» projektura.org
    nature of the business strategy

Strateška uloga ITa
Dizajn organizacije
Dizajn usluga (PSO)
Upravljanje projektima
 

Nathan Myhrvold was hired from one of Microsoft’s most strategic acquisitions; DSR. He was head of MSR for 5+ years and Rick Rashid was his understudy. Nathan is a colleague of Steven Hawkins and known for his futurist, significant thinking. He was significant in setting direction as a company – part of Microsoft’s core inner leadership circle.

April 3, 1999

Introduction

In the last ten years at Microsoft I have participated in innumerable product reviews and internal presentations - both giving them, listening to them and passing judgment on them as a manager. There are a number of pet peeves that I have with the way people do these presentations. After years of grumbling about this with no appreciable effect, I decided to write a memo[1] on the topic - specifically how to avoid some of the pitfalls in giving an effective presentation.

In many cases I speak from experience on both sides of these topics - I’ve committed all these sins myself. Many of the traps which mar an otherwise good presentation are incredibly easy to fall into, so I don’t in any way mean for the advice in this memo to be condescending. There is no shame in falling afoul of these maxims, and I am certainly not trying to single anybody out.

These points are very much personal and correspond to my taste in the sort of presentations and meetings I like to have. I don’t claim that they are perfect for all audiences. In some cases I think that any senior Microsoft manager would agree with what I recommend, but in other places this might not be the case. Maybe your manager will hate this approach - who knows? I can’t know what everybody wants, so I’m going to plunge ahead with my personal views.

Also, it is worth noting that most of the points below are really aimed at internal Microsoft meetings. I view meetings such as this as a collaborative process. It might not always seem that way, but the goal is to communicate the essential features of the project to management and/or coworkers and get valuable feedback. This is quite different than trying to sell somebody something and clinch the big deal, or how you impress people with industry speeches. I have views on this topic too, but they are very different than internal meetings.

I’d like to acknowledge a lot of helpful comments from a lot of people over the years in helping me make better presentations myself, and also in helping me understand their projects by giving great presentations. I can’t list everybody by name, but I certainly am grateful for what I have learned along the way - after all that is the essence of having great meetings - learning from people and having them learn from you.

The Truth Shall Set You Free

The first and most important thing to remember about a presentation or review is to be completely honest. It might seem odd to belabor such an obvious point, or even insulting to suggest that people at Microsoft might be anything other than honest. Don’t be fooled by the apparent simplicity - being forthright is probably the single most important thing to focus on. People fail at this simple thing in a very high percentage of reviews, hurting both the efficiency of the meeting, and in the long run their personal credibility.

It is incredibly easy to fall into the trap of spin doctoring and putting something in the best possible light. There is a natural tension between the goal of the presenter and that of the manager who is listening to it. The presenter has a responsibility to his or her team to do a good job, and this means being positive and upbeat. If they presenter is a marketing person, then their full time job is painting the project in the best possible light. I don’t mean to say that marketing or sales people are bad presenters or are dishonest, but it is only human nature to use our skills in selling products and concepts and showing them off to their best advantage. To a degree this is a good thing in a presentation. However, it is very easy to take it too far, and once you do it becomes hard to recover.

Meanwhile, a guy like me who has to go to a lot of these meetings knows that it is easy for the person presenting to me to get into a spin doctoring mode, so we get good at trying to detect it. It doesn’t take much for the bullshit detector to go off and have me thinking “gee, is that really true?” Sometimes this gets triggered by an outlandish claim; sometimes it is because a remark is made that does not seem to square with other experience or information that I have.

Example - I once was in a meeting where a bright young guy had been given a new and important area. It was pretty screwed up at the time, and he had only been on the area for a week or so. Nevertheless he had a big deck of slides to talk about the great new things we would do in this area. Mostly, this consisted of recounting the previous and rather confused steps taken before this guy was in charge. Since nobody blamed him for this, he should have stood up and talked about principles going forward. Instead, he had slide after slide that tried to paint the old stuff as somehow being on track toward the new strategy. At the same time he made grandiose claims about what plans were ongoing.

