Sorted by what the summary formulas actually do, read in the live sheet with the file signed out. Half of them summarise a block of rows that was fixed when the file shipped — type a thirteenth task under their twelve and nothing counts it, and nothing warns you.
That is the whole decision, and it takes three minutes to check on any template before you commit a quarter of reporting to it. The rest of this piece is how I measured it, what the 38 files actually contain, and the check itself.
Where the files came from
I searched the phrase people search, google sheets dashboard template, and took the pages that answer it: Smartsheet, HubSpot, rows.com, datapad.io, Coupler.io, ClickUp, Polymer Search, Spreadsheet Point and six others. Google's own results page refuses automated requests, so the list came from a US-locale search on a second engine and was cross-checked against a third. Those fourteen pages link to 71 distinct Google Sheets files between them. Forty-four are presented as dashboard or KPI templates, and those 44 are the sample.
Six never opened. Five returned a permission error and one is gone, which is worth knowing before you write a listicle into your bookmarks. The remaining 38 opened with no Google account at all, because a spreadsheet shared to anyone with the link can be downloaded through the export endpoint, so I read every one as a file instead of clicking through it.
Those 38 files turned out to be 34 distinct spreadsheets. One customer relationship file is handed out by datapad, HubSpot and rows.com under three different IDs, and two other templates appear twice. Between them they hold 12,039 formulas across a median of four tabs.
=SUM(I27:I) in the live sheet arrives in the downloaded file as SUM(I27:I81), and it does the same inside the formulas it preserves as text. Measured that way every template looks like it has a ceiling, because the exporter puts one there. So the files were used to decide which cells to look at, and every range claim here was then read in the live spreadsheet from the formula bar, signed out: 53 cells across 26 of the 38 files, including every file whose verdict could move. Two templates the download called fixed turned out to be open, and the counts below are the corrected ones.The number that decides whether a template survives your data
Smartsheet's basic dashboard is the clearest example, and it is the first result most people meet. The task table occupies rows 12 to 23, twelve rows of sample tasks. Two rows under it, the summary block begins:
- B27 Not Started C27
=COUNTIF(G12:G23, "Not Started") - B28 In Progress C28
=COUNTIF(G12:G23, "In Progress") - B29 Complete C29
=COUNTIF(G12:G23, "Complete") - B32 TOTAL C32
=SUM(C27:C31)
Every count reads rows 12 to 23. There is no thirteenth row waiting: row 24 is the blank line before the summary block, and row 25 is a heading. The template has room for exactly the number of tasks it shipped with. That is not a bug in Smartsheet's file, it is what a hand-drawn range means, and all seven Smartsheet templates in the sample are built the same way. I read the summary cell of each of the seven in the live sheet, signed out, and the formula bar shows what the file shows.
Across the 38, the split looks like this.
| Where it was listed | Files opened | Every range ends at a fixed row | At least one range grows | Summarises nothing |
|---|---|---|---|---|
| datapad.io | 11 | 3 | 5 | 3 |
| rows.com | 9 | 3 | 2 | 4 |
| Smartsheet | 7 | 7 | 0 | 0 |
| HubSpot | 4 | 2 | 2 | 0 |
| Coupler.io | 2 | 0 | 2 | 0 |
| Polymer Search | 2 | 2 | 0 | 0 |
| Spreadsheet Point | 2 | 1 | 0 | 1 |
| ClickUp | 1 | 1 | 0 | 0 |
| Total | 38 | 19 | 11 | 8 |
Of the 30 templates that summarise anything at all, 19 have no range that grows: every total stops at a cell reference somebody typed once. Seventeen of those 19 have at least one summary range that ends before row 100, and 100 rows is a bit over three months at one row a day, or a fortnight if you log a row per appointment.
Eight of them do not calculate anything
The other end of the range surprised me more. Eight of the 38 contain no formula that summarises a range at all: no SUM, no COUNTIF, no QUERY, nothing that turns rows into a number. Two of the eight contain no formulas whatsoever.
- ABM Report (rows.com) has 63 formulas and every one is a cell mirror, in the shape
='Raw Data'!K2. The dashboard copies values across one cell at a time, so whatever aggregation happens, happens somewhere you cannot see. - The Ultimate Ecommerce Metrics Dashboard Template (datapad) contains exactly one formula, and it is a HYPERLINK to a blog post.
