard, and it is the reason a group report can safely be opened by entity managers.
What is a star schema and does my model need one?
A star schema means the ledger transactions sit in one central fact table, with separate dimension tables around it for period, entity, cost centre and the chart of accounts. It matters because a flat export of everything turns every new question into a rebuild, while a star schema turns it into a filter. For anything beyond a single simple report, yes, insist on one.
Why does my P&L come out in the wrong order in Power BI?
Because a bookkeeping system returns accounts in account order, not in statement order, and subtotals like gross margin and EBITDA are not simple sums of the rows above them. The fix is a separate layout table that holds the statement order, the subtotals and the sign conventions. It is standard work for anyone who has built a finance model before, and a surprise to anyone who has not.
Who maintains the dashboard after it is built?
Someone has to, and it is a real if small job: two to four hours a month for new accounts, a changed budget or an extra entity. Either you appoint an owner in the finance team and get them trained, or you leave it with the party that built it. What does not work is assuming a dashboard is finished on the day it goes live.
Want to see your own figures in a dashboard first?
Send us a trial balance and one month of data and we will show you what a first version looks like, before you commit to anything. Half an hour is enough to tell you whether Power BI is worth it in your situation.
Book a no-obligation call →Related articles
This article is general information and not financial or tax advice. Licence prices are Microsoft list prices for the Netherlands as at July 2026, excluding VAT. For advice on your own situation, contact Oakhill Financial Services.