It’s hard to cover all of these circumstances, so we’ll focus on one specific yet common situation. Let’s say you have a unique list of items, say “fruit”. Each type of fruit has a numerical value associated with it — perhaps the quantity or price of the item. Somewhere else in your spreadsheet you want to enter the fruit item you’re interested in and have a formula return the specific value for this fruit item. Let’s further assume that the items are not sorted in any particular order.

The table below gives you the idea.

In this table, we want to enter the name of the fruit in cell B10, and we want the value in cell C10 to automatically pick out the correct quantity in the source data.

So, now that we’ve set the scene, let’s look at 7 different ways to do this in Excel.

Method 1 – SUMIF

C11: =SUMIF(B4:B7,B11,C4:C7)

SUMIF is a very useful way to do conditional sums. It lets you specify the range of items (i.e. the list of fruit names), the criteria (i.e. the particular fruit you want to find), and the result range (i.e. the quantities associated with the items). It does not require the list to be sorted in any order, but for this to work as a lookup each item in the list needs to be unique. Furthermore, it will only work for numerical values in the result range. If you wanted to return say the colour of the fruit, SUMIF would not work.

Method 2 – INDEX and MATCH

D11: =INDEX(D4:D7,MATCH(B11,B4:B7,FALSE))

INDEX and MATCH is probably the method I use most often to perform exact lookups. For some reason it just makes the most sense to me, and has clear advantages over VLOOKUP and HLOOKUP. If the items in the list are not unique then it returns the associated value for the first one it finds. Although this indicates a problem with your source data, this seems better to me than SUMIF which would effectively aggregate all the items it finds. Furthermore, INDEX and MATCH works with non-numeric data, so would work fine in the situation discussed above regarding picking a colour out of the list. Finally, the ranges do not have to be contiguous, and there is a more advanced version of INDEX and MATCH which can match criteria in two dimensions i.e. rows and columns.

Method 3 – VLOOKUP

E11: =VLOOKUP(B11,B4:E7,4,FALSE)

This is similar to the INDEX and MATCH approach but with one significant disadvantage — it requires you to specify the column number of the data you want returned (in this case the 4th column of the range). It is possible to calculate this column number dynamically, but this is hard, opaque and error-prone. Furthermore, the formula requires that the left-most column of the range contains the items you’re searching. This is unnecessarily restrictive and is not required by the INDEX and MATCH approach.

Method 4 – DGET

F11: =DGET(B3:F7,F3,B10:B11)

This is a neat method and probably should get more use. By all accounts it is a computationally efficient method, and when working with arrays it allows for flexible criteria and boolean logic e.g. “apples” OR “bananas” (e.g. DSUM, DCOUNT, etc.). One drawback is that it requires a criteria range to be set up in a particular way (B10:B11 in the formula) and this may not suit your worksheet layout and may break later if you insert rows in the wrong place. In general, I would avoid this approach and still go with INDEX and MATCH.

Method 5 – Array IF

G11: {=SUM(IF(B4:B7=$B$11,G4:G7))}

Arrays are incredibly powerful and intuitive when you really get them. Sometimes thinking about them can cause your brain to leak out of your ear though, so be careful. In this case, using an array IF is no better than SUMIF and has the disadvantage of being more advanced – so less skilled users of your model may be confused. Also arrays can only be entered by holding down Ctrl and Shft at the same time when hitting Enter. Again, this is not something less skilled users will know about — not that they should be messing with your formulas anyway.

Method 6 – Range Names

H11: =INDIRECT(B11)

This assumes that you’ve created range names for your source data. You can do this easily / automatically by selecting your source data (B4:I7) and hitting Ctrl Shft F3, then choose the Left column option and hit Enter. This will create range names out of all item names. The INDIRECT formula is the way to get Excel to use the input value you type in B11 as if you’d written =Bananas in cell H11. This method is bad design! The main problem is that you now need to maintain and ensure the integrity of a whole lot of range names. If you later change the item Bananas in your source list to be Kiwi Fruit, you’ll need to delete your Bananas range name and add one for Kiwi Fruit. There are so many things that can/will go wrong with this method, and we generally recommend not using range names at all in your models.

