Report View: Canvas, Visualizations & Filters
Report view is where you will spend most of your time once a model exists. This lesson covers the canvas and its real size, the Visualizations pane and its field wells, and the Filters pane — including the three filter scopes, which is the single most misunderstood thing in Power BI and the cause of a great many wrong numbers.
Learning Objective
Report view looks simple and has three genuine depths to it: the canvas has a real fixed size, field wells determine what a visual can show, and filters operate at three different scopes that behave very differently.
- Explain the report canvas's fixed logical size and why it matters to your reader
- Add, resize, align and layer visuals deliberately rather than by dragging
- Use field wells correctly, including changing a field's aggregation
- Distinguish the three filter scopes and predict which visuals each one affects
- Manage report pages: rename, duplicate, hide, and reorder
Why This Matters
Two failures that come straight out of not understanding this view.
The canvas size problem. You build on a 27-inch monitor with the view set to Fit to page. Everything looks generously spaced. Your reader opens it on a 13-inch laptop and the text is unreadable, because the canvas has a fixed logical size of 1280×720 and everything scales down together. You did not design for a small screen; you designed for a scaled-up version of a small screen.
The filter scope problem. A number looks wrong. You clear the slicers and it is still wrong. The cause is a filter someone left at report level six months ago, silently applying to every page. Nobody looks in the Filters pane because nobody knows there are three sections in it.
Concept
The canvas has a real size
A report page is not an infinite surface. By default it is 1280 × 720 logical pixels — 16:9. What you see is that canvas scaled to fit your window, which is why it looks different on different monitors.
| Setting | What it does | When to use it |
|---|---|---|
| Fit to page | Scales the canvas to fill your window (default) | Building and arranging — you see the whole page |
| Fit to width | Scales to the window width; you scroll vertically | Long-form pages designed to scroll |
| Actual size | Shows the canvas at 1:1 | Checking your work. This is what your reader gets on a standard screen |
Canvas size itself is changed under Format › Canvas settings when nothing is selected. Options include 16:9, 4:3, Letter, Tooltip (a small canvas for report-page tooltips) and Custom.
The Visualizations pane
Three modes, switched by three icons at the top of the pane. Knowing all three exist eliminates most "that feature isn't there" moments.
| Mode | Icon | Contains |
|---|---|---|
| Build | bar-chart icon | The visual gallery, and the field wells for the selected visual |
| Format | paintbrush | Everything about appearance: colours, fonts, data labels, axes, title, background, border |
| Analytics | magnifying glass | Reference lines, constant lines, min/max/average lines, trend lines, forecasting |
Field wells
When you select a visual, the Build mode shows its wells — named slots that determine what the visual can display. Different visuals have different wells, and the well a field goes into changes what it means.
| Well | Effect |
|---|---|
| X-axis / Y-axis | The axes. On a bar chart, the category goes on Y and the value on X |
| Legend | Splits the visual into series by this field, each with its own colour |
| Small multiples | Splits into a grid of small copies, one per value — often clearer than a legend |
| Tooltips | Extra fields shown on hover without cluttering the visual |
| Values | The numbers, on visuals that do not have a value axis (cards, tables, matrix) |
| Rows / Columns | Matrix visual only — what goes down the side and across the top |
| Drill-through | Page-level well, not a visual well. Covered in Module 9 |
Changing a field's aggregation
Every numeric field dropped into a well gets a dropdown. Clicking it lets you change from the default Sum to Average, Minimum, Maximum, Count, Count (Distinct), Standard deviation, Variance or Median.
Count versus Count (Distinct) on an ID field is the specific case that produces the most confidently-wrong reports — Module 0 Lesson 5 showed this costing a factor of two.The Filters pane — three scopes
This is the important part of the lesson. The Filters pane has three sections, and they behave very differently.
Filters versus slicers
A slicer is a visual that filters. A filter in the Filters pane is configuration. Both filter; only the slicer is visible to your reader by default.
| Slicer | Filters pane | |
|---|---|---|
| Visible to the reader | Yes — it is a visual on the canvas | Only if you enable the pane and unhide the filter |
| Reader can change it | Yes | Only if the pane is visible and the filter is unlocked |
| Takes canvas space | Yes | No |
| Best for | Choices you want the reader to make | Constraints that define the report |
Report pages
Pages are tabs at the bottom, like Excel worksheets. Right-click a tab for the options:
- Rename — do this immediately. "Page 1" in a delivered report is sloppy.
