This is a guest article by Theodor on how to Compare Sales of One Product with Another
Ok, now here’s one for you.
Suppose you’d like to come up with a sales report on different products, comparing their evolution on the same period of different years (say Jan ’09 vs. Jan Jan ’10). At the same time, you’d like to keep an eye on their yearly trend (entire 2009 vs. entire 2010).
No big deal, you’ll say, but here’s the twist: the products have not been available for the entire time span taken into consideration. Let’s say you’ve only had Product 1 available for sale for Feb ’09 onwards, while it had been discontinued from October ’10. If you’re really looking for a Like-For-Like (LFL) comparison, you’d only want to compare the months where you have data for both years. It’s false to claim you’ve had a sales boost of 300% when you entered the market with Product X in October 2009, selling 1000 units over 3 months and compare that to the full results of 2010, when you’ve sold 3000 units. In the first scenario you were averaging some 333 units/month, while later you’ve dropped to a mere 250/month. Nothing to brag about there, is it?
Ah, but we also have different product classes. One is aimed for the high-profile buyer (A-Class products), the second for the middle level (B-Class) and so on. Given that different products were added to each class’s portfolio and then later discontinued, we should see the total LFL development of each product class in the same graphical representation.
Hold on another second. One country is defining its quarters as Jan-Mar, Apr-Jun etc, while other might relate a quarterly result to a specific day in the company history (such as the company launch date, or since the new CEO took over or whatever). Wouldn’t it be nice to be able to compare equivalent datasets in any user-defined time span?
So how do you compare sales of one product with another?
Now I’ve always said that the second hardest thing mankind has ever done was to send men on the Moon and safely return them home. That’s only because the MOST difficult thing in the world has become to compare apples with apples. There are so many subtle differences between one dataset and the other (even though they both relate to the same source), that if one reporting template is to have a long life, it should first and foremost come with the built-in ability to allow the end user to drill down through the data and change criteria in order to get relevant results for today’s issue. And all that will change tomorrow, as there will lay a new and unexpected issue on the table.
With that in mind, when I create my templates I follow the self-made golden rule (which later I found many others have applied for themselves long before I knew Excel ever existed) – keep the raw data in one sheet, preferably hidden; use a second sheet for calculation, ALWAYS hidden, and provide a simple and useful graphical interface for the end-user in the third sheet. This will avoid any mishaps such as “Could you please put your formulas back in, I donno which button I pressed and….!!”
Comparing Sales of One Product with Another – Demo:
First see the demo of this technique. Then, we can learn how I created it.
Coming to the attached example – which only works in Excel 2007 and later, by the way:
- Your data is in sheet ‘data’, ordered by product and timeline (Jan-Dec, 2009 and 2010). I’ve created the values using the =randbetween() formula, and then copy-pasted the values only so they will not change anymore.
- To keep things more clear, I’ve placed the calculation formulas in the same sheet as that with the graph, just so you can compare values and figure out formulas more quickly, without having to switch between sheets all the time.
How the Sales Comparison Chart is made?
Now, to bring up values of a particular product, I’ve created a list in C44:C70 (values in column B are just for guidance). We can compare two products, which can be chosen from a couple of drop-down boxes linked to cells B6 and B8. Here’s where the values in column B help: basically, they tell me which item index from the drop-down corresponds to a product. I then placed the same item indexes in data!A7:A46. This is all because I am lazy and I find the sumifs() formula a blessing: all I have to do now is to add up the results that correspond to (1) the chosen Product in the drop-down, which is looked up by the index, and (2) the year, which is in data!E6:E45. [More on INDEX Formula]
An alternative in Excel 2003 would have been to concatenate the value of “Product 1″&”2009” for example, to get a unique identifier and not return the sales value of 2010 by mistake. Then vlookup() after the concatenated value. [Related: How to lookup based on multiple conditions]
These calculations are placed in ‘Yr sls’!F51:Q54. Note there’s an initial IF() there, to only display the values if the respective month is selected. There are two sliders up in the second row, which can help you ‘cut’ your desired portion of the year for comparison.