Method 7 – PivotTables

I11: =GETPIVOTDATA(“PivotTable”,B13,”FRUIT”,B11)

PivotTables are very useful for analysing data set up in a database format. They can certainly be used to lookup data as we’re doing here, but they require you to create a PivotTable and ensure that it is being refreshed when your source data changes. This does not happen automatically like normal spreadsheet functions. It also requires you to use and understand the GETPIVOTDATA formula, which can be a bit fiddly. In this case, using a PivotTable approach is a bit like using a sledge-hammer to kill an ant, so it’s not our preferred approach.

Conclusion

So, while none of these methods are foolproof, the winner is clearly method 2 – INDEX and MATCH. To further improve your exact lookups, you should 1) check your source items are unique, 2) use data validation to ensure the result item is from this list, 3) deal with potential errors with IFERROR or ISERROR. Finally, although not done in the above examples, remember to get your $ signs in the right places in case you need to copy and paste this formula to other cells.

One question at the end of my talk caught my attention. I had discussed how traditional spreadsheets such as Excel actually build in complexity and opacity, and I gave the example of a formula from a mortgage calculator spreadsheet I searched for and downloaded from the web. This formula calculates the repayments required for each payment period:

=IF(A55=”",”",IF(A55=nper,I54+D55,MIN(I54+D55,IF(C55=C54, E54,IF($D$10=”Acc Bi-Weekly”,ROUND((-PMT(((1+C55/ CP)^(CP/12))-1, (nper-A55+1)*12/26,I54))/2,2),IF($D$10 =”Acc Weekly”, ROUND((-PMT(((1+C55/CP)^ (CP/12))-1, (nper-A55+1)* 12/52,I54))/4,2),ROUND(-PMT(((1+C55/ CP)^(CP/periods_per_year))-1,nper-A55+1,I54),2)))))))

I suspect that this formula gives the right results, but the issue I have with it is “how do you know”? I challenge anyone to dissect this formula and fully understand what it is doing and whether it is correct. Just look at it! It ends with no less than 7 closing round brackets!

Following on from this I ran through Sumwise and how it dramatically simplifies your models and lets you express what would be complex, coded, formulae into easy-to-write, easy-to-read, formulae in your native language. The final question from the floor was a really good one: “what would this formula look like in Sumwise”?

Firstly, the key components of the formula should be broken out into separate columns. I would have a column that calculates the correct effective rate using a lookup table approach. (I will describe how Sumwise deals with lookup tables in a separate post.)

Once this is done, the equivalent formula in Sumwise would be something like:

=PMT([effective rate],[periods remaining],[opening balance])

For those who are interested, basically the PMT function is a replica of the Excel PMT function, which is just a short-cut for the mathematical equation: PxR(1+R)^N/((1+R)^N-1), where P = principal, R = rate, N = number of periods.

• rate of inflation
• number of employees
• revenue
• number of units sold
• salaries and wages

So what is the best way to set up your model to achieve this? The method we prefer is the lookup table approach. The key components of this approach are:

1. Set out all the cost drivers you’ll use in a section at the top of your model. (Rows 5-8 below.)
2. A column that will contain the cost driver for each cost line. (Column E below.)
3. A column containing the base levels of the cost drivers and the associated base levels of each cost line. (Column G below.)
4. Write a formula to calculate the cost forecast based on the driver specified. This is the same formula for every cost line and follows the design in blocks approach discussed earlier. (Cells I11:K16 below.)

Basically what this formula is saying is “find the cost driver level for this period, divide this by the cost driver level for the base period, and multiply this by the base cost level”. The use of INDEX and MATCH is my preferred approach for looking up specific values in another table.

You can download this model and play around with it yourself. Do you have any ideas for improving on this technique?

In Excel you might write the formula =Sum(X31:X37) to add up the items under the Revenue heading. One problem is that this is cryptic and doesn’t mean anything unless you confirm that the range X31:X37 refers to the items you want. And what if the items actually extend to row 38? It is very hard to check and correct this unless you carefully check and review your formulae.

In Sumwise though, there is no A1-style grid. Rather, rows and columns are referred to by user-defined labels such as Revenue, Hardware, or Jan. In Sumwise, the formula =Sum([Revenue.]) would add up all the rows underneath the Revenue row. The formula means what it says and does not need to be translated to be understood or checked.

