How to Identify a Good Power BI Consultant Before You Hire One

Why I Wrote This - Three Conversations That Stuck With Me

This blog was originally going to be a short LinkedIn post, but when I started I just could not stop. I had more to say.

So here it is... over the last few months, I had several conversations that all pointed in the same direction which stuck with me. Before we answer the title of this blog though, below is a snippet of three of those conversations:

Conversation One: “Power BI Reports Are Just Easy to Build”

One was with someone leading an internal, decent sized BI and Analytics team who said Power BI reports are just easy to build. The emphasis in that conversation was all on the data platform. Pipelines, notebooks, engineering, orchestration, so everything before Power BI and sprinkled with AI. But hold on… a six-figure data platform is not the end goal, surely?! It is there to produce INSIGHTS. To tell us what's working and not working, to help us achieve BUSINESS objectives or overcome BUSINESS challenges. If the reporting layer is simply dismissed as the easy bit, the drag/drop onto a canvas part, you risk forgetting the part that actually determines whether the investment was worth it in the first place. It is not the pipelines under the hood that justify the spend. It is the decisions made at the end. Don't believe me, take this to your CEO and ask them.

Conversation Two: AI Does Not Remove the Need for Fundamentals

Another was with someone learning Power BI, Microsoft Fabric and other tools, who debated that AI now removes the need to learn what I would call the core fundamentals. To name a few, things like dimensional modelling, evaluation context, the principles behind good DAX and the core elements of data warehousing. I mean, from a Power BI perspective, there was no plan to learn or ever pick up The Data Warehouse Toolkit by Ralph Kimball, Agile Data Warehouse Design, The Definitive Guide to DAX or Fundamentals of Data Engineering. And look, I am not saying AI is all bad. In fact, I have written other blogs telling Power BI experts they need to adapt and integrate into the business with the way AI is advancing. But here is the thing... Relying on AI too heavily might help someone get through a few discovery conversations or put together a proposal deck. Then they get into a real project, with real people, real pressure and real messiness and suddenly you realise ChatGPT is just agreeing with them on everything. If they do not understand the fundamentals themselves, they will not know when the answer is weak, when the model is fragile or when the logic is quietly going off course. Plus, if you are the buyer, that becomes your problem too.

Conversation Three: When AI Becomes the Pitch

The third came from a potential prospect. They told me a competitor had said they would use AI to derive most of the work for them. Honestly, that stopped me for a second. In over 10 years of discovery calls, proposals and competitive bids, I had never heard of another consultancy position that so directly as part of the sell. And that, to me, says a lot about where the market is heading. Think about that for a second. AI can absolutely help accelerate parts of delivery. Used properly, it is brilliant. But when that becomes the pitch, it should raise questions (at least for now), especially if you are the buyer looking for genuine Power BI expertise. You should not be impressed that someone can generate things quickly. You should want to know whether they understand what good actually looks like, what assumptions sit underneath the output, how they think about value for the audience the solution is being built for and how they will stop speed turning into rework. You need someone who really feels comfortable talking to the business, getting involved, asking the right questions, facilitating focused workshops and knowing how to pushback.

Okay, so there you have it. Three recent conversations that pushed me to write this blog and some of the things to look for.

One more thing, if you are the buyer, here something else to be aware of. I am seeing more and more polished content on social media, more people teaching Power BI in X days, more people being endorsed by the communities they are active in and more people positioning themselves as experts overnight.

That does not automatically mean they are not good. But it does mean you need to go into the market with better questions.

Because if you are looking for genuine Power BI or MS Fabric expertise, you do not just want someone who can post well or demo something shiny from a script. You want someone who can help you create value, navigate complexity and build something that actually helps your business move forward. It is about finding a consultancy, a partner who can help you create something that genuinely moves a business objective forward, DELIVERS VALUE and when needed, has the honesty to tell you not to build anything yet. And yes, I have done this over the years, and I will continue to do it. More recently, we had a potential client we could have signed, but after some thought, we recommended another partner with deeper Microsoft specialism in other areas to take it on instead.

