Adding Comments in Fabric IQ Planning - Making Budgets a Conversation
Ask any finance team in Australia how budget season goes and you'll hear a version of the same story. The numbers live in one spreadsheet. The explanations live in emails, Teams chats, sticky notes and the heads of three people, one of whom is on leave. By the time the CFO asks "why is marketing up 18% in Q3?", nobody can find the answer without a round of messages.
That's the problem comments in Fabric IQ planning are meant to fix. The idea is simple: put the explanation next to the number. When someone adjusts a forecast, they leave a note on the cell saying why. When a reviewer has a question, they ask it on the cell. The conversation stays attached to the data instead of drifting off into someone's inbox.
Microsoft's documentation on adding comments in Fabric IQ planning covers the mechanics. I want to cover how this plays out in practice, because we've now set up planning workflows for a few clients and comments end up mattering more than most people expect.
Quick context on Fabric IQ planning
If you haven't looked at it yet, Fabric IQ planning brings budgeting, forecasting and what-if planning into Microsoft Fabric. You build planning sheets on top of your semantic models, users enter or adjust numbers, and those values get written back so they can be reported on alongside actuals. It sits alongside the other Fabric IQ pieces like ontologies and data agents.
The pitch is that you stop exporting data to Excel, planning offline, then importing it back in. Everything happens against governed data in Fabric. Comments are one of the features that make that pitch believable, because a planning tool without collaboration is just a fancier spreadsheet.
How comments work
At a high level, you can attach comments to cells and rows in a planning sheet. A user selects the cell, adds a comment, and it's saved against that data point. Other users with access can see the comment, reply to it and build up a thread. Comments are visible as indicators in the sheet so reviewers can spot which numbers have context attached.
The useful bit is that comments are tied to the data context, not just a screen position. So a comment on "Marketing, Sydney, March forecast" stays with that intersection even if the sheet layout changes or someone sorts and filters differently.
The doc covers the specific steps and options. I won't repeat them click by click, because the UI keeps evolving and the doc will stay more current than this post.
Where we've seen comments earn their keep
Variance explanations. This is the big one. Every monthly review, someone asks why actuals differ from forecast. With comments, the person responsible for the line writes the explanation once, on the cell, and everyone sees it. One of our clients, a professional services firm, used to spend most of a day each month chasing variance commentary from cost centre owners by email. Moving that into cell comments didn't make the owners any keener to write explanations, but it did make it obvious who hadn't. Visibility does a lot of the work.
Assumption tracking. Forecasts are full of assumptions. Headcount growth, price increases, a big contract that might land in February. When those assumptions live in a comment next to the number, the next person to touch the forecast knows what they're looking at. When they live in someone's head, the forecast becomes untouchable the moment that person leaves.
Review and approval. Reviewers can raise questions directly on the numbers they're questioning. "Is this including the new Melbourne office?" sits right there on the cell. The submitter answers in the thread. Done. Compare that with an email chain where someone screenshots a section of a spreadsheet and draws a red circle on it.
Audit trail. For organisations that need to show why decisions were made, comments give you a record attached to the data. It isn't a full audit system and I wouldn't sell it to an auditor as one, but it's far better than nothing.
What to watch out for
I like this feature. I also have a few honest reservations.
Comments need conventions. Without some agreement on how to use them, you get noise. People write "updated" or "see email" or "per Dave." That's worse than useless, because it looks like there's an explanation when there isn't. Set a simple standard before rollout: what changed, why, and any source. One or two sentences. Show people a good example and a bad example.
Don't let comments become the workflow. Comments are good for context and questions. They're not an approval engine. If you need formal sign-off steps, deadlines and escalations, you need a proper workflow on top, whether that's Power Automate, a planning process with clear stages, or something else. I've seen teams try to run approvals entirely through comment threads. It works for about two cycles and then falls apart when someone misses a reply.
Notifications matter. A comment nobody sees is a comment nobody acts on. Check how notifications behave in your setup and make sure reviewers actually get alerted. If people have to go hunting for new comments, they won't.
Think about permissions. Comments are attached to data, so think about who can see what. If your planning sheet has row-level security, consider whether comment content could reveal something to a user that the numbers themselves wouldn't. In most cases this isn't a problem, but in HR or salary planning it absolutely can be. Someone writing "increase reflects Jane's promotion" on a cost centre line is sharing information that might not be appropriate for everyone with access to that sheet.
It's still a newer product. Fabric IQ planning is moving quickly, and some of the edges are still being smoothed. Check the known limitations before you commit to a design, and test comment behaviour with your actual data volumes and security setup. Don't assume.
How we set this up for clients
Our usual approach for a planning rollout looks roughly like this:
- Agree on which cells or rows require a comment. Usually any manual adjustment above a threshold, say 10% or $50k, whichever matters for the business.
- Write a one-page commenting guide. What a good comment looks like. Who answers questions. How long people have to respond.
- Pilot with one or two cost centres for a single cycle. Watch how people use comments, then adjust.
- Build a report showing comment coverage. Which significant variances have explanations, which don't. This is the bit that changes behaviour, because now it's measurable.
- Roll out more broadly once the conventions stick.
That fourth step is the one most people skip. If you can see which cost centre owners are leaving their variances unexplained, you can have a direct conversation. If you can't, you're back to chasing by email.
Where AI comes into it
This is where it gets interesting for us. Once explanations live as structured comments attached to data, they become something you can actually work with. You can summarise all variance commentary for a division into a single paragraph for the board pack. You can flag comments that don't actually explain anything ("updated per discussion"). You can spot common themes across cost centres, like everyone blaming the same supplier price increase.
We've been building this kind of thing for finance teams, pairing Fabric data with AI agents that draft commentary summaries and highlight gaps. It's a practical use of AI that saves real hours at month end, and it works because the comments sit right next to the data instead of scattered through email. If that's something you're exploring, our Microsoft Fabric consultants and our AI agent development team work on exactly these sorts of projects. We also do a fair bit of this for financial services organisations, where explaining numbers is half the job.
My take
Comments are not the headline feature of Fabric IQ planning. Writeback, driver models and forecasting get more attention, and fair enough. But in our experience, comments are what turn a planning tool into something finance teams actually prefer over Excel. The numbers were never the hard part. The hard part was knowing why the numbers are what they are.
If you're rolling out Fabric IQ planning, don't treat comments as an afterthought. Set conventions early, make coverage visible, and keep approvals in a proper workflow. Do those three things and you'll spend a lot less time in March chasing explanations for numbers somebody changed in January.