‘Yr sls’!F61:Q68, using sumifs() again, I added the sales values for each product class. Finally, in ‘Yr sls’!F45:Q48 are the final calculation, where if an item index lower than 8 (corresponding to Product 1) is selected, the values in F61:Q68 are brought up, else the values in F51:Q54.
So now we see our resulting values above the chart, in cells F6:Q9. The deviation is calculated in F5:Q5. But for the yearly totals, I only want to compare apples with apples, i.e. months in which sales have been recorded in both years. For that I used cells U6:AF9. The totals in R6:R9 are based on these isnumber() results. This allows you to have the exact deviation between similar months over an user-defined time span.
Ok, time to close. But not before your boss knows the exact portfolio of each product class! Look shortly in data!B6:B45. This is where, using countif(), we have the number of occurrences for each product class. Knowing that product class “A” will be repeated say 3 times, we’ll use this knowledge to look up the third occurrence of “A” and bring up the product next to it. Now take a peak in sheet “Legend”. Knowing we have to lookup for A 1, that’s how I wrote the formula. But also knowing that “A” will be repeated twice for each product (once for 2009, another for 2010) and not wanting to see duplicates in my product list, there’s a very simple solution: just use odd numbers!! This will only bring up every 2nd occurrence of a product. As I said, I like it simple 🙂 I just left the numbers in C5:C15 visible so you don’t have to fish around for them, the rest are simply I the same color as the background. A bit of conditional formatting does the rest.
Of course, before presenting this to any decision maker, you’d hide the rows and columns they’re not supposed to touch and present them with a clean looking table.
Download the Excel Workbook:
Click here to download the workbook with this example. Examine the formulas and chart in “Yr Sls” worksheet to understand how this is weaved together.
[Added by Chandoo]
Thank you Theodor
Thank you so much Theodor for teaching us some valuable techniques on how to compare apples with apples. I am sure our readers will find these ideas very useful.
If you like this post, say thanks to Theodor.
Do you compare & analyze sales data?
I do this all the time. As part of running my small business, every couple of months, I would take up sales data and see if something odd is going on. I make line charts comparing sales of this year with previous year, understanding the overall trend and compare one product with another.
What about you? Do you analyze sales data? What techniques do you use use? Please share using comments.
Learn more from these pages:
If you work a lot with data & do similar work as above, go thru these articles to learn more.
39 Responses to “Make a Quick Thermometer Chart to Compare Targets and Actuals”
You'll probably have some readers insist on bullet charts, which in my experience are no easier to read.
Note on the case where actuals may exceed targets, the target has to be the second series in the chart, not the first, so it appears in front of the actual.
@Jon.. good point. And yes, readers are already saying bullets are the way to go. Atleast @dmgerbino said it on twitter: http://twitter.com/dmgerbino/status/6761754333
But I feel the same as you did. Bullets need orientation to get started and not that easy to construct (here is a tutorial btw... http://chandoo.org/wp/2008/07/21/dashboard-bullet-graphs-excel/ )
When you just have to compare 2 sets of values, a chart like above is good and easy enough.
And yes, thank you for saying that data series order should be correct to show the target on top.
I think bullet charts are a good alternative. I'm not a huge fan of the formatting that you used above where the outline is so thick.
Another option would be to combine a line graph (plan/goal amounts) with the columns (actual) and select the option to remove the line. This leaves just the value (marker), which can be increased in size to leave only a line about the size of the bar. It's an easy and cleaner way to show actual to plan/goal. Does that make sense?
Tony -
I would use columns (or area) for goal, and lines and markers for actual.
What about if you go over the target? The chart doesn't work so well then.
The technique described today is a near bullet chart. As I stated early this morning on Twitter (link: http://bit.ly/4K3yPM ) , I am a fan of Stephen Few's Bullet Graph.