The usual response to this is “don’t Excel’s range names do the same thing?” The short (and long) answer is NO! In terms of making your spreadsheet model more understandable and reducing potential errors, Excel’s range names actually compound, rather than solve, the problem.

The key issue is that Excel’s range names need to be set up and then constantly checked and maintained to ensure that they are current and accurate. If I have named a whole range of rows Revenue, I need to be sure at all times that this range name references the correct rows. What if I add another row to the end? What if I move things around? What if I change my Revenue section to Expenses? All of these things can impact the definition of the range name. Ultimately, it is just another thing that can go wrong, and another thing that needs to be checked and maintained.

This view is supported by some research presented at EuSpRIG’s 2009 conference which found that the use of range names in Excel models actually reduces one’s ability to find and fix errors in an Excel model.

Of course, there are good uses for range names (global variables is one) and there are ways to get range names to automatically extend and adapt to model changes (dynamic ranges using OFFSET). But, on the whole, and especially in the hands of non-experts, they are pretty dangerous.

Here are some of the takeaways from this session:

• Marketing through social media (Facebook, Twitter, LinkedIn, etc) is the new “word of mouth”. But, social marketing is high risk—high reward as it takes time and effort but can lead to nothing. Alternatively if it works, it can lead to something really big and self sustaining.
• Online marketing done well should consist of the following elements:
  • paid search
  • search optimisation (organic search)
  • online directories
  • social media
  • database marketing
• Why online marketing is better than offline:
  • targeted
  • test lots of things
  • measurable
  • fast
  • value for money
• Some DO’s:
  • target people looking for your product
  • build a database of prospects
  • become an expert in your field
  • stimulate repeat customers
  • track, measure and test
• Some tools to help you get started:
  • Google Keyword Tool, Insight, Trends
  • Doubleclik Ad Planner
  • Alexa
• Your website’s most important attributes:
  • relevancy
  • trust
  • call to action
  • important things ‘above the fold’
• Practical things to achieve this:
  • third party endorsement
  • contact us box
  • testimonials
  • FAQ’s
  • unique and valuable content
• Paid versus organic:
  • paid can get you visitors quickly without much fixed cost
  • organic search takes time and effort (website content, links, etc) with more fixed costs but if you rank highly then low marginal cost, highly engaged traffic. But it requires continual maintenance and attention to stay on top.

Finally, it is worth noting that Ari invited people to this session via social media only. He expected around 10 people to respond but got about 3x this number saying they would like to attend. So, Ari changed the venue to a larger room and increased the catering. But, fewer than 10 people actually attended.

Makes you realise that marketing through social media is not really as social as you think – in that it is not “socially” acceptable to say you’re coming to something and then not turn up. Clicking the “yes I would like to attend” button in Facebook is just a free option, and you should only expect 1/3 of people to actually follow through.

Of course there’s also the fact that people generally value what they pay for. Perhaps Ari should have charged $50 for this session. He might have only received 10 responses, but I am pretty sure 10 people would have turned up. He’d have stuck with his original venue, not over-catered, and earned $500. (Hopefully Ari won’t read this blog and charge me for his next advanced session.)

1. DO NOT have separate sections for Inputs, Calculations and Outputs.

Many people say you should have physically separate sections for Inputs, Calculations and Outputs. While this sounds catchy and pleasantly modular, it leads to bigger problems than it solves. Every part of your model should be easy to understand and review in context. That means when you read a formula you should be able to see the inputs to this formula without having to flip back and forth to other worksheets. Keeping things together and in context does not mean confusing inputs from calculations though. Inputs should still be clearly identified (cell shading, colours, etc.) and where possible, separated from calculations using a blank row or column.

2. DO NOT use named ranges.

Naming cells and ranges using your native language sounds terrific. Surely a formula that says Revenue – Expenses makes more sense than one saying X51-X173? Well, this is all nice in theory, but what if your range named Revenue is pointing to the wrong range? Naming ranges takes time, as does checking whether your range names are correct. Named ranges just add another layer of complexity between the underlying logic and the way this is expressed in your spreadsheet.