Why This Matters More Now

Power BI (and other tech) has become easier to pick up and easier to showcase. That is not a bad thing. But it does mean more people can sound credible after a few builds or a certification. For a buyer doing due diligence, that creates a problem. The shortlist starts to look... well the same.

There is another layer to this as well. More people can now produce decent looking solutions without really understanding what sits underneath it. That matters, especially if you are the buyer. At minimum, ask them to show you the actual report and walk you through it properly. You will be surprised how many Power BI examples online are really just AI-produced images with a Power BI logo slapped on them, not real solutions anyone has built, used or delivered.

But, let me be clear. AI is brilliant for speed, productivity and reducing friction. Used well, it makes good consultants better. But it does not remove the need for proper grounding. If someone is relying on AI to fill gaps in first principles they never really learned, that is not acceleration.

This is why generic questions do not help much when looking for good Power BI resource. Asking whether someone knows Power BI, has built reports and dashboards before or can work with stakeholders will get you polished answers from almost anyone. The better approach is to ask questions that reveal judgement.

Start Here: Listen Before You Ask Anything

Before the questions themselves, listen to how they frame the work. A weaker consultant will usually centre the conversation around tooling. So, reports, dashboards, features, visuals, capabilities. Maybe some Fabric terms for extra effect.

A stronger consultant will usually start elsewhere. Audience, business objectives, decisions, process bottlenecks, data quality, ownership, adoption and what success actually looks like.

They also tend to work with the client, not just for them. That means helping drive the conversations that need to happen, not disappearing into the background and resurfacing later with a reporting solution. A genuinely good consultant gets properly involved in the business context, because that is usually where the real issues are.

That does not mean technical depth is optional. It means technical depth on its own is not enough. Plus, technical depth does not mean being able to produce something that works today. It means understanding why it works, what assumptions it depends on and what will happen when the model, the data or the business questions get more complex.

8 Questions That Reveal Judgement, Not Just Competence

1. How do you start a Power BI project when the requirements are still vague?

This is a very useful question because most real projects do not begin with perfect clarity. A good answer usually includes some version of this thinking: identify the audience, understand the business decisions involved, clarify the most important measures, challenge assumptions, prioritise a first use case and avoid overbuilding too early. In fact, ask them to walk you through their methodology and process, how they applied it to real client scenarios.

A weaker answer often leads with simply calling a meeting, waiting for the right person to be pointed to them and asking "what do you want" style questions like what would you like to see, what visuals, do you have an existing report we can view, etc.

What you are listening for is whether they know how to turn messy and vague requirements into a focused solution. Whether they can facilitate large workshops with senior stakeholders, handle objections, drive discussions and get the outcomes needed to move into build. 

2. Tell me about a Power BI project that became more difficult than expected. What happened?

Most consultants can talk about success. This question tells you far more.

A good consultant will not give you a polished fairy tale. They will usually talk openly about ambiguity, stakeholder misalignment, data issues, changing priorities, access delays, conflicting KPI definitions or adoption problems. More importantly, they will explain how they navigated it.

A red flag answer either avoids the question, blames the client for everything or pretends nothing meaningful has ever gone wrong. I mean come on, something always goes wrong or at least not to plan. The best consultants respect complexity and they do not act surprised by it.

3. How do you decide what should be handled in Power BI and what should not?

This is one of the better filters because it exposes whether someone can think beyond the tool.

Good answers acknowledge trade-offs. They might talk about when logic belongs upstream, when a semantic model is being overloaded, when report interactivity is hurting usability, when ingesting and authenticating agaisnt a API is not secure or when self-service ambitions need governance rather than encouragement.

Poorer answers often sound like this: yes, we can do that in Power BI. That is not always the answer you want. Sometimes you may need to use an Azure Function or a Dataflow Gen 2 or something else. Also, to add a bit more, good answers also touch on data modelling decisions, for example choosing an appropriate grain, structuring a clean star schema, avoiding unnecessary many-to-many relationships and knowing when logic should sit upstream rather than being forced into the semantic model. The semantic model is now more important then ever with AI being in the mix. At Metis BI, we do not agree with any idea that the semantic model is no longer needed or loses it's importance due to AI and NLP.