- Simple Dashboard Template (Spreadsheet Point) has a dashboard tab with no cell values on it at all, just three charts pointed at a data tab.
- Project timeline and TMR: Klaviyo Overview Report (both rows.com) have zero formulas.
This is the part the screenshots cannot show you. A template that summarises nothing is a layout, and a layout is a real thing to want, but you supply the arithmetic. If you were planning to paste in a month of exports and read a number off the top, that template is not going to give you one.
The file names say the same thing more bluntly than I can. Among the 38 are files called PG TEST 2, TMR: shopify and 2022-2023 Mcubed Annual Budget. One paid-traffic dashboard still has a recruiting checklist sitting in its cells, in Portuguese, left over from whatever the author was doing before. These are working spreadsheets that were shared, not products that were built.
The eleven that grow with you, and what they do differently
Eleven templates use ranges with no end row. The pattern is small and it is the whole difference:
- grows:
=SUM(I27:I) - grows:
=counta(People!C4:C) - grows:
=ARRAYFORMULA(if(F2:F="","", countifs('Data-New Accounts'!D:D, F2:F, ...))) - fixed:
=COUNTIF(G12:G23, "Complete") - fixed:
=SUM(D4:D13)
A range written I27:I means "from row 27 to the bottom of the sheet". Paste in six months instead of one and the total follows. These eleven came from four of the eight sites.
Eleven counts templates with at least one growing range, and the wording is doing real work, because one file often mixes both habits. Two of these sites hand out the same KPI dashboard. Both versions total their column with =SUM(I27:I), which grows, and both fill the tables underneath with a QUERY over the data tab. In one copy that query reads 'Data Sheet'!A1:Z and follows the data down; in the other it reads 'Data Sheet'!A1:Z104 and stops. Same layout, same headline number, different answer three months in.
There is also a trade-off worth saying out loud, because open ranges are not free. A range that runs to the bottom of the sheet picks up anything else you type down there, so a note under the table, or a hand-typed total, quietly joins the sum. The rule that makes it safe is one table per tab, with nothing underneath it.
The three-minute check, on any template
You do not need to read 12,000 formulas. You need four things, and you can do all of them in a copy before you enter a single real number.
Click the headline number and read the formula bar.
If the range ends in a row number, that number is your capacity. If it ends in a bare column letter, the template grows.
Press Ctrl and the tilde key
to show every formula on the sheet at once. Google lists it as "show all formulas" on both Windows and Mac. One glance tells you whether the dashboard computes or copies.
Find where the data is supposed to live.
Median in this sample was four tabs, and the ones I would build on keep raw rows on a tab of their own. If the numbers and the layout share a tab, adding data means editing the design.
Type one row past the sample and watch.
This is the test that settles it, because it also answers the question I could not answer from someone else's read-only file: inserting a row inside a table and typing a row under it are not the same event, and only one of them moves the range end. Add your row the way you actually plan to add rows, then check whether the total changed.
If you think you will move it to Excel later
Fifteen of the 38 use functions that exist only in Google Sheets. SPARKLINE appears 456 times across the sample, SPLIT 60 times and QUERY 36 times. Download one of those files as .xlsx and you get something worth seeing before you rely on it:
=IFERROR(__xludf.DUMMYFUNCTION("""COMPUTED_VALUE"""),1.0)
The value beside the placeholder is what the cell held the last time Google calculated it, and in Excel that is what the cell shows from then on. Sometimes the original formula survives inside the placeholder as text, and sometimes it does not survive at all: across the sample, 366 of these cells carry the literal string COMPUTED_VALUE where the formula used to be, against 319 that keep it. Add the range rewriting described above and the picture is consistent. A Sheets dashboard converts to a photograph of a Sheets dashboard.
If the template pulls from another file
Some of these templates are wired to a second spreadsheet with IMPORTRANGE, and a copy does not inherit the connection. Google's documentation is direct about it: "Spreadsheets must be explicitly granted permission to pull data from other spreadsheets using IMPORTRANGE." Your copy is a new spreadsheet, so it arrives with a #REF! and a prompt to allow access, and if the source belongs to someone else you are waiting on them, not on Sheets.