- Duplicate page — the fastest way to build a variant, and it copies all formatting.
- Hide page — the page still exists and still works as a drill-through target or a tooltip page. Hiding is not deleting.
- Delete page — permanent, and not undoable after a save.
Visual Explanation
The three filter scopes, shown as what each one actually reaches.
| Filter placed at… | Affects the selected visual | Affects other visuals on the page | Affects other pages |
|---|---|---|---|
| Visual level | Yes | No | No |
| Page level | Yes | Yes | No |
| Report level | Yes | Yes | Yes |
And what the same measure shows under each. Using the course sales data, total revenue is about $6.31M unfiltered:
Canvas size, honestly
| Viewing at | Effective size of 10pt text | Readable? |
|---|---|---|
| Actual size, 1080p monitor | 10pt | Yes |
| Fit to page, 27-inch 1440p | roughly 13pt | Yes — and this is what flattered you |
| Fit to page, 13-inch laptop | roughly 7pt | Marginal |
| Power BI mobile, phone, no phone layout | roughly 4pt | No |
Power BI Demonstration
Open your report from Module 0 Lesson 5, or any .pbix with a few visuals.
NetSales into a card. Click the dropdown next to it in the well. Cycle through Sum, Average and Count and watch the number change dramatically. Set it back to Sum.Channel into Filters on this visual and select only Online. Note that only that visual changes.OrderDate into Filters on this page, set it to a relative date or a range. Every visual on the page responds.Channel into Filters on all pages. Add a second page and confirm it is affected too. Then remove it — and notice how easy it would have been to forget.Example Dataset
This lesson is conceptual and needs no download — it uses the sample data shown inline above. The next lesson that requires a file will link it here.
You can still browse every dataset in the course from the dataset library.
Step-by-Step Exercise
A filter scope drill. Build a two-page report and use it to prove to yourself how each scope behaves.
Expected Result
| Step | What changes | Why |
|---|---|---|
| Visual filter on the page 1 bar chart | Only that bar chart | Visual scope reaches nothing else — not even the card beside it |
| Page filter on page 1 | All three visuals on page 1; nothing on page 2 | Page scope reaches every visual on that page only |
| Report filter on Channel | All six visuals across both pages | Report scope reaches everything |
| Combined on the page 1 bar chart | It shows nothing — an empty visual | The three filters intersect: Channel = Online AND year = 2025 AND Channel = Retail Store. No row can be both Online and Retail Store, so the result is empty |
You should also have noticed how easy it was to forget the report-level filter. In a real report, six months later, with a different person maintaining it, that is exactly how mystery numbers happen.
Common Mistakes
Fix: Check every page in View › Page view › Actual size before you consider it finished. It takes seconds.
Fix: Make checking Filters on all pages part of your pre-publish routine. If a report-level filter is genuinely intended, document it in a text box on the first page.
Fix: Check the dropdown on every field you drop into a well. Better still, fix it at model level with Summarization so it cannot happen again.
Fix: Ctrl+click to multi-select, then Format ribbon › Align and Distribute. Turn on snap to grid.
Fix: Hide filters that are structural and uninteresting (excluding test rows). Never hide one that changes what the numbers represent. If in doubt, state it in the page title.
Professional Tip
Better still, make the title dynamic so it updates with the selection — Module 9 Lesson 5 covers exactly how. Until then, a static title that states the constraint is far better than a title that quietly omits it. A reader who cannot tell what a number covers will either misinterpret it or stop trusting the report; both are bad outcomes and both are avoidable in ten seconds.
Two more habits from professional practice:
- Name every page before you build on it. "Page 1" and "Page 1 (2)" in a delivered report signal carelessness, and page names become the navigation menu in the service.
- Build one visual perfectly, then copy it. Format the fonts, colours, data labels and title on a single visual, then
Ctrl+C/Ctrl+Vand swap the fields. Consistency comes free, and it is far faster than formatting each visual from the default.
Knowledge Check
Knowledge Check
5 questions. Answer each one to see why the right answer is right.
A visual on page 1 shows blank. Its measure works fine elsewhere. What should you check first?
You apply a page-level filter on page 1. Which visuals are affected?
