Questions to Ask a Power BI Consultant

The questions to ask a Power BI consultant are not about whether they know the tool. Almost anyone who reaches a proposal can build a chart. What differs between firms is who actually does the work, who ends up owning it, how the price is set and what happens when the project is over — and none of that shows up in a demo.

This article assumes you have already decided to hire help. If you are still weighing whether an outside consultant is worth it at all, start with why companies hire Power BI consultants. If the open question is budget, our Power BI implementation cost guide breaks a first project into its line items, so the rate questions below have something to be measured against.

Each section gives the question, why it matters and what to listen for. Ask them of every firm on your shortlist, including us.

Who will actually build our model?

Ask for the name of the person who will write the semantic model and the measures, and whether that is the person on the call. In many firms the senior consultant sells and scopes the work and a junior team or subcontractor delivers it. That is not automatically bad, but you should know before you sign, and you should meet the builder.

The model is where structural mistakes are made and where they are expensive to undo. A good answer names one person, says how much of their time the project gets, and explains who covers for them. A vague answer about "our team" or "our delivery centre" is worth a follow-up question about who reviews the model before it reaches you.

• Will the person scoping the work also build it?
• Is any part of the work subcontracted, and to whom?
• Who reviews the model before we see it?

Can you show us a model you built, not just a dashboard?

Dashboards are the easy part to show and the least informative. Ask the consultant to walk you through the model behind one: the tables, the relationships, where the measures live and how they are named. Most of a consultant's real work is there, and it is the part that decides whether two reports agree on a number next year.

Listen for a star schema — fact tables of events joined to shared dimension tables — and for measures defined once in the model rather than calculated separately inside each visual. A consultant who can explain why a model is shaped the way it is, in plain language, can usually build one that your own staff can maintain. Client data will rightly be off limits, so a demo model or an anonymised walkthrough is a reasonable substitute.

Certifications come up at this point too. An exam certification shows that someone has studied the product and passed a test, which is useful. It does not show how they design a model under a real deadline with messy data, so treat it as one input and ask to see the work as well.

Who owns the files, the workspace and the data when we are done?

This is the question buyers most often forget to ask. The report files, the workspaces they are published to and the connections to your data should all sit in your Microsoft tenant, under accounts your company controls, from the first day. If a consultant builds in their own tenant and "hands over" at the end, you depend on them for every change until that handover actually happens.

Ask where the files will be stored, whose account owns the published workspace and the scheduled refresh, and whether the credentials used for data connections belong to your organisation. A good answer is unambiguous: everything lives in your tenant, you hold admin rights, and the consultant works as a guest whose access you can remove.

• Report and model files stored in your own storage or source control
• Workspaces and refresh schedules owned by your organisation's accounts
• Data connection credentials that belong to you, not to the consultant
• Consultant access as a removable guest, not as the owner

How is the scope set, and what happens when it changes?

Every project changes once real data is in front of people. Ask how the consultant writes scope down — the specific dashboards, sources and measures in the first delivery — and what the process is when you ask for something outside it. A written change process protects both sides; its absence usually means either an argument or a surprise invoice later.

Ask too whether the price is fixed for the agreed scope or billed by the hour, and what the consultant does when they discover the data is messier than expected. A good answer explains how they look at your sources before quoting, so the number reflects what is really there, and how they tell you early when the estimate is at risk.

What are your rates, and are they published?

Ask for the hourly or daily rate even if the project is quoted as a fixed fee, because it is what any extra work will be billed at. Ask whether there is a minimum commitment, whether unused hours expire, and whether support after launch is a monthly retainer or something you draw down only when you need it.

We publish ours: Power BI Flex Support is $100–$125 per hour in prepaid blocks whose hours do not expire. Whichever firm you choose, a rate you can compare is more useful than a project total you cannot break down — and a consultant who will not state a rate at all is making the comparison harder for a reason.

What will we be able to change ourselves after launch?

A good engagement leaves you more independent, not less. Ask what documentation you will receive — where each source comes from, what each measure means, how the refresh is scheduled — and who on your side will be trained to maintain it. Ask what an internal person will be able to do alone next month: add a column, change a filter, build a new page on the existing model.

Then ask what support looks like after go-live and how quickly a broken refresh gets looked at. The honest answer to "what happens when you are gone" should be specific. If the only plan is to call the consultant, you are buying a dependency rather than an implementation.

How do you handle security and access?

If different people should see different data — managers their own teams, regions their own numbers, clients their own accounts — ask how the consultant implements and tests row-level security. A good answer describes a security table, roles and testing as a specific user before anyone is given access.

Ask also which licences your viewers will need, because that recurring cost often outlasts the project fee. A consultant who raises licensing before you ask is usually one who has been through a rollout before.

Further reading: Microsoft Learn — star schema guidance for Power BI.