Hubert Urruttia and I started with Charlie Kyd's method, but as Jon Peltier and Chandoo said, they are not easy to contruct. We moved onto prototyping with Fabrice Rimlinger's SPARKLINES FOR EXCEL and now use XLCube's (BonaVista) Micro Chart tool. Both of these tools allow you to create bullet charts just as easy as any Excel chart type.
As far as reading and interpreting them, this chart type has been the easiest for us to present.
There are many chart types. Today's "Make a Quick Thermometer Chart to Compare Targets and Actuals" is fine for a start, but your ultimate goal should be to create Bullet Graphs. AS Stephen Few states in his overview, "The bullet graph was developed to replace the meters and gauges that are often used on dashboards. Its linear and no-frills design provides a rich display of data in a small space, which is essential on a dashboard. Like most meters and gauges, bullet graphs feature a single quantitative measure (for example, year-to-date revenue) along with complementary measures to enrich the meaning of the featured measure. Specifically, bullet graphs support the comparison of the featured measure to one or more related measures (for example, a target or the same measure at some point in the past, such as a year ago) and relate the featured measure to defined quantitative ranges that declare its qualitative state (for example, good, satisfactory, and poor). Its linear design not only gives it a small footprint, but also supports more efficient reading than radial meters."
@dmgerbino
Since @dmgerbino had to bring my name up I guess I should throw in my two cents.
@dmgerbino and I have both implemented Bullet Charts with great success. What is most interesting about this fact is that we have had a harder time implementing Sparklines than Bullet Charts. The reason for this revolves around the simple fact of familiarity. I will explain. People look at a Sparkline and they think it is a really small Line Chart and it is not. People are familiar with Line Charts since they have been around since 1786 when they were created by William Playfair. Bullet Charts on the other hand are different so they almost demand an explanation. Because of this there was a lot of face time that was needed to explain these charts but once people got them they understood the concept. This is similar to when I introduced Cycle Plots http://bit.ly/87ydVG (Thank you @nbrgraphs!) or Horizon Charts http://bit.ly/6PVavj.
Now about the Thermometer Charts… The first thing I want to address is Tony Rose’s statement. I totally agree that the outline on the chart is too think. It might come of as being a whole new series or a new variable. What I have done in instances like this is I have created a Bar Graph and Scatter Plot mixture. Then I have turned off the Data Series on the Scatter Plot and turned on the Horrizontal Error Bars on the Scatter Plot. The new horizontal line stands for the Plan and the Bar is the actual. The reason why I find this more useful is because this technique works if you have exceeded plan. Actually, I do not understand how Chandoo’s method would display the data if Plan is surpassed.
This reminds me of another blog post that @dmgerbino, @Jon_Peltier, and myself commented on over a year ago. http://bit.ly/PNdO Actually, I talk about similar things in regards to familiarity to charting techniques.
- @hubert_urruttia
[...] we have a post on using thermometer charts to quickly compare actual values with targets. Today we follow up the post with 10 charting ideas you can use to compare actual values with [...]
Hi Chandoo
How do I increase the width of the bar chart and also make the long axis labels come in the same line?
Thank you,
Rajiv
@Rajiv
Select the outer part of the chart "Chart Area" and note the cursor will change to arrows
drag the edges to what ever size you want
You can hold the Alt key as you drag and the chart will snap to the cell boundaries
Now click on the chart area inside the chart "Plot Area" and note that a box with small circles appears around it
drag the circles on the edge of that box to suit
You can hold the Alt key as you drag and the chart will snap to the cell boundaries
@ Hui
Thank you for your comments. But my question was not for the "Plot Area" instead I wanted to know about how should I increase the width of the individual bar charts because with my data all the individual bars are coming to be thin and I want to make them appear broader.
Thank You
@Rajiv
Right click on the Series you want to change and select Format Data Series
Under Series Options goto Gap Width and decrease it to suit
[...] Make a Quick Thermo-meter Chart using Excel [...]
