Free KPI dashboard maker
A KPI dashboard puts the few numbers you steer by on one screen, each with its change since the last period. Paste a CSV with a date column and your metrics, and this tool builds one card per metric from your last complete week, month or quarter, right in your browser.
| Column | Roll up by |
|---|---|
| date | Date |
Oct 2026 vs Sep 2026
Nov 2026 isn't counted because your data stops on Nov 13, before the month ends.
signups
1,459▲ +5.3%
1,386 in Sep 2026
Sum · trend over 5 months
revenue ($)
$27,410▲ +11%
$24,623 in Sep 2026
Sum · trend over 5 months
trial conversion (%)
3.71%▲ +6.4%
3.48% in Sep 2026
Average · trend over 5 months
Why an unfinished month makes your KPIs look like they crashed
The example data in the tool runs from June 1 to November 13, so the cards compare October with September: signups up 5.3%, from 1,386 to 1,459.
Now check "Include the latest month, even if it isn't over". The cards compare November with October, and signups fall 54%, from 1,459 to 678. Nothing went wrong in November. Those 678 signups came in 13 days, and October had 31. Any dashboard that sets this month so far against all of last month shows a drop like this for most of every month.
This tool won't make that comparison unless you ask it to. It skips a period your data stops partway through, and it goes by your rows, not today's date: with daily rows, a month is complete once it has its last day; with monthly rows, every month with a row is complete.
How each KPI card is worked out
- Step 1RowsOne row per date, one column per metric
- Step 2PeriodsRows grouped into weeks, months or quarters
- Step 3Roll-upEach metric summed, averaged or taken last
- Step 4CardLast complete period against the one before
The tool finds the column of dates and the columns of numbers on its own. It reads dates like 2026-10-31, 10/31/2026 and 2026-10, and numbers like 1200, $1,200, 3.4% and (450). Weeks start on Monday; months and quarters follow the calendar.
Then each metric rolls up one of three ways. Pick by what the metric counts:
- Sum for counts and money that add up over time: signups, orders, revenue.
- Average for rates and scores, which can't be added: a 3% conversion rate on 30 days isn't 90%.
- Last value for balances that stand at a point in time: MRR, headcount, subscribers.
The tool guesses each one from the column name and the values (a % sign, or "rate" in the name, gets Average). Change any guess in the table. One catch with Average: a day with 10 trials counts as much as a day with 200, so the average of daily rates can drift from the month's real rate. For the exact figure, divide the month's total conversions by its total trials.
The change is plain percent change: the new period minus the old one, divided by the old one, times 100. When the old value is zero or below, the card shows "n/a". A percent change from zero can't be worked out, and from a negative number the sign flips: going from −10 to −6 gives −40%, though the number went up. Blue means up and red means down, not good or bad, so a rise in churn shows in blue too.
How many KPIs should be on a dashboard?
Fewer than you think. HubSpot's advice is five to nine metrics, max, and that's a good rule for a screen people look at every week. The tool shows up to 12 cards. Keep the ones you'd act on when they move and untick the rest.
One definition per KPI, in a repo your team owns
The CSV you paste here probably came out of a warehouse query. Every choice you just made (the period, the roll-up, whether to count an unfinished month) is part of what the metric means. When those choices live in one person's spreadsheet, the next person makes different ones, and you end up with two numbers for the same KPI.
In Querio, you write those choices down once. Metric definitions are plain files in a git repo your team owns, synced both ways with the app. Dashboards live in the same repo and refresh on a schedule. Open the dashboard, ask the Slack agent for the daily numbers or ask Claude, and you get the same number from the same definition. Each dashboard is marked experimental, team or trusted, and promoting one opens a pull request your reviewers approve.