3. DO NOT waste time formatting your spreadsheet.

Formatting your spreadsheet with lots of fonts, colours, borders and shading can make you feel like an artist and certainly lets you exercise the right side of your brain for a change. But formatting takes time and effort which can be better spent designing, testing and documenting your model. Also, formatting can sometimes mask errors and obscure the real meaning of your data—from innocuous mistakes like formatting the wrong row in bold, to misleading practices like using a white font on white background to “hide” data. The key part of this recommendation is the word “waste”: do not “waste” time formatting. Formatting should be brief, informative and efficient. Never format cells using ad hoc cell formatting, and always use styles (see here).

4. DO NOT use Excel’s error checking.

Excel provides native background error checking* but it really doesn’t work well. Turning it on and blindly trusting that it’s protecting you against errors would be a big mistake. To test this feature I built a simple model, seeded it with some common errors and turned on Excel’s error checking. It correctly identified 1 out of 7 errors, missed 6 and suggested 5 errors which were not in fact errors at all (click here to download this model). It’s like hiring a security guard to protect your home, then feeling comfortable leaving your front door unlocked and windows open, only to discover a week later that your guard has been at the pub down the road all week.

* Tools>Options>Error Checking in Excel 2003 or File>Options>Formulas>Error Checking in Excel 2010.

5. DO NOT use VBA/macros.

When models are to be used and possibly maintained by others, you should only use VBA/macros when you have no other choice. Most users are not knowledgeable enough to understand and review VBA code, yet most people can work their way around the normal workbook interface. What happens when something goes wrong with the macro? Who will fix it? Who will check it is working properly? Sometimes macros are the only way to achieve the desired result—in these cases the macros should be concise, anticipate changes that may be made to the model over time, and be well documented.

6. DO NOT use multiple worksheets.

Multiple worksheets (i.e. tabs) in a workbook are a great innovation and useful when different model structures need to be combined in one financial model. I still remember the bad old days when the only choice was to have multiple diagonally offset blocks of data in one worksheet to deal with differing model structures. However, like all good things, this feature has become abused. I’ve seen serious financial models with over 50 worksheets, many of which were only used for scratch pad-like calculations, redundant scenarios, and half-built models. Bulky models are confusing, time consuming to check, slow to recalculate, and have a higher chance of containing errors. Only use multiple worksheets where your model structure requires it. Delete redundant worksheets. And try not to replicate worksheets just to model alternate scenarios.

7. DO NOT trust your model.

When all is said and done, you should never blindly trust your model, and you definitely should never unquestioningly trust someone else’s model. Most spreadsheets contain errors (see here). Many errors will be trivial but some might be material. A significant part of your model design and building should be directed towards protecting against errors (both real present errors and possible future errors). Things to consider include: good documentation, data validation, conditional formatting, range testing, high-level sense checks, cross checks and balance checks, and where possible get another person to check your work.

What are your key DON’Ts? We’d love to know.

Say you’ve got a monthly profit and loss statement with many rows of revenue and expense information. Once you’ve got all of your data in a worksheet you now want to analyse the information. Often you’ll want to select a particular line in the P&L and see a chart of the data. Well, you could just create a nice looking chart of that row of data and position it somewhere on your worksheet. Then you could repeat the process for every row of data that you’re interested in. Well this is lot of manual effort, plus it adds clutter to your worksheet. After all, you may only be interested in viewing one chart at a time. (Let’s assume for now that you don’t need to print these all out.)

One method to help reduce the clutter is to use a list box or combo box control on your worksheet and link this to your range of P&L line items. That way you can select the row you are interested in from the list box and see the relevant data displayed in the chart.

There are some problems with this technique though. You need to set up the List Box control and then make sure it continues to refer to the correct data when you change the structure of your model. It also requires that all of your row labels are in the same column – as the List Box control only looks in one column for its data. I like to use multiple columns from my row labels, to give a kind of nesting or hierarchy effect – so this List Box technique does not work all that well for me.

