Labs · Power BI

Lab: Find the Slowest Visual with Performance Analyzer and Make It Faster

By Ravindra Bagale · Last updated: 8 October 2026

Suppose we are…

Suppose the Croma report takes several seconds to open and the director complains. Our 24-row sample is too small to be slow by itself, so we simulate a badly written measure, the kind that on a real 50-million-row table makes a page crawl. Then we do what a performance engineer does: record, find the slowest visual, understand why, fix, and record again. The numbers are sample data made up for practice.

Goal of this lab

By the end you will have:

• Recorded a page with Performance Analyzer and sorted visuals by duration.
• Identified the slow visual and seen that its time is in DAX query, not in Visual display.
• Replaced the slow measure and recorded a big improvement.

What you need (all free)

• Power BI Desktop and a file with pbi_sales, pbi_cities and Total Sales (Lab 11 or later).
• Sample data if you need to rebuild: Download pbi_sales.csv (24 orders) Download pbi_cities.csv
• 30 minutes.

The data: before and after

Before. Example Performance Analyzer results: one visual takes almost 3 seconds, nearly all in the DAX query.

After. The slow measure replaced: the same table loads in under 0.1 seconds.

These are example timings from one laptop. Yours will be different; what matters is which visual is slowest and where its time goes.

The formula

The deliberately slow measure:

Slow Sales =
VAR _busywork = SUMX ( GENERATESERIES ( 1, 3000000 ), MOD ( [Value], 7 ) )
RETURN [Total Sales] + 0 * _busywork

GENERATESERIES ( 1, 3000000 ) builds a 3-million-row table and SUMX loops over it, for every cell of the visual. The answer is the same as Total Sales (0 × anything = 0), but the work is huge. In real reports the same pattern appears as SUMX or FILTER over a whole big fact table when a simple SUM or a CALCULATE filter would do.

The fix:

Total Sales = SUM ( pbi_sales[Sales] )

How to read Performance Analyzer:

Steps

1. Open your file, Save as Lab-27-performance and add a page Perf test.
2. Right-click pbi_sales → New measure and paste the Slow Sales measure.

3. Add a Card (Total Sales), a Bar chart (pbi_cities[City], Total Sales) and a Table (pbi_cities[City], Slow Sales).

   👁 What you should see: the table takes a moment to appear; its numbers are the same as Total Sales (Mumbai 439,000…).

4. Click View → Performance analyzer. In the pane, click Start recording, then Refresh visuals.

   👁 What you should see: one line per visual with a duration in milliseconds.

5. Click the Duration (ms) header to sort, largest first. Expand the top line with the +.

   👁 What you should see: the Table at the top, with most of its time in DAX query (often 1 to 5 seconds, depending on your laptop). The card and bar chart take tens of milliseconds.

6. Under the table's entry, click Copy query. Open DAX query view, paste it and click Run. Read the query: you will see Slow Sales inside it. This tells you which measure to look at.

7. Back in Report view, select the table and replace Slow Sales with Total Sales.

8. In Performance analyzer, click Clear, then Refresh visuals again and sort by duration.

   👁 What you should see: the table now takes about as long as the bar chart (tens of milliseconds). The slow measure is gone from the page.

9. Click Stop. Delete the Slow Sales measure (right-click → Delete from model). Save.

Common mistakes

Self-check checklist

Try-at-home challenge

Turn off Auto date/time for this file (File → Options and settings → Options → Current file → Data load). Compare the file size before and after saving. Why does a file with only 24 rows still change size?

Samjla ka? Measure first: DAX query vs Visual display tells you where to fix. Aata pudhe jaauya: the Blinkit quick-commerce mini project, from raw CSV to report.