How to Use Claude for Excel (Step-by-Step Tutorial)

0:00 I'll show you how to use Claude directly inside Excel to clean data,

0:04 fix formulas, and build models automatically.

0:07 This spreadsheet is a mess.

0:10 Broken formulas, inconsistent data.

0:12 Let's use Claude to fix it.

0:15 This video is sponsored by Anthropic, the team behind Claude.

0:18 Hi, I'm Kevin, and let's dive in.

0:21 Let's start by getting Claude set up in Excel.

0:23 This only takes about a minute.

0:26 First, open Microsoft Excel, and make sure you're on the Home tab.

0:30 All the way over on the right-hand side, you'll see an icon for add-ins.

0:33 Let's click on that, and here at the top, let's search for Claude.

0:37 You should see it as the best match.

0:39 Then, over here, click on Add.

0:41 Once it's installed, you'll either need to log in or sign up.

0:45 Claude for Excel requires a paid plan.

0:47 It does not work with the free version.

0:50 Right over here, I'll log in.

0:51 And that's it.

0:52 You now have Claude in Excel.

0:54 To open Claude in Excel, you can press the shortcut key Ctrl-Alt-C,

0:58 and here we get a dialog telling us

1:01 that that shortcut is currently being used for multiple features.

1:04 Let's set it to show the Claude sidebar, and then click on OK.

1:08 Over on the right-hand side, let's close Claude.

1:10 Now, alternatively, you can also open up Claude

1:13 by clicking on the icon on the ribbon.

1:15 So here, I'll click there, and that also opens up the side pane.

1:18 If for whatever reason the shortcut key did not work for you,

1:21 you can always reset it.

1:22 Up on the top search field, type in reset office add-in shortcut preferences,

1:27 and then right here, select this option.

1:29 When you click on that, now, when we press Ctrl-Alt-C,

1:32 I get that prompt again, and here I can set it to the Claude sidebar.

1:36 I'll click on OK, and there we see it.

1:38 This opens the sidebar directly inside your spreadsheet, where down below,

1:40 you can now type what you want it to do.

1:43 All right, let's start with this spreadsheet.

1:45 If you want to follow along, you can download this spreadsheet from the link

1:50 right down below in the description.

1:51 Here, I have sales data for the Kevin Cookie Company,

1:54 and you'll notice that it's pretty messy.

1:57 If we look here at the order date column,

1:59 you'll notice that we have inconsistent dates.

2:02 In fact, one of these dates may not even exist.

2:05 Here, we have a missing customer name.

2:07 We have different casing for all the different product names,

2:09 and the same goes for the sales rep.

2:11 And in fact, if I look closely,

2:13 I think there may even be a duplicate order entered here,

2:16 and that can mess up our reporting.

2:19 We definitely don't want that.

2:20 Now, normally, this would take quite a bit

2:22 of time and manual work to clean this up.

2:25 Instead, let's have Claude take a first pass at fixing everything.

2:29 Let's highlight this entire table of data and then

2:31 let's use that shortcut key that we just learned.

2:34 Control, Alt, C, and that opens up the Claude pane over on the right-hand side.

2:38 Now, right down below,

2:39 you'll see that it's going to work on the data that we just selected.

2:42 Right down at the bottom, let's type in the first prompt.

2:45 Clean up this dataset, standardize the date formats,

2:48 remove duplicates, and fix any obvious issues.

2:51 Once you type that in, let's send that through.

2:53 On the right-hand side, you can now see it working through the sheet.

2:57 If it needs to make any changes, it'll ask for your permission first.

3:00 It looks like it identified the first issue.

3:02 Here, it wants to delete range 10, or row number 10.

3:05 Now, if I look at row number 10,

3:07 this is the duplicate entry, so I'm OK removing that.

3:10 Now, the nice thing is Claude asks for permission

3:13 before it makes any destructive changes to the sheet.

3:15 I'll click on Allow.

3:16 All right, let's take a look at what Claude did.

3:19 Claude has now gone through the sheet.

3:21 Over here in the right-hand pane, at the very top,

3:24 we can review all the different issues that it found,

3:26 and if we scroll down, we can also see the changes it made,

3:30 and there were quite a few in this dataset.

3:32 Looking at the sheet now, it's also much cleaner.

3:35 Here, if we look at the dates, they're all consistent.

3:38 The customer names are properly formatted,

3:40 and the same goes for the product names and the sales reps.

3:44 You'll also notice there are comments in the sheet.

3:47 These are changes Claude made but flagged for us to review.

3:50 Earlier, this date was entered as 8/1,

3:53 but based on the sequence of dates, it should be 1/8.

3:57 Claude made that correction and flagged it.

3:59 The same thing happened here.

4:01 What I like about this is instead of guessing silently,

4:04 it flags anything that needs review, like these invalid dates.

4:08 Looking at the customer names, this one is still blank,

4:11 which makes sense since Claude wouldn't know what belongs here,

4:14 so I'll go ahead and fill that in.

4:17 It also flagged this quantity.

4:19 Earlier, this was blank,

4:20 and Claude inferred the value based on the unit price and total.

4:24 Now, if I look at the total column, there's one more nice improvement.

4:28 Originally, these values were hard coded, so if I change the quantity,

4:32 the total wouldn't update, but now Claude has converted this into a formula.