An alternative method is to get Excel to figure out which row you’re in as you move around your model, and then to use that row’s data in the chart. This is quite neat as it means I can just click on a cell to get the chart for that row of data. To do this you need to create a macro – but don’t worry, it’s probably the simplest bit of code you could write – in fact it is only 1 line of code. Here it is:

Private Sub Worksheet_SelectionChange(ByVal Target As Range)
    ActiveWorkbook.Names.Add Name:="xRow", RefersToR1C1:=ActiveCell.Row
End Sub

All you need to do is go to the VBA editor (Alt-F11) find and double click the relevant worksheet in the navigation panel, and copy and paste this code. What it does is create a defined name “xRow” which stores the row number of the currently selected row in your worksheet. Every time the selection changes in your worksheet, this code will run and the row you’re in will be stored in the xRow defined name. You can then write a formula in your worksheet that refers to xRow.

Here’s the end result:

I also used xRow in conditional formatting to highlight in yellow the currently selected row.

One drawback of this approach is that this code to update the xRow defined name is running every time you move in the worksheet. This adds some overhead to your model, and may not be appropriate or desired for really big models.

Once you know how certain keyboard actions behave you’ll actually start to design your spreadsheets in a way that lets you take advantage of them. For example, you may know that holding down Ctrl and then hitting an arrow key on the keyboard will jump to the next “block” in the selected direction. This is demonstrated below. When you hold Ctrl and hit the down arrow key the selection moves to the end of the current block or the start of the next block — a block being a group of non-blank contiguous cells.

This technique can be used to give you a better way to navigate through a complex model. For example, if you have a model with rows in sections, sub-sections and sub-sub-sections, you might want to set up your row labels in various columns, rather than just having these in just one column. That way you can more easily navigate to the next section or sub-section. The example below demonstrates this. In the example on the left we’ve put all the row labels in column A, wheras in the example on the right we’ve split out the structure of our rows across columns A, B and C. You can see that starting at the Income Statement row it takes 6 steps to get to the Balance Sheet row (left), whereas it takes just 1 step to get there if our model is structured as per the right.

This is a very simple example, but if you’ve worked with large complex models before you’ll recognise how this could help you navigate up and down through the rows in a much easier way.

Some of my other favourite keyboard shortcuts (for actually getting things done) include:

• Shift arrows and Ctrl Shift arrows  … to select multiple cells.
• Ctrl [  … to select the dependent cells of the active cell, then tab to toggle between them, and “F5 Enter” to return to where you came from.
• Ctrl ˜  … to toggle between normal view and formula view.
• F4  … to repeat the last action – especially helpful for repetitive formatting.
• F5  … goto and to access the goto special menu.
• Alt F11  … to access the VBA editor.
• Ctrl R and Ctrl D  … to fill right and fill down the active cell – often much better than copy and paste.
• Ctrl . (Ctrl dot) … to toggle to the corners of a selected block of cells – click and drag to select a range of cells and then try this.

This list just scratches the surface, but for me they probably represent the most useful keyboard shortcuts, purely in terms of time-saving and efficiency. If you can master navigating around your worksheet using just the keyboard, you’re well on your way to becoming a power user.

What are your favourite keyboard shortcuts? We’d love to know.

People who work in private equity, venture capital, or any other area where investment time horizons are measured in years rather than days, weeks or months – are generally pretty interested in the IRR of their investments. You’ll often hear a VC say something like “… we need a 35% IRR to justify so and so investment.” There’s plenty of discussion and debate elsewhere about the relevance of IRR as an investment measure, and the benefits of using IRR versus NPV (net present value) – so I’ll avoid getting into this debate here.

The problem with the above rule of thumb is that it doesn’t help you work out the result you’re interested in for other combinations of inputs. This is because the relationship between the variables is exponential rather than linear. You either need a calculator, a ready reckoner, or a large brain. In any case, it is useful to first explore the relationship between IRR (r), years the investment is held (t), and the sale multiple achieved (m).

The mathematics is pretty simple:

m = (1+r)^t

So, if you invest $100 at a 10% compound annual rate of return for 7 years, you’ll get back $195 at the end. That is:

100 * (1+0.1)^7

Which is pretty close to 2x your money − the rule of thumb mentioned above. Alternatively, if you know the multiple you require/achieved and you know the number of years, you can calculate the IRR as follows:

r = m^(1/t)-1

If you’re into spreadsheets you can pretty quickly knock up a table to reflect this. You can do this by either writing the required formula in every cell of the table, or by writing the generic formula once and creating a data table to present all of the scenarios you’re interested in. Either way, the end result may look like this.