This would have been OK if there had been some real substance, but unfortunately there was nothing there. The ideas were not thought through at all - they were just motherhood and apple pie without any details. Yet they were presented as a “plan”. At first I thought that this just exuberance and over stating things a bit. Then I started to think “this guy is trying to fool me into thinking he has this all figured out”. A while later it, after some interchanges where he would not give up with this posturing I start to think “damn, not only is this guy trying to lie to me, he is also stupid enough to think that I am falling for it”.

People who try to spin doctor an internal product review are really only fooling themselves. In every case that I am aware of[2], the audience has checked out long before the speaker realizes that there is a problem. Anybody with experience in these meetings has a good bullshit detector, and you are just not going to fool them. The more you try to paint a happy face on a bad situation, or claim an overly optimistic schedule, the less likely it is that anybody believes you.

Given the general etiquette in group meetings, very few if any people will have the courage to tell you to your face that they don’t believe you. Loss of credibility is a silent disease. People who are in full spin mode won’t catch the subtle signs that their audience thinks they are full of shit, and the audience likes it that way. It is embarrassing to be put on the spot and tell somebody something like this. It creates confrontation and stress, so your audience will shy away from telling you so.

You might think, that Microsoft meetings are frank to the point of being brutal, so this could not occur. This is wrong. Some people are incredibly open about doubting you and saying how, however most are not. Even those people famous for calling a spade a spade are very likely to be holding back to one degree or another - even if they say rude things in the meeting you can be sure that they thought them long before they were said.

In fact, one of the really awkward things you can do to senior management in such a meeting is to ask them to comment directly as to whether they believe somebody. I am often in a situation where I think that somebody is clueless and their plan is full of shit - however I don’t really want to say, “well, in my opinion you are clueless and your plan is full of shit”. Sometimes I have to, or I have to make a decision that amounts to that. However, in most cases I try to do it is a way which avoids being personal - not only is that more comfortable for them, it is for me too. Sometimes this can’t be avoided because people in the meeting are very insistent on pushing the issue. The worst combination is somebody who is insecure enough to both push the limits of credibility and then seek reassurance.

Even if you do mange to fool people in one meeting, you are not going to keep it up. You might get by for a while, but when the product does actually ship late, or the other things do eventually come out, you will be found out. This will have an adverse effect on your personal credibility. One of the most career limiting things you can do is to get a lingering reputation for not being credible.

The key to being truthful and honest is to go out of your way to make sure that any bad news, and any downside risks are clearly articulated and understood by your audience. Be forthright about screw-ups! Remember two clear principles:

Nobody was ever fired at Microsoft for making an honest mistake.

Nobody was ever promoted at Microsoft who didn’t make a few prior mistakes.

We have tough jobs and it is inevitable that we will make some mistakes. Trying to be Mr. or Ms Perfect is not the way to be successful here. Management understands that there are errors which are made. People who have the maturity to admit errors are the people who can learn from them and that is much more important in the long run than being truly error free.

In addition to errors, there are many times when random factor mean that you do not always win. If somebody offers you a chance to bet on a fair coin toss at 2:1 odds, you should take the bet. After all, if the true probability is 50% and the win is greater than the loss, you have a positive expectation value. It is a good bet. However, it is still a bet and half the time you will lose that bet. This is more than compensated by the times when you win, but you cannot look for a perfect record.

Success at Microsoft is not about being perfect every time, it as about winning in the long run, so it is much better to be honest than to try apply spin and revisionist history to attempt to have a perfect record.

The same is true for schedule overruns. It is hard to manage software schedules, and inevitably we will miss some. I am reminded of one of my favorite quotes on the topic -

Anybody who thinks that software engineering is screwed up clearly has not built a house. - Charles Simonyi

We should be honest about the best estimates that we can reasonably make about getting schedules to be accurate. When we think that we will slip, we should be honest with ourselves and with management about what the consequences are. I have seen many people destroy their credibility by holding to the “plan of record”, even when it was obvious to everybody in the room that we were in deep shit. There is a certain nit-picking logic that says, “well, the official plan is X, and since we have not done a detailed, thorough, bottom up analysis of the plan we won’t officially slip”. When they finally get around to the official review, this sort of group will slip far further, and with far worse consequences than those that know how to take a read of the group informally and start to communicate even if they do not have an “official” answer. The hallway gossip mill will have told everybody about the problem long before then anyway, so once again the people trying to stick to the party line are only fooling themselves.

