How to Connect Excel to a SQL Database

1. Why connecting beats pasting

Before the explanation: think about the last report you rebuilt from an export. How many of those minutes were spent on judgement, and how many on moving data from one window to another?

Pasting is not slow because of the paste. It is slow because the paste is only step four of seven, and the other six get repeated every single time: run the query, export it, open the file, fix the columns that came in as text, paste into the right place, check the formulas still point at the right range, re-apply the formatting.

A connection replaces all seven with one button, because Power Query records the whole sequence as steps and replays them. And it removes an entire class of error rather than just saving time. A paste that lands one column to the left produces a report that is wrong and looks completely normal, which is the worst kind of wrong there is.

2. Find Power Query, whatever your Excel looks like

Power Query is built into Excel and has been for years, but it hides under different names depending on how old your version is. On the Data tab, look for one of these, in this order:

• Get Data, on the left of the Data tab. This is the current name and what the rest of this page uses.
• New Query, in older versions. Same tool, same menu underneath.
• A Get & Transform group with buttons rather than one menu. Also the same tool.

If you find none of these, you are probably on Excel for Mac or a very old version. Mac Excel has fewer connectors and a different arrangement; older versions may need the Power Query add-in installed separately. Section 5's advice about exporting a file is the route that works everywhere.

3. Connect to SQL Server, Postgres or MySQL

The path is Data, Get Data, From Database, then your engine. Take SQL Server as the worked example, because it is the one most workplaces have.

A dialog asks for a Server and, optionally, a Database. Server is the machine name your colleagues use in their own connections, which is the fastest way to get it right: ask for a screenshot of someone's working connection rather than guessing. Leaving Database blank shows you everything on that server you are allowed to see, which is a useful way to look around the first time.

Then it asks how to sign in. Windows means your existing network account and needs no password typed anywhere, which is both easier and safer. Database means a separate username and password issued for that server. Use Windows unless you were specifically given credentials.

Excel then shows a Navigator listing the tables and views. Click one to preview it, tick it, and you are one dialog from finished. Prefer a view over a raw table when one exists with a sensible name, because a view is usually somebody's tidied, agreed version of that data, and using it means your numbers match theirs.

4. Load it, and the choice in that dialog

The Navigator offers Load and Transform Data, and the difference matters more than it looks.

Load drops the table onto a worksheet as it is. Transform Data opens the Power Query editor first, where you can remove columns, filter rows, rename headers, fix data types and split fields. Every action you take is recorded as a step in a list on the right, and that list replays on every future refresh.

That list is the real prize. Cleaning done in the editor happens again next month automatically; the same cleaning done by hand in cells has to be redone by hand next month, by whoever is holding the file. Choose Transform Data by default, even if the only thing you do is remove three columns you do not need.

When you press Close & Load, you also get a choice of where it lands. A Table on a worksheet is right for anything you want to look at. Only Create Connection is right when the data is only feeding a PivotTable, and it keeps a million rows out of your worksheet while still being fully usable, which is the trick behind spreadsheets that stay small and quick.

5. SQLite, and the ODBC gap

Say why a database might be missing from a menu that lists a dozen others, before reading on.

The menu lists the connectors Microsoft chose to build and maintain. SQLite is not one of them, which surprises people who have just learned SQL on SQLite because a tutorial sensibly started there. Nothing is broken and you have not missed a setting.

There are two honest routes, and for most beginners the second is better.

Install an ODBC driver. ODBC is a translation layer: install a small driver for your database, register it once, and it then appears in Excel under Get Data, From Other Sources, From ODBC. It works well and it needs permission to install software, which on a work machine often means a ticket.

Export the answer and connect to the file. Run your query in your database tool, save the result as CSV, and use Get Data, From File, From Text/CSV. You keep the refresh mechanism, because re-running the query and overwriting the same file makes Refresh work exactly as before. What you give up is the automatic part.

The second route sounds like a defeat and usually is not, because you should be summarising in SQL anyway. Which is the next section.

6. Let the database do the heavy lifting

The most common mistake here is connecting to a raw transaction table with millions of rows and then grouping it in Excel. It works until it does not, and the failure is a spreadsheet that takes two minutes to open and eventually refuses to.

Databases are built for exactly this and Excel is not. Summarise first, then connect. One row per day per product instead of one row per sale usually turns millions of rows into thousands, and every chart you were going to build works identically on the smaller table.

You can write that summary directly into the connection: the SQL Server dialog has a SQL statement box for exactly this purpose, and whatever you put there runs on the server before anything travels to your machine.

SELECT order_date,
       product_category,
       COUNT(*)     AS orders,
       SUM(amount)  AS revenue
FROM sales
WHERE order_date >= '2026-01-01'
GROUP BY order_date, product_category;

If that query is unfamiliar, GROUP BY and HAVING is the guide for it. One caution worth knowing: a query typed into that box is fixed text, so Excel cannot be clever about folding your later filters back into it. That is a fair trade when the query is already doing the summarising, and a poor one if you paste in SELECT *.

Now picture the report you rebuild most often. At what grain is one row of its source data, and does anything in the finished report actually need that grain? If the answer is no, the summary belongs in the query.

7. Refresh, and what breaks on other people's machines

Refreshing is Data, then Refresh All, or right-click the table and choose Refresh. To make it automatic, right-click the table, choose Table Properties or Connection Properties, and tick Refresh data when opening the file.

Then send that file to a colleague and watch it fail, which it will, for a reason worth understanding. The workbook stores the address of the database and the shape of the query. It does not store your permission to read it. Your colleague needs their own access to the same server, and if they are outside the office they may need the company network as well.

So a connected workbook is a tool for people who can reach the source. For everyone else, send them the numbers instead: paste the finished summary as values into a fresh sheet, or export a PDF. That is not a workaround, it is the correct answer to a different question.

What goes wrong, and the fix

Six that account for most of the pain.

"Cannot connect" or a login failure. Ask a colleague for the exact server name from their working connection. Guessing at server names is how afternoons disappear, and the name often includes an instance or a port that nobody mentions out loud.

Your database is not in the menu. Only built-in connectors appear. Use ODBC with a driver, or export to CSV and connect to that. Section 5.

Refresh is very slow. You are pulling more rows than you need. Filter and group inside the query rather than after it, so the work happens on the server.

Leading zeros disappeared from codes. A column of postcodes or account numbers arrived as numbers. Set the column type to Text in the Power Query editor, not afterwards in cells. The CSV import fix covers the same problem coming from files.

Dates are wrong by a month, or refuse to sort. Day and month have been swapped, or the column is text that looks like a date. Set the type explicitly in the editor and pick the right locale when it asks.

It works for you and not for anyone else. Permissions, not the file. Section 7.