IRRs for various Multiple and Years combinations

This table tells you the IRR for various combinations of Multiple and Years.

I’ve seen a table like this on the back of someone’s business card. If you’re in the investments industry, it’s a pretty cool idea. It’s useful – so the person you give your card to might actually bother to keep your card on their desk. It also conveys immediately what you do, and what you’re interested in.

An alternative representation would be to show the Multiple for various combinations of IRR and Years. This table looks like this.

Multiples for various IRR and Years combinations

Here I’ve highlighted in yellow the Multiples that are close to 2.

I achieved this using conditional formatting, so that I am able to play around with the value I am interested in and also the IRR and Years values I want in the table, without having to manually adjust the shading.

If you’re interested, here’s a link to the Excel spreadsheet I created for this.

Circular references are the financial modeling equivalent of the old philosophical causality dilemma, typically stated as “which came first, the chicken or the egg?”

Circular references in spreadsheets can be either a) intended, or b) unintended. If they are the latter, then there is a mistake in your model and the circular reference should be eliminated as soon as possible. The longer it persists the harder it is to find and fix. I could write a whole post on clever techniques you can use to find a rogue circular reference – but that’s not what this post is concerned with.

Today, we’re concerned with the problem of “intended” circular references. These are still problems for a number of reasons. Firstly, they require that you turn iteration on (Tools>Options>Calculation) so that Excel can hopefully hone in on an answer. If iteration is not turned on then Excel will not be able to calculate an answer and will display an error, pop up various windows telling you all about circular references, and even display a toolbar dedicated to helping you fix it.

Excel circular reference dialog

It’s pretty annoying when it happens, but okay, we turn iteration on and our problem is solved. Well, not quite. Leaving iteration on slows everything down. Now your model recalculates multiple times every time a recalculation event occurs. If you’re like me and prefer having your model recalculate automatically, then this verbose recalculation happens frequently. I have seen large models grind to a halt because of intended circular references.

The next, more serious, problem, is that if you ever get a #REF, #VALUE, #NAME, or other similar error in your spreadsheet, this error propagates throughout your model and does not go away even when you fix the (often innocuous) cause. For example you may want to use a defined name in a formula, but if you make a spelling mistake you’ll get #NAME errors throughout your spreadsheet. Even if you fix the spelling mistake, the #NAME errors will remain as they are trapped in the circular reference vortex that you have created. Short of reverting to a saved version without this error in it, the only solution I can think of is to delete the contents of the cells that are involved in the circular reference and then reinstate them. It’s a bizarre solution to a bizarre problem.

Finally, the worst problem is that leaving iteration turned on for intended circular references can sometimes mask unintended circular references. With iteration turned on, Excel is happy again, and won’t pop up annoying dialogs telling you that you have a circular reference. This is terrific until you or someone else mistakenly introduces another circular reference into your model. Now Excel won’t tell you about it, and you may never know you have an error in your model.

So, what to do? Well, you may find that you can eliminate your intended circular references from your model using basic algebra. This is best explained using a simple example. Let’s say your’e modelling subscribers to a service such as mobile phone customers, or subscribers to a DVD home delivery service. The general equation describing your subscriber forecast is: closing subscribers = opening subscribers + additions – losses. Often losses are forecast as an assumed percentage of average subscribers for the period i.e. losses = churn % x (opening subscribers + closing subscribers) / 2.

So here is our circular reference − closing subscribers depends on losses, which depends on average subscribers, which in turn depends on closing subscribers.

However, if we take our two equations:

1. C=O+A-L; and
2. L=P(O+C)/2                 … where P = churn %.

We can rewrite these equations to eliminate closing subscribers (C) and thereby break the circular reference. Our rewritten equation 2 would be:

L=P(2.O+A)/(2+P).

So now we can rewrite our formula in cell B4 to be:

=-B7*(2*B2+B3)/(2+B7).

This gives the correct result but this time without any circular references.

You can see that the formulae to calculate losses now refers only to cells which are not themselves reliant on closing subscribers or average subscribers.

It may not be possible to solve all of your circular reference problems in this way, but the more you can avoid building circular references into your models, the better off you’ll be.

If you’ve got any interesting or twisted circular reference examples please share them with us.