Having said all this, I hasten to add that you should not try not to over do this by being too dour and too pessimistic. Being honest is important, but don’t stand up and piss on your own team and project unless it is warranted. Being honest does not mean that you can’t be an optimist to some extent, and you certainly shouldn’t think that being a “Dr. No” type will enhance your credibility with management. The key is to balance the natural optimism with an honest and fair appraisal of the facts.

Mean What You Say - Say What You Mean

A milder version of honesty is to mean what you say on slides. This is a simple thing, but it is violated time and time again.

Example - many groups will have a “lessons learned” section where they talk about what things have been learned from previous experience. I once saw a presentation which started with a great set of “lessons learned” which clearly laid out the principles which made a product successful. I was really impressed by these insights. Then the next slide came presenting a follow on product which broke every “lesson” that had supposedly been “learned”.

It was clear that nobody really believed the so called “lessons learned” - it was just some damn set of slogans and rationalizations which were created for the purpose of sounding good to management. My respect for the insights (which I still have) was balanced with contempt for the notion that they wouldn’t even read (or take seriously) their own slides. In that case, why should I take them seriously?

Many cases are less extreme than this example, such as:

Example - many slides have irrelevant claims or comments made on them which detract from the central focus. The presentation will say “We are doing project X because of facts Y and Z”. I’ll be sitting in the audience thinking, well, I disagree that Y and Z are factual, each are unrelated, and that isn’t why we are doing the project. This has often happened to me in cases where I love the project! What do you in a case like this - waste time by discussing the irrelevant side issues and red herrings, or let them go and have the noise be perpetuated?

The upshot of this is that you shouldn’t put something on your slides unless you really, truly and clearly mean it, and it is really essential to the points you want to communicate. Many people like to get their own rationalizations for the “real” motive that we are doing a project, or the “real” reason that we will win, when in actual fact they have only fixated on one of several possible reasons. Overemphasizing one reason over others will only incite people in the meeting who believe in a different reason.

The same is true for controversial points. If you know that there is a diversity of opinion about a topic, but that controversy does not matter to your key points, then don’t bring it up. Or, you could bring it up (in the name of honesty) but don’t take sides on it unless it is materially important to your project. Doing so only wastes time and hurts your main goal of getting more important issues on the table.

The converse of meaning what you say is that, everything that you really and truly mean to say should be on the slides. When I go over slides with people early in the process (even my own slides) there are always some which are worded in an awkward way, or seem to be off target. When I ask about them the person says, “what I really mean is…” or “this is about…”. My response is always “then why doesn’t it say that?”

This is usually a disease of formality. When we create a nice looking PowerPoint slide we tend to be on our best behavior and tend to put things in formal and official sounding terms. Sometimes this is tolerable, but when this formality causes you to create a poorly worded slide it is a mistake. It is much better to say something directly, even if it looks odd to see something that plain in print, than to beat around the bush and risk losing your audience.

The Audience Can Read

The single biggest comment for improving the delivery of presentations is to firmly implant in the mind of every presenter the following point.

The audience can read.

This has an immediate corollary.

You don’t need to read aloud every damn word on every damn PowerPoint slide!!!

It is really rather amazing that people fell compelled to recite every word to an audience that has been staring right at the same words for far longer than it takes to read them. Taken on the face of it, it makes no sense whatsoever.

This belies that fact that there is a strong compulsion to do so. I admit that I am drawn to say those words just as much as anybody else. It is stressful to stand up in front of an audience and talk. The PowerPoint slide serves as an outline, and it is very natural to make sure that every important point is covered. Once we have put things on our slides, by God, we’re going to say them, audience be damned!

The result is that the audience is sitting there knowing just what will be said, waiting for the presenter to get around to saying it. This is no fun. It makes the audience impatient and antsy, yet the presenter is usually oblivious to this. The worst is when the presenter reads just what is on the slide. It is a big let down because the audience knows perfectly well how to read the material. A big delay for nothing new. When the presenter reads the material and then supplements it with something new, that helps, but the audience is still bored by hearing a recitation of the obvious.

