Dynamic Format Strings for Power BI Measures - One Number, Many Formats

August 31, 2026•6 min read•Michael Ridland

Here is a problem that used to have no clean answer. You build a report with a slicer that lets people flip between "Revenue", "Growth %" and "Units Sold", all driven off one selector. The chart updates fine, the numbers are correct, and then the whole thing looks broken because your growth percentage is showing as "0.14" instead of "14%", or your revenue is missing its dollar sign, or your unit count has two decimal places for no reason. One measure, one format string, three metrics that each want to be shown differently. For years the workaround was ugly, and most people just lived with it.

Dynamic format strings fix this properly. Power BI now lets you drive a measure's format with a DAX expression instead of a fixed setting, so the same measure can render as currency, a percentage, a whole number or custom text depending on the filter context. Microsoft's documentation on dynamic format strings walks through the mechanics. I want to talk about where this actually earns its keep on real client reports, because it solves a specific class of annoyance that comes up more than you would think.

What it actually does

A normal measure has one format string. You set it once, in the measure's properties, and every value that measure produces gets formatted the same way. Dynamic format strings break that link. Instead of a static setting, the format itself becomes a DAX expression that gets evaluated in context, and whatever format string it returns is applied to the result.

There are two places this shows up. The first is on any regular measure: you select the measure, switch its format to "Dynamic", and write a DAX expression that returns a format string. The second, and the one that started it all, is calculation groups in the model, where dynamic format strings have been available longer through the format string expression on the calculation item. Both do the same job. You are writing logic that decides the format, not just the value.

The case it was built for

The classic use is a field parameter or a disconnected slicer that switches what a chart is measuring. Say you give users a selector with three options and a measure that returns the right value for each:

Selected Metric =
SWITCH(
    SELECTEDVALUE('Metric Selector'[Metric]),
    "Revenue", [Total Revenue],
    "Growth %", [Revenue Growth %],
    "Units", [Total Units]
)

That returns the correct number every time. The trouble is the format. Revenue wants "$#,##0", growth wants "0.0%", and units wants "#,##0" with no decimals. A single static format cannot serve all three. So you write a matching format expression:

SWITCH(
    SELECTEDVALUE('Metric Selector'[Metric]),
    "Revenue", "\$#,##0",
    "Growth %", "0.0%",
    "Units", "#,##0"
)

Now the chart shows "$1,240,000" when Revenue is picked, "14.2%" when Growth is picked, and "8,430" when Units is picked, all from the one measure. This is the thing that used to be genuinely hard, and it is now a five-minute job. The first time you show a client this working, they visibly relax, because they have usually been staring at a slightly wrong-looking dashboard for weeks and assuming it was just how Power BI worked.

Beyond the switch trick

The switch pattern is the obvious one, but the more interesting uses are subtler.

You can scale a number based on its size. Large financial figures read better when they collapse to thousands or millions, and you can format that conditionally:

VAR v = [Total Revenue]
RETURN
SWITCH(
    TRUE(),
    v >= 1000000, "\$#,##0,,\" M\"",
    v >= 1000, "\$#,##0,\" K\"",
    "\$#,##0"
)

A value of 2,400,000 shows as "$2 M", 15,000 shows as "$15 K", and 320 shows as "$320". On an executive summary where the exact digits do not matter but the order of magnitude does, this is much easier to read than a wall of long numbers. I use this a lot on board-level pages where the point is the shape of the business, not the cents.

You can also format by currency when a report covers multiple regions. If your model knows a row is in AUD, USD or NZD, the format string can put the right symbol in front without you splitting the measure. That one is genuinely useful for the multi-market clients we work with, where the same revenue measure needs to display honestly depending on which market the user filtered to.

Where it goes wrong

I have cleaned up a few of these, so here is what to watch.

The biggest trap is that a dynamic format string has to return a valid format string for every possible context, including the ones you did not think about. If your SWITCH has no fallback and the selector is cleared, the expression can return blank, and your beautifully formatted number falls back to a raw unformatted value or something that looks off. Always give the expression a sensible default branch, and test it with nothing selected and with multiple things selected, not just the happy path where exactly one option is chosen.

The second is escaping. Format strings use characters like the dollar sign and backslash that mean something to DAX, so you end up escaping them, and it is fiddly. It is easy to get "$#,##0" subtly wrong and not notice until a stakeholder points at a number missing its symbol. Build the format expression slowly and check each branch against real values.

The third is that this feature makes it very tempting to cram three unrelated metrics into one visual. Just because you can flip a chart between revenue and a percentage does not always mean you should. Sometimes two clear charts beat one clever shape-shifting one, especially for an audience that is not report-savvy and will not notice the axis meaning changed underneath them. The technology being able to do it is not the same as it being the right design. Use judgement, not just capability.

The fourth is performance, though it is minor. The format expression evaluates alongside the value, so on a very heavy page with a complex format expression you are adding a little work. For a simple SWITCH this is nothing. Keep the format logic light and you will never notice it.

Where this fits

On its own, dynamic format strings are a small feature that removes a specific, nagging kind of ugliness. In the context of a report built carefully, they are one of the touches that separate work that looks considered from work that looks like a first draft. A single number showing in the right units, with the right symbol, formatted the way a human would expect, is the sort of detail people do not consciously notice but absolutely feel.

This kind of thing comes up constantly in the report work our Power BI consultants do. The DAX to make it work is rarely the hard part. Knowing when a shape-shifting measure genuinely helps versus when it just confuses people, and getting the format right for a mixed-currency, mixed-metric audience, is the part that takes experience. We treat report design as a real discipline, which is also why a lot of our broader AI and data work starts with getting the reporting layer honest and readable before anything clever gets built on top of it.

If you have Power BI reports where the numbers are right but they look slightly wrong, or you want reporting built properly from the start, that is squarely what we do. Take a look at our services or get in touch and we will have a look at what you have got.