Power BI in New Zealand

The dashboard is the last fifth of the job

Power BI Desktop is free. Anyone can connect it to a spreadsheet and produce something that looks like a management report before lunch. That is why the quotes surprise people.

What actually takes the time:

1. Getting the data out. Some systems have a clean API. Some have a database you can query. Some have a CSV export a person runs on the first of the month, which is not a data source, it is a habit.
2. Making it agree with itself. Your CRM, your accounting system and your operations system each have a customer list, and they do not match. Somebody has to decide how they are joined.
3. Agreeing what the numbers mean. This is a business problem wearing a technical costume, and it is where projects stall.
4. Building the model. Dates, relationships, measures — the part that decides whether the reports stay fast and correct.
5. The reports. The visible fifth.

Judge quotes on how much attention they give to the first three. A consultant who spends the whole conversation on visuals has not yet met your data.

Write down what the numbers mean

Ask three people in your business what "revenue" means and you will get three answers. Invoiced or paid? Including GST? Including freight? What happens to a credit note issued in a later month? Is a refund negative revenue now or a correction then?

Same again for "active customer", "margin", "on-time delivery" and "open job".

These have to be decided by your people, not your consultant, and they have to be written down somewhere the next person can read them. A project without this produces dashboards that finance and sales disagree about, and the moment that happens both teams quietly go back to their own spreadsheets and you have bought nothing.

Do this before the project rather than during it. It is the cheapest week of work available to you.

Warehouse, or connect direct?

Power BI connects directly to plenty of things — SQL databases, Xero, many SaaS products, files in SharePoint. For one or two sources that is genuinely enough, and a consultant who insists on building a warehouse anyway is overbuilding.

A warehouse starts to earn its cost when:

• You are combining four or more systems, and the joining logic is getting complicated.
• You need history that source systems overwrite — what the pipeline looked like last March, not just now.
• Refreshes are becoming slow, or hitting API limits on source systems.
• Something other than Power BI needs the same numbers.
• Reporting queries are slowing down the systems people are trying to work in.

Ask each consultant which side of that line you are on, and why. Both answers can be honest; the reasoning is what you are buying.

Licensing is a permanent cost and it is not theirs

This catches New Zealand businesses out regularly, because it is easy to build before anyone works out what it costs to share.

• Desktop is free, for building.
• Pro is around NZD 20 per person per month, and everyone who views a report needs one. Fine for a team of ten, painful at two hundred.
• Premium Per User costs more per head and adds larger models and more frequent refreshes.
• Capacity — Fabric or Premium — is a fixed monthly cost that lets people view without individual licences. Above roughly a hundred viewers it usually becomes the cheaper answer.
• Embedded, for showing reports to your customers inside your own product, is a different model again and needs pricing deliberately.

Ask for licensing to be stated separately from the consultant's fee, with the number of viewers you expect in year two rather than year one.

Decide who owns it before it is built

The failure nobody plans for is not a broken dashboard. It is a dashboard that is quietly wrong.

A source system adds a field, someone renames a category, an integration starts sending nulls — and the report keeps rendering, with numbers that are no longer true. Nobody notices for six weeks, and then nobody trusts anything again.

Guard against it with:

• A named owner inside your business, not just at the agency.
• Refresh failure alerts that reach a person.
• A few sanity checks in the model — row counts, totals that must reconcile to the accounting system.
• Documentation of where every number comes from, so a wrong figure can be traced rather than argued about.

Start with the Monday spreadsheet

If this is your first Power BI project, do not start with a company-wide platform. Find the report somebody rebuilds by hand every week, automate that one, and put it in front of the people who use it.

It pays back immediately, it teaches you what your data is actually like, and it earns the credibility to fund the larger piece. Consultants who suggest this rather than a six-month programme are usually the ones who have watched a six-month programme fail.

Comparing five quotes

Read for:

• Who asked which systems the data lives in, and how they would get it out?
• Who raised the question of agreed definitions before you did?
• Who priced Microsoft licensing separately and for the right number of people?
• Who explained whether you need a warehouse — either way — rather than assuming?
• Who described handover, documentation and training your own team?

Describe the report you currently rebuild by hand, which systems it comes from, and how many people need to see the result. Consultants who do this work regularly will tell you quickly whether it is a $15,000 job or the start of a $90,000 one.