What is the default canvas size for a Power BI report page, and why does it matter?
Where do you find forecasting for a line chart?
When is it appropriate to hide a filter from the reader using the eye icon?
Practical Challenge
Diagnose a report with three planted faults
Build a two-page report from the course sales data, then plant the three faults below. Swap with someone else if you can — otherwise leave it for a day and come back to it. For each fault, write down the symptom a reader would see and the fastest way to find the cause.
- Fault 1: a report-level filter restricting to one channel, hidden with the eye icon
- Fault 2: a card whose aggregation is Count rather than Count (Distinct) on
OrderID - Fault 3: a visual-level filter on one chart that contradicts a page-level filter, producing a blank visual
- For each: write the symptom a reader would notice
- For each: write the fastest diagnostic step
- Rank the three by how dangerous they are, and justify the ranking
For the ranking, think about detectability rather than magnitude. Ask of each fault:
- Would a reader notice something was wrong?
- If they noticed, could they work out what?
- If nobody noticed, what would the consequence be?
The most dangerous fault is rarely the one that produces the biggest number change.
Solution
Try the challenge properly before opening this. Reading a solution you have not attempted feels like learning, but it is not.
Diagnosis and ranking
Fault 1 — hidden report-level filter
Symptom: every number in the report is lower than expected, consistently, across both pages. Nothing looks broken. The report is internally coherent — the channel bar chart simply shows one bar, which a reader may assume is correct.
Fastest diagnostic: open the Filters pane and read Filters on all pages. If the filter is hidden, you need the pane open in Desktop — hidden filters are hidden from readers in the service but still visible to an author.
Fault 2 — Count instead of Count (Distinct)
Symptom: the order count reads 15,067 instead of 7,607. Roughly double. Any average-order-value figure derived from it is roughly halved.
Fastest diagnostic: click the field in the visual's well and read the aggregation dropdown. Alternatively, cross-check against the row count in Table view — if your "order count" equals the table's row count exactly, you are counting rows.
Fault 3 — contradictory filters
Symptom: one visual is blank. Everything else on the page works.
Fastest diagnostic: select the blank visual and read all three sections of the Filters pane. The contradiction will be immediately visible.
Ranking by danger
| Rank | Fault | Why |
|---|---|---|
| 1 — most dangerous | Hidden report-level filter | It is silent and plausible. Every figure is wrong, nothing looks broken, and the report is internally consistent so cross-checking within it will not reveal the problem. It could run for a year before an external reconciliation exposes it. Crucially, a reader has no way to detect it at all. |
| 2 | Count instead of Count (Distinct) | Also silent and plausible, but somewhat more detectable: a business user who knows roughly how many orders they take per month has a decent chance of noticing that the figure is about double. It depends entirely on someone having that domain knowledge and bothering to say something. |
| 3 — least dangerous | Blank visual | It is obvious. A blank visual is visibly broken, so it gets reported and fixed. Faults that announce themselves are the ones you want. |
The general principle
Rank faults by detectability, not by magnitude. A fault that makes a visual obviously break is nearly harmless, because it will be fixed within a day. A fault that produces a confident, plausible, slightly-wrong number is genuinely dangerous, because it reaches decisions unchallenged.
This is the same conclusion as the Module 0 Lesson 5 challenge, arrived at from a completely different direction — and it is why so much of professional BI practice is about systematic validation rather than about building things that look right.
Lesson Summary
- The canvas has a fixed logical size — 1280×720 by default. Check your work in Actual size, not Fit to page.
- The Visualizations pane has three modes: Build (fields), Format (appearance), Analytics (reference and trend lines, forecasting).
- Always check the aggregation dropdown on any field you drop into a well. Sum is the default and is frequently wrong.
- There are three filter scopes: visual, page and report. They combine as an intersection — they do not override each other.
- A forgotten report-level filter is one of the most dangerous faults in Power BI because it is silent, plausible and internally consistent.
- A slicer is a visible filter your reader controls; the Filters pane is configuration they usually cannot see.
- Use Align and Distribute, not dragging. Alignment is the cheapest quality signal available.
- Put the constraint in the title whenever a filter materially scopes a visual.
Syllabus topics delivered
Progress
Marking a lesson complete stores your progress in this browser and updates your BitWithBite learner dashboard. Nothing is uploaded anywhere.