4:35 So, if I change this to, let's say a three,

4:38 you can see the total updates automatically.

4:39 I'll change that back to a two.

4:41 Now that the data is all cleaned up, let's take a look at the formulas.

4:45 At first glance, everything here looks fine.

4:47 The totals are all filled in, and there aren't any obvious errors,

4:51 but just to simulate a real-world scenario,

4:54 let's go ahead and change this formula here.

4:56 Now, you'll notice the total is the quantity by the unit price,

5:00 but let's make a little correction here.

5:02 Let's change that to the unit price times the unit price.

5:05 Now, everything still looks correct at a glance, but the math is actually off.

5:11 Let's have Claude validate everything for us.

5:13 Here again, I'll highlight this entire table, and right up on top,

5:17 let's click on Claude, and down below, we can now type in a prompt.

5:21 I'll say validate the formulas in this sheet, identify any issues, and fix them.

5:26 Let's run that.

5:27 All right, let's take a look.

5:28 Right here, it looks like it caught the issue,

5:31 and then it also made the fix, so everything is now calculated properly.

5:34 And here, if I click into the cell,

5:36 it looks like it's returned it to the quantity times the unit price.

5:40 Now, this is a great example of where everything can look correct at a glance,

5:44 but there may still be some underlying issues.

5:47 Now, we have clean data,

5:49 and we can also trust that the calculations are correct.

5:51 Now that we have clean data and we can trust the calculations,

5:55 let's take this a step further and actually build something on top of it.

6:00 Instead of creating a report from scratch,

6:02 let's have Claude generate a quick summary for us.

6:04 Right up on top, let's open Claude, and down below, I'll type in my prompt.

6:09 Create a summary of total sales by product and region,

6:11 and add a chart to visualize the results.

6:13 Let's run that.

6:14 It created a summary table showing total sales here by product,

6:18 and right down below, we also have it by region.

6:21 And to the right of that, we also have associated charts,

6:25 helping me visualize the data.

6:26 Now, the nice thing is, this is all built directly in Excel.

6:30 You'll notice when I click into the table,

6:32 this is a formula that ties back to the underlying data on sheet number one.

6:35 So, if I modify any of it,

6:37 that'll flow through to the table, and then also the chart.

6:40 Now, personally, I prefer using pivot tables,

6:43 since I think they're more flexible.

6:45 So let's actually see if we could have Claude convert it to a pivot table.

6:49 Back over here, let's follow up by saying,

6:51 can you turn this into a pivot table instead?

6:53 And then let's run that.

6:54 Here, it asks me a follow-up question.

6:56 Let's go with two separate pivot tables for each view.

6:59 Here, it wants to replace the current data

7:01 that's on this sheet with the pivot tables.

7:02 Let's go ahead and allow that.

7:04 And there we go.

7:05 We have one pivot table for product, and one pivot table for region,

7:08 and then we also have associated pivot charts.

7:10 From here, you can keep iterating with Claude

7:13 until you get exactly what you're looking for.

7:15 Now that we have a clean data set and a report,

7:17 let's take this another step further and actually

7:20 understand what's going on in the data.

7:22 Instead of manually digging through the numbers,

7:24 I'm going to have Claude analyze all of this for us.

7:27 Let's open Claude.

7:28 I'll press the shortcut key, and right down below, I'll type in my prompt,

7:31 analyze this data, and highlight the key trends,

7:34 top-performing products, and any notable patterns.

7:37 Let's run that.

7:38 All right, let's take a look.

7:40 I'll scroll up, and right here, it summarizes the key insights from the data,

7:44 including information on the product mix.

7:46 Right down here, we could also see which products sell the most,

7:51 chocolate chip, or the least, Snickerdoodle.

7:53 If we scroll up a little bit, we could also click into the summary,

7:56 and this brings me to the summary sheet.

7:57 And here, it constructed a table with some of the key metrics.

8:01 This way, it helps me quickly understand what's happening in the business.

8:04 And of course, if I want to go deeper,

8:07 right over here, I can continue asking follow-up questions.

8:09 This makes it really easy to go from just all of my raw data to actual insights.

8:15 And this is just scratching the surface.

8:17 Claude also supports things like custom instructions.

8:19 Right up here, I'll click on the three dots,

8:21 and then go into settings, and here, I could type them in.

8:24 For example, if you always want results formatted in a certain way,

8:27 like including a thousand separator, you could set that once.

8:30 Here, I'll go back.

8:31 Now down below, you could also click on this plus icon,

8:34 and you could turn on something called connectors.

8:36 This allows you to pull in external data from sources like FactSet or Moody's,

8:41 and you also have built-in skills.

8:43 Now, these are pre-built workflows that you can trigger with a command.

8:47 For example, earlier, when we cleaned the data,

8:49 you could also use a skill that's already optimized for that task.

8:52 Now, you could add a skill here, or in the prompt field,

8:55 you can simply enter the forward slash, and here,

8:57 you get access to all of your different skills.

8:59 But even just using the simple prompts like we

9:01 did here can help you save a lot of time.

9:04 So, in just a few steps, we cleaned up the data, verified the formulas,

9:08 built a report, and generated insights, all directly inside Excel using Claude.

9:12 If you want to try this for yourself,

9:15 I've included a link in the description and also in the comments.

9:18 Thanks for watching, and I will see you in the next video.