The point of having written material on a PowerPoint slide is to allow this visual channel to compliment the auditory information, and visa versa. A good presentation uses these two independent channels into the brains of the audience without having complete duplication.

The text on a set of slides should serve as an outline which frames the discussion and lists all of the really important points. Many of these points do not need any further discussion verbally. You can introduce the topic without reading every bullet point, discuss it generally, and then move on. If the audience wants more information, they will ask. When you do want to make verbal comments they should as much as possible supplement what is written. Even if you have to repeat what is said - say it a different way so that it is not simply rote repetition of what the audience has already read.

Another aspect of understanding that the audience can read is to make sure that the text covers all of the important points you are going to make on your slides. Leaving a major point off the slides and bringing it up orally should be avoided. One reason is that it is very easy for the presenter to forget to bring it up.

Even if it is brought up however, it will be inaccessible to people whom either read ahead in the presentation, or who are not present at the meeting. It is very common for somebody who needs to hear your message not being present, or a meeting will drag on so you don’t get to go through your slides.

Some people don’t like the idea of the audience reading ahead in their presentation, and actually admonish the audience not to do so. This sort of presentation fascism is ridiculous and counterproductive. It might decrease the level of control that the presenter has, but at the same time it is pretty strong hint that the presentation is not being given rapidly enough. If your goal is to perform, then look up one of the many amateur theater groups in the Seattle area. If you want to inform your audience, let them read ahead. Better yet, present at the fastest pace you can and they won’t have time to.

It is hard to go too fast because the audience will stop you, but it is easy to go too slowly because for some reason people very shy about saying “hurry up”, but less shy when they ask about something you skip over quickly. This asymmetry means that you are much more danger of going too slowly than too quickly. Invariably, going to slowly occurs during the least interesting and least important part of the presentation. It is very rare that we have a lot of extra time in a product review or other internal meetings. Even if your project has budgeted time perfectly (which is incredibly rare!) it is likely that the meeting before you didn’t, so you’ll start late and won’t have enough time.

In any event, the fact that your audience can read is a tremendous help in getting through material quickly. Obligatory points that are of little interest can be covered in slides which you bounce over very quickly without talking about every point. This let’s you race through the standard stuff. If you race all the way to the end, count yourself lucky - there is no dishonor in finishing early. More that likely you will race up to the point where you and the audience get to engage in some really substantive conversation.

Not only can the audience read, but they can do so before your presentation. The most effective presentations that I have seen are those where the slides can be distributed before the meeting and read through. This allows very rapid progress through the material, so you can focus on the key points that are controversial.

Reading up front is not limited to slides. Big points of controversy are often best presented in email or memos before the meeting. As you probably know, I am a big fan of using memos or email to discuss topics. There are a lot of issues which are really better discussed this way. Many product reviews would be improved by having email or a memo on an important topic precede the review. Normal status is not a good candidate for this, but if there is a single important issue, I’d advise writing it up before the meeting.

It takes more work to type than to talk, so effort put into writing up an issue in email or a review is greater than just talking about it. However there are some advantages, which I believe, more than compensate. You get to present your material in a complete fashion. You might not be able to get enough to time to do this in a review meeting, or you might wind up getting cut off by conflicting points and strong personalities in the meeting. A memo or email lets also you involve more people and do so sooner than you could schedule them to all be in room.

The Forest and the Trees

Suppose that you’ve taken all of the material above to heart and have a painfully honest, concise presentation which makes effective use of text. This already puts your presentation into the top 10%, but a sticky issue remains - what do you present? When you separate out presentation form, what remains is the content, and unfortunately that often leaves something to be desired.

Another way to put this is that you have to understand the goal of your meeting or presentation. Often there is an immediate goal such as getting headcount approval, or getting through the 3 year planning process. However there is always an implicit goal which you must keep in mind while making the presentation - management wants to understand the fundamentals.

We are a company based on having every level of the management chain understand what we do. Bill is not the sort of guy that feels comfortable making decisions about things he does not understand. The flip side of this is that he is very good at learning about things, and he tries to do so at every opportunity. Bill has created an atmosphere in the company where this trait is passed on recursively through the management chain down. We like to know what we’re doing and have a foundation of understanding on which to base our thinking about product and business strategy.

