Power BI Dashboard Tiles - How to Use Them Well and When to Skip Them
Dashboards are the part of Power BI that most organisations set up once, in their first enthusiastic month, and then stop thinking about. Two years later there's a dashboard called "Exec Overview" with 34 tiles, five of which are broken, three of which point at a report nobody owns any more, and one which is a picture of the company logo. I've seen that exact dashboard, give or take a few tiles, at a logistics company in Melbourne, a council in regional Queensland and a professional services firm in Sydney.
Tiles are simple. That's their strength and the reason they get abused. Microsoft's documentation page on dashboard tiles covers the mechanics, and I want to add the part the docs don't: how we decide what belongs on a dashboard and how to keep one useful after the launch party.
What a tile actually is
A dashboard in the Power BI service is a single canvas made of tiles. Each tile is a snapshot of something, most often a visual pinned from a report, but it can also be:
- A whole live report page
- A Q&A result
- A visual from an Excel workbook
- An image, text box, video or web content you add directly to the dashboard
- A streaming data tile for real-time feeds
- A tile pinned from another source like a paginated report
The important thing to understand is that a tile is not the visual itself. It's a pointer to the visual plus a cached rendering of it. Clicking a tile takes you somewhere, by default the report the visual came from. That makes dashboards a front door. They summarise, and then they send people to the detail.
Dashboards are also a service-only feature. You can't build them in Power BI Desktop. People coming from Desktop sometimes go looking for them there and conclude dashboards have been removed. They haven't; they live in the browser.
Pinning and what happens afterwards
You pin a visual from a report by hovering over it and clicking the pin icon, then choosing an existing dashboard or creating a new one. Simple enough.
What catches people out is what pinning preserves. The tile keeps the filters and slicer state that were applied when you pinned it. If you pinned the "Sales by Region" chart while the slicer was set to FY26, the tile shows FY26, and it will keep showing FY26 after someone changes the slicer in the report. That is by design, but it surprises users every single time. I've had a CFO ask why the dashboard number didn't match the report, and the answer was that the tile was pinned with a filter on "NSW only" eight months earlier.
Pinning a live report page is different. The tile shows the whole page and stays interactive, including slicers. It's useful when you want one area of a dashboard to behave like a mini report, but it takes up a lot of room and makes the dashboard slower to load.
The tile settings worth knowing
Each tile has a menu (the three dots) with a handful of options. The ones that matter in practice:
Edit details. You can change the title and subtitle, and set a custom link. The custom link is underrated. By default a tile clicks through to its source report, but you can point it at a specific report page, a different dashboard, or an external URL. We use this to send people from a headline KPI straight to the drill-down page that explains it, rather than dropping them on the first page of a 12-page report.
Resize and move. Drag the bottom-right corner to resize. Tiles snap to a grid, which keeps things tidy but also means you can't do pixel-perfect layouts. Don't fight it.
Set alerts. This is the feature that makes dashboards worth having for some teams. On card, KPI and gauge tiles you can set data alerts: notify me when inventory drops below 500 units, or when daily web orders exceed a threshold. Alerts are per user and they fire after the data refreshes. Note the restriction to single-number visuals. You can't alert on a line chart.
Open in focus mode. Expands the tile to fill the screen. Handy for presenting.
View insights. Power BI runs its automatic analysis over the data behind the tile. Occasionally interesting. Mostly I'd ignore it.
Delete tile. Use more often than you do.
When tiles go stale or break
Tiles refresh when their underlying semantic model refreshes, and there's also a tile cache refresh process the service runs. For import models that usually means the dashboard is as fresh as the last scheduled refresh. For DirectQuery and live connection models, the service refreshes the tile cache on a schedule (roughly every hour by default, and you can adjust it in the semantic model settings).
The common causes of broken tiles we see: the source report or semantic model was deleted or moved, someone renamed a field the visual depended on, or row-level security means the viewer can't see the data behind the tile. We wrote a separate post on troubleshooting Power BI tile errors if you're dealing with one of those right now.
When a dashboard is the right tool
My honest opinion is that most organisations don't need as many dashboards as they have. Reports have caught up a long way. You can build a one-page summary report that does almost everything a dashboard does, with better formatting control and real interactivity.
Dashboards still earn their place in a few situations:
Combining multiple reports on one screen. A report is tied to one semantic model (composite models aside). A dashboard can pull tiles from a finance report, an operations report and a customer service report side by side. For an executive who wants a single screen across functions, that's the main reason dashboards exist.
Data alerts. If someone wants an email when a number crosses a line, dashboards are the simplest built-in way to give them that.
Mobile. Dashboards have a phone layout you can arrange separately, and the Power BI mobile app handles them nicely. For field managers who want five numbers on their phone each morning, a dashboard with a tidy mobile layout is hard to beat.
Real-time feeds. Streaming tiles still live on dashboards.
If none of those apply, build a good summary report page instead and skip the dashboard. One less thing to maintain.
Rules we give clients for dashboards that stay useful
These are the habits that separate the dashboards people look at every morning from the ones that rot:
- Keep it to one screen. If you have to scroll, there's too much. Eight to twelve tiles is plenty for most audiences.
- Lead with the numbers that drive decisions. Card tiles for the four or five KPIs that matter, top left, where eyes land first. Charts and context below.
- Retitle every tile. The default title is often the visual's original title, which made sense inside the report and means nothing on the dashboard. "Sum of Net Amount by Month" should become "Monthly revenue (ex GST)".
- Set custom links deliberately. Each tile should click through to the place that answers the next obvious question.
- Note the filter context in the subtitle. If a tile is pinned with "FY26, Retail channel" applied, say so in the subtitle. It prevents most of the "the numbers don't match" arguments.
- Give the dashboard an owner and a review date. Every six months someone should go through and delete anything nobody uses. Usage metrics in the workspace will tell you which dashboards are actually being opened.
That last one is the one that never happens unless someone is named. It's not a technical problem.
Where AI fits now
A quick note, because people ask. Copilot in Power BI is getting better at summarising report pages and answering questions over semantic models, and the Q&A tile was an early version of the same idea. In practice, we find the AI features work best when the underlying model is clean: good names, sensible measures, proper descriptions. A dashboard tile pinned from a messy model gives you a messy tile, and a Copilot summary of a messy model gives you a confident wrong answer. The boring modelling work still pays off first.
If your Power BI estate has grown a crop of dashboards nobody trusts, our Power BI consultants can help audit and rebuild them. For organisations thinking about where AI fits into their reporting, have a look at our AI for business intelligence work, or contact us for a chat.
Reference: Introduction to dashboard tiles for Power BI designers - Microsoft Learn