Thank you for the great chart and explanation!
How do I show two amounts (Signed Revenue and Pipeline) as stacked within the Target amount?
@CL... you can use stacked column charts and follow the same technique to get this. See attached file for an example - http://img.chandoo.org/playground/thermo-meter-with-additional-details.xlsx
Chandoo - thanks for the quick response! What if I want the data label for the pipeline to be the actual pipeline value, not the signed rev + pipeline value? i.e. 15 instead of 55
Thanks!
How would i do this in excel 2003?
[...] Thermo-meter charts are very good to show how actual value compares with target (or budget). But how can we add another point for say Last Year value to the chart with out cluttering it. [...]
Hi Guys,
As Matt said,
"What if you if you go over the target?"
Is there a way to make it change color? or at least to show what the target was?
I am planning to use this with a "Forecasted vs Real" production chart but I do not know how to show overproduction.
Any clue?
Thanks
How do I do this if I have 2 bars I want side-by-side? ie 2012 Mean with 2012 benchmark overlapping and then 2013 mean with 2013 benchmark overlapping? I want the 2012 and 2012 mean bars sie by side to compare multiple categories.
Sorry, I meant to say the 2012 and 2013 mean bars side by side
I have a problem in that my PM wants a chart that shows a stacked column (Labor and Expense) and then have the overall buget shown as a thermo.
Everytime I try to do this, I either end up with all three being stacked or all of them being seperated.
Help?
Or if someone knows how to only outline the top and sides of a chart series....then I would have this solved. (Make a stacked column with labor, expenses, and remaining budget, then clear the fill and outline only the top and sides.) I just can't figure out how to do that/ not sure if excel will let me only outline part of a chart series.
[...] Thermometer chart to show budget vs. actual performance [...]
Your home is valueble for me. Thanks!...
I've created the thermometer chart as the Chandoo tutorial described. How do I move my columns closer together? I don't want wider columns; I want to move my narrow columns closer together. Thank you!
Dear Elite members,
could you please let me informed whether we could incorporate color formating in this thermometer approach i.e. if my actual performance is <Min then meter color sud go Red, in between min & target it sud change to Amber & target and above sud change to Green. pls advise. thanks,
I think the only way to do that would be with VBA programming.
@Abhinav
Yes, Simply use a stacked column chart, colored appropriately
Or
You may also want to read about Bullet Charts
@ Hui,
Could you pls demonstrate this with the help of an example.
let's have the below sample data
Actual=12
Min=10
Target=15
Max=20
if Actual>=Min then bar color sud be Red
in between Min & Target= Amber
between target(inclusive) & Max = Green
greater than or equal to Max= Blue
Thanks in advance
Abhi
Great blog post with awesome sample data. I've implemented two of the top "power tips" by changing the colour of the actual values, AND setting Actual to be 40% transparent. Looking good.
[…] easy with these charts. Use them sparingly. As a rule a thermo-meter chart would be better (easy to make, takes less space, scalable) for situations like […]
[…] easy with these charts. Use them sparingly. As a rule a thermo-meter chart would be better (easy to make, takes less space, scalable) for situations like […]
I recently purchased the template bundle and love the ease of use - thank you!
I would like to ask if it is possible to add an important 'block' to the dashboard to illustrate an important status for my executive team; 'billing status'? (ie budget / amount billed) something like that?
Thank you!
@Cheif449.. Thanks for your purchase and kind words.
You can add this easily to the dashboard. Follow below steps.
1. Unprotect the dashboard worksheet.
2. Add a text box (Insert > Drawing Shapes) to the dashboard
3. Put any text inside it as per your need.
4. Format it as needed.
5. Protect the dashboard again.
How do you do this in Excel 2010 - I am not seeing that option in Format data series.
how would we check target and actual sale for multiple years
Select any of the bar, right click and format data series