The Power BI Revenue Opportunities Sample - What to Steal and What to Ignore

Every sales leader I have ever sat across from wants the same thing from their reporting: not a list of deals, but a read on which deals are actually going to close and what to do about the ones that might not. That is a harder question than most pipeline dashboards answer. Most of them show you the pipeline as a total and call it a day. The good ones tell you where the risk sits and where the upside is hiding.

Microsoft ships a sample report called Revenue Opportunities that is a decent teaching tool for exactly this problem. It is one of the built-in samples you can open inside Power BI without wiring up a single data source, and Microsoft's page for it is the Revenue Opportunities sample. I point clients at it fairly often, not because the report itself is something you would ship, but because it demonstrates a few pipeline modelling ideas that most homegrown sales dashboards get wrong. This post is the version I would give a colleague: here is what to take from it, and here is what to quietly leave behind.

What the sample actually is

The Revenue Opportunities sample is a fictional dataset built around a company selling through a network of resellers. It tracks opportunities through a sales funnel, with each opportunity carrying a revenue figure, a stage, a probability of closing, a region, a product category, and the partner or reseller attached to it. The report layered on top slices all of that a few different ways: revenue by sales stage, opportunity count by region, the mix of new business versus existing, and a breakdown by partner type.

On the surface it is an ordinary sales dashboard. What makes it useful as a reference is the modelling underneath. The data is shaped so that a single opportunity can be viewed as a raw dollar value, as a probability-weighted value, and as a count, and the report moves between those three lenses cleanly. That sounds obvious written down. In practice it is the single thing most sales reports fail to do, and getting it right is most of the value here.

The pattern worth stealing: weighted pipeline

Here is the idea the sample gets right and most teams get wrong. A pipeline number on its own is close to meaningless. If a rep has ten million dollars of "open" opportunities, that tells you almost nothing, because a deal at the "qualified" stage with a 10% chance of closing is not remotely the same asset as a deal at "proposal sent" with an 80% chance. Adding them together as raw dollars produces a headline figure that flatters everyone and predicts nothing.

The sample handles this by carrying a probability against each stage and letting you view revenue as a weighted figure. Ten million of raw pipeline might be two million of weighted pipeline once you account for where each deal actually sits. That weighted number is the one a CFO can plan against. The raw number is the one a salesperson quotes in a good mood.

When we build sales analytics for clients, this weighted view is nearly always the first thing we introduce, and it is nearly always the first argument. Sales teams like the big raw number because it looks like progress. Finance wants the weighted number because it is the only one that has ever correlated with what actually landed. The report becomes a negotiation between optimism and history, and the honest answer is you need both figures on the page, side by side, so the gap between them is visible. The gap is the story. A team whose raw and weighted numbers are miles apart has a pipeline stuffed with early-stage hope. A team where they track closely has a pipeline you can bank on. That single comparison is worth more than most of the fancier visuals people ask for. If you want a steer on getting this foundation right in your own environment, it is the kind of thing our Power BI consultants spend the first week of an engagement nailing down before anyone builds a chart.

The second pattern: funnel shape over funnel total

The sample also does a nice job of showing the pipeline as a shape, not just a sum. A funnel visual that shows how many opportunities and how much value sit at each stage tells you something a total cannot: whether your pipeline is healthy or clogged.

A pipeline that balloons at the early stages and thins to almost nothing at the closing stages is a warning, not a triumph. It usually means a lot of deals are getting created and very few are progressing, which points at a qualification problem or a sales process that stalls somewhere specific. The sample lets you see that shape at a glance. In real engagements, the stage where deals go to die is often the most valuable finding in the whole project, and it only shows up when you look at the funnel as a profile rather than a number. We have had clients discover that 70% of their stalled value was sitting in a single stage that everyone assumed was a formality. Nobody had ever charted it that way.

Where the sample quietly misleads

Now the honest part, because I would not be doing my job if I just told you it was great.

The first thing to watch is that the probabilities in the sample are clean, fixed, and tied neatly to stages. Real pipelines are not like that. In the sample, "proposal" always means the same probability. In your business, a proposal to a repeat customer you have sold to for years is a very different bet from a proposal to a cold prospect you met at a conference, even though both sit at the same stage. Stage-based probability is a starting point, not a truth. The better organisations we work with eventually move to probabilities informed by their own historical close rates by segment, by rep, by deal size, rather than a flat number bolted to a stage name. The sample cannot show you that because its data is too tidy. Do not mistake the tidiness for realism.

Second, the sample presents a snapshot. It shows you the pipeline as it is right now. What it does not show, and what almost every real sales question actually needs, is movement over time. Is this deal advancing or has it been stuck at "negotiation" for four months? A snapshot cannot tell you that a deal has been "closing next month" for six consecutive months, which is one of the most reliable signals that it is never closing at all. Any pipeline report you actually deploy needs a time dimension and the ability to track how individual opportunities move between stages. That is a meaningfully harder data modelling job than the sample lets on, because it means capturing the history of each opportunity, not just its current state. It is the difference between a photo and a film, and the film is where the insight lives.

Third, and this is a general caution with all the built-in samples, the data model is small and forgiving. It has been shaped to make the visuals look good. Your CRM export has not. Real sales data arrives with duplicate opportunities, deals with no owner, currencies that need converting, resellers whose names are spelled three different ways, and a "close date" field that half the team treats as aspirational. The sample teaches you what a clean pipeline model looks like. It does not prepare you for the two weeks of unglamorous data work it takes to get your actual pipeline into that shape. That gap between the demo and the reality is where most sales analytics projects either succeed quietly or fail loudly, and it is a big part of what a business intelligence engagement is actually spending its time on.

How I would use it

Treat the Revenue Opportunities sample as a reference model, not a template. Open it, poke at how the weighted revenue is built, look at how the funnel visual is put together, notice how the report lets you pivot the same opportunity between dollars, weighted dollars, and counts. Steal those ideas. They are genuinely good and most people rebuild them badly from scratch.

Then throw away the assumptions. Your probabilities should come from your history, not from a stage label. Your report needs time and stage movement, which the sample does not have. And your real data will need far more cleaning than the sample suggests. If you copy the sample's structure and feed it your messy CRM export, you will get a dashboard that looks professional and lies to you, which is worse than no dashboard at all.

The larger point, and the one I keep coming back to with clients, is that a sales pipeline report is only as honest as the probabilities and the data feeding it. A beautifully designed weighted-pipeline chart built on made-up close rates is a very convincing way to be wrong. Where this gets genuinely interesting is when you start letting AI look at your historical close data and suggest probabilities that reflect what actually happens in your business, rather than a flat number someone picked years ago. That is increasingly where our AI for business intelligence and broader AI strategy work sits: not making the chart prettier, but making the numbers under it trustworthy.

So go open the sample. Learn the weighted-pipeline pattern and the funnel-shape pattern, because they are the two things it teaches well. Just remember it is a clean-room demonstration of a messy real-world problem, and the mess is where your actual project lives.