The same logic applies to anything else the original was authorised to do. A template that promises automatic refreshes is usually promising something the author's account was set up for.
So which one should you copy
If you want a layout to argue with, take the Smartsheet ones. All seven compute, all seven are built the same way, I read the summary cell of every one of them in the live sheet, and their ceilings are visible in ten seconds now that you know where to look. If you want something that keeps working as the data grows, start from the eleven, and check the parts of them that do not grow before you trust the screen as a whole.
What none of them will do is know what your practice should be measuring. That part is not a file problem, and it is why the honest answer to "which template" is usually "whichever one you are willing to rewire".
Related reading on this site: how to build a KPI dashboard in Google Sheets if you would rather start from an empty sheet, what you actually get when you download a dashboard template for the Excel version of the same teardown, and what a good sample KPI dashboard looks like for the layout question.
A fixed range is not the only thing that makes a dashboard lie quietly. So does data that has not finished arriving, which is exactly what Google Ads conversions do.
Frequently asked questions
Not one called a dashboard. Google's help page describes the gallery as offering "Google-created templates like resumes, budgets, and order forms", and that is the shape of it: budgets, invoices, schedules, timelines. Three of the files handed out as free dashboard templates in my sample are titled "Annual business budget", "Project timeline" and "Purchase order", which are budgets and schedules with a dashboard label attached by the site listing them. You reach the gallery from the Sheets home screen by clicking Template gallery at the top right.
They vary more than the listicles suggest. Of the 38 I opened, 8 contain no formula that summarises a range and no pivot table either, so the numbers on them are typed or mirrored from another tab. Of the 30 that do calculate, 19 are pinned to fixed ranges and 11 have at least one range that grows with your data. None of that is visible in a screenshot, and it is the difference between a template you can use for a year and one you outgrow in a month.
Click the number you care about and read the formula bar. A range like G12:G23 counts twelve rows and only those twelve; a range like I27:I runs to the bottom of the sheet. Ctrl and the tilde key shows every formula on the sheet at once, which is the fastest way to see whether a dashboard computes or just copies. Then add one row the way you actually plan to add rows and check that the total moved.
Partly, and less than you would hope if it was built with Sheets functions. Fifteen of the 38 templates I measured use functions Excel does not have, mostly SPARKLINE, and downloading those as .xlsx wraps each one in a placeholder that carries the last value Google calculated. The cell then shows a number that never changes. Open-ended ranges are also rewritten with a fixed end row during export, so a template that grew with your data in Sheets stops doing that in Excel.
Almost always the summary range ended before your new rows started. Select the cell that looks wrong, read the last row number in its range, and compare it with the row your data now reaches. If your rows go further, either widen the range to a bare column letter so it runs to the bottom of the sheet, or add future rows by inserting inside the existing block instead of typing underneath it.
Sharing settings. Of the 44 dashboard templates linked from the pages I checked, 38 are shared to anyone with the link, five require permission from the owner and one has been deleted. A link in a "free templates" roundup is only as durable as one stranger's Drive.
WRITTEN BY
Olha · clinic data analyst
I build the reporting our managers open every morning at a multi-branch medical clinic — and package it so other practices don't have to start from scratch.
Published 13 August 2026. The method here was rebuilt twice. The first version measured the downloaded copy of each template and produced a tidy finding that was an artefact of Google's exporter, which writes an end row into any range that does not have one. Checking a single cell in the live sheet caught that. The second version assumed the formulas the exporter preserves as text were safe to read, and a second live check caught that too: a QUERY over A1:Z reaches the file as A1:Z81. Everything about ranges was then read from the formula bar of the live spreadsheets, 53 cells across 26 files, which moved two templates from fixed to growing and the totals from 21 and 9 to 19 and 11. An earlier pattern-matching bug had also been reading AP2:AP35071 as an open range, and was fixed before any of these counts were taken. Two claims were dropped for lack of evidence: that exporting drops ARRAYFORMULA, which it does not, it converts it, and any statement about what happens when you insert a row inside a table, which cannot be tested on a file you are not allowed to edit and is handed to the reader as a test instead.