4. What would make you push back on a client request?

In workshops I run, this is where you start to see whether someone is there to advise or just to comply.

A good consultant will have no issue describing situations where they would challenge the request. Maybe the measure is poorly defined, the audience is too broad, the timeline is unrealistic, the business is asking for pixel-perfect visuals before agreeing what decisions the report should support.

The key thing is this. Good consultants do not just nod along, take the brief and hide behind "well, that is what was requested" later on. They ask why. They challenge early. They would rather surface friction at the start than leave you paying for it later.

A weaker consultant often avoids pushback because they think agreement sounds commercial. It does in the short term. It does not in delivery. I like to say, you are not paying us to come in and just agree with everything.

5. How do you handle competing stakeholder opinions on what the report needs to do?

This matters because many reporting projects fail quietly at this point.

Good answers usually involve structured facilitation, bringing people back to the audience and objective, separating must-haves from preferences and creating a basis for decision-making rather than endless opinion trading.

Weak answers drift toward gathering every request and trying to include everything to satisfy everyone. That is not stakeholder management, this is backlog inflation.

6. What do you do if adoption is poor after go-live?

This is where delivery maturity shows. I alwasy say, the work doesn't stop when we publish, in other words give access to the audience or end users.

So, a strong consultant will not treat go-live as the finish line. They will talk about relevance, usability, trust in the numbers, training, ownership, feedback loops and whether the solution is actually embedded into business routines.

They will also talk about making your team stronger, not more dependent. That usually shows up in the way they explain things, involve your people, transfer knowledge and help build confidence inside the organisation rather than keeping all the thinking to themselves.

A weak answer tends to frame low adoption as a user problem. Here is the thing. If a report is technically correct but operationally ignored, the project is still underperforming.

7. Can you walk me through a past solution and explain the insight it gave the audience it was built for?

This is one of the most useful things you can ask.

Do not just ask to see a portfolio or for them to send you a few images. Ask them to demo a solution and explain what it actually helped the intended audience understand, decide or improve. That changes the conversation completely.

A weaker consultant will often stay close to the surface. They will show you visuals, interactions, layout choices, bookmarks, drill-through, maybe a few bits of DAX. None of that is bad. But on its own, it does not tell you whether they know how to embed value into a reporting solution.

A stronger consultant will explain who the audience was, what challenge they were dealing with, what decisions needed to improve, KPIs they needed to track that was aligned to the business objective and how the solution helped move that forward. They will talk about the thinking behind it, not just the appearance of it.

Think about it. We are now at a point where AI can help people build reports quickly and make them look polished. That is useful. But a polished report is not the same as a valuable one.

You do not want reporting for reporting’s sake. That is what we use to do and now with AI it's coming to an end - thankfully. You want a solution that genuinely helps your business progress an objective, overcome a challenge or make better decisions with more confidence.

That is why this question matters so much. It exposes whether someone builds dashboards or whether they build reporting solutions with purpose.

8. Show me your most visually impressive dashboard, live and talk me through why it worked.

Yes, insights come first. Yes, business value should always be front and centre.

But let us not pretend the visual side does not matter - IT REALLY DOES!

A genuinely good Power BI consultancy should be able to show you at least one proper showstopper. Something visually strong, polished, memorable and clearly built with care. One that truly stands out compared to all the rest you see online. Not because visuals are the goal on their own, but because strong visual design is often part of what gets people to engage, explore and actually use the solution.

The key is how they present it. Do not let them talk you through screenshots. Do not let them hide behind a static portfolio. Ask them to show it live, in action and explain why it was designed that way, who it was for, what the audience needed from it and how the visual experience supported the insight rather than distracted from it.

A weaker answer will stay at the level of effects, animations, slick tiles, fancy navigation, dark backgrounds, movement for the sake of movement.

A stronger answer will still appreciate craft, but they will connect the visual choices back to clarity, usability, audience engagement and decision-making. In other words, they will show you something impressive, but they will also prove they knew why it needed to be impressive.

As I said above, this really does matter as AI can now help people produce reports and dashboards that look good very quickly. What you are trying to test is whether the person in front of you understands how to combine visual impact with real reporting value.