I personally think that it is great to be in a company based on understanding, however there is one important footnote, which is that understanding comes a cost - namely those of us in management need to be informed and educated in product reviews and internal meetings. No matter how much we want to understand, and how good we are at learning, Bill, or the BOOP members, and most managers down the line do not have enough time to immerse themselves in every detail. Product reviews are our opportunity to try to glean the fundamental points we need to understand from the myriad of details which are best left to the people whose full time job is working with those details.

Too many presentations put all of their priority and space on the wrong things - focusing on the trees to the extent that it is impossible to see the forest.

Example - a group that reported in to me had just finished a review with Bill where almost all of the time was spent on a couple of technical issues, mentioned on one slide out of a deck of 30 or so, most of which were not even presented for lack of time. The manager of the group turned to me and said “gee, I’m sorry that we got ratholed on this one little issue and didn’t get to cover our material”. I laughed and said that his apology should be in the other direction - Bill had asked about the only interesting and important thing in the whole presentation. “How come”, I asked, “you wasted time and preparation on a bunch of boring details that nobody but you should care about?”

The goal of a presentation is not to “get through” a bunch of details, or petty facts. The implicit goal is to update the understanding of the people involved and help them reason effectively about the product. So what if a few details get missed - as long as they do not pertain to the key issues, they are not really of interest to anybody except the people in the group who must deal with them.

The kind of details that people tend to focus on are often self congratulatory. Negative details are usually skipped, but it is hard to resist bragging about details that seem to make you look good. Poor presentations will have slides on the details of the focus group that shows how good the product design is, detailed projections out to the last decimal point on things that are really wild ass guesses, exhaustive lists of problems that have already been solved and a host of other specious detail trying to substitute for meaningful content. A good presentation will focus on the key issues that matter and then be prepared to dive into them to the depth necessary to get a real understanding of the issue.

Another form of needless detail is when reams of data are presented without much in the way of a conclusion. It is great to gather lots of data, but unless you can distill some learning about general examples from it, there is not much point in going through it. Conversely, if you do distill the general lessons, then they are what should be presented first and foremost, with only enough supporting data to illustrate them. It is very important to focus on giving the most important, most fundamental issues the highest priority in terms of both your thinking and your presentation.

When I say fundamentals, I do not mean a trivialized, 50,000 foot view of an issue. In any organization with hierarchical management, the more senior the manager, the more things the manger is responsible for. In order to cope with the informational load this entails, one approach is for successive levels of management to treat issues with successively more shallow and trivialized treatments of a topic. Microsoft takes a different approach. Bill tends to focus on issues in complete depth - all the way down to the basic technology if need be. Instead of getting more shallow, successive layers of management at Microsoft have to get more prioritized because we can’t deal with everything at complete depth. Every product has a few key issues that are unique to in and which are the “high order bits” to its success. These are what Bill, and by extension other levels of management below Bill, need to focus on.

Financial Shenanigans

If I had to pick one area which as being the most rife with bad presentation, it would have to be financial data. More sins of omission or commission exist for finance than essentially any other area. Perhaps this is because finance relies so heavily on numbers that people forget the ideas behind them, but whatever the cause, I have more quarrels with financial data than nearly any other aspect of a presentation.

The first issue with financial data is to understand why you need to have in a presentation. Before you thrust a page of fine print numbers in front of your audience, try to decide what they are going to get out of it. There are many kinds of accounting rules in use - GAP, SEC reporting etc. In most cases for internal Microsoft meetings the rules to use are those that illustrate management issues. This principle should guide what you present.

In most cases this means that the standard P&L statement is not the best way to present the data, or at most it is something to be used in summarizing after the real information has been presented in a different way - either by a modified P&L or some other means.

Example - I was doing a budget review a couple years ago and was surprised to see large variances for the people expenses. Upon asking it turned out that there had been a tax ruling which mean that the amount of money we’d budgeted for taxes was incorrect. This totally distorted and obscured what I was really trying to see, which is how individual groups were