When Technical Skill Isn't Enough: The Delivery-Readiness Gap

This can be harder to spot, because on the surface they may still come across as strong. They may know DAX, build attractive pages, and answer technical questions with confidence. But once you listen more carefully, certain patterns start to show.

They talk far more about building than about deciding and they describe requirements as something they receive rather than something they help shape. They tend to use the language of features more than the language of outcomes and they often steer away from the awkward but important subjects like ownership, governance, adoption, poor source data, treating solutions as products or conflicting stakeholder agendas.

You will also notice that they treat almost every problem as a reporting problem and they make hard things sound suspiciously easy. They may be able to produce code or modelling patterns, but struggle to explain the principles behind them. Plus, while they often sound very comfortable with AI-generated answers, that confidence can disappear quite quickly when you start probing things like like filter propagation, evaluation context, why a model has been structured a certain way, how to optimsie for NLP, when to use a lakehuse or warehouse, how to authemticate against a trickly API.

Now look, none of that automatically makes them bad at Power BI. It just means you may be hiring a builder when what you really need is a consultant.

What Good Consultants Actually Sound Like

Good consultants usually sound a little less slick and a lot more grounded and that is often a very good sign. They are comfortable saying "it depends", but they do not leave it there. They explain what it depends on, they bring up risks before they become issues, they talk honestly about trade-offs instead of hiding them and they naturally steer the conversation towards the audience, the decisions that need to improve and what the organisation will realistically be able to maintain after handover.

They also tend to care about what sits behind the report, not just the report itself. So, you will hear them talk about how the model is structured, whether the definitions can actually be trusted, whether governance will hold once usage grows and whether the solution is being built in a way the organisation can genuinely sustain it themsleves. They can talk about failure points without becoming negative, because they are not trying to sound pessimistic, they are trying to stop problems before they become yours.

Plus, let me add one more thing here. The better consultancies do not force everything into a lowest-cost conversation. Cost matters, of course it does, but cost on its own tells you very little. A cheaper option that takes longer, misses the point, needs rework or delivers something that looks good in a demo but falls flat in the real world is not cheaper at all. You need to look at cost in context of speed, quality, judgement, challenge and whether the work is actually going to create value once it lands.

That is usually what strong answers sound like. They are not trying to oversell Power BI, oversell AI or oversell themselves. They just sound like people who understand the work, understand the business context and understand what it really takes to deliver something worth building.

How to Make the Final Call

Do not just ask what they have built and leave it there. Ask how they think when things are unclear, where projects normally go wrong, what they would challenge, how they deal with trade-offs and what they do when adoption is poor after go-live. Ask them to demo a past solution and explain the business insight it produced, not just the features it contains. Ask them to show you their most visually impressive solution live and explain why the visual design actually worked for the audience it was built for. And ask whether their goal is to make your team more capable over time or more reliant on them.

Then, when you get to commercials, do not reduce the decision to day rate or total project cost in isolation. Look at what that cost is actually buying you. Faster clarity, better judgement, stronger challenge, better quality, less rework. A cheaper option that misses the point, takes longer or needs fixing later is not really cheaper at all.

Because the consultancy you choose will shape far more than the report. They will shape the conversations, assumptions and decisions that happen before the first visual is ever built. Opening Power BI is the easy part. Choosing someone who can help you deliver something worth opening is the harder bit.

The best consultancies are not the ones who make everything sound exciting and effortless. They are the ones who stay close to the business problem, challenge weak assumptions, tell the truth early and care whether the solution still makes sense six months after go-live.

Need a Second Opinion Before You Commit?

If you are about to engage a Power BI consultancy, there is real value in pressure-testing the scope before work starts. Sometimes the biggest win is not a better dashboard. It is avoiding the wrong approach, the wrong assumptions or the wrong delivery model in the first place. That is particularly true now, when AI can help people move faster but also makes it easier to sound more capable than the underlying foundations really are. The right partner will not just move quickly, they will know when to slow down, challenge the brief, strengthen the fundamentals and make sure the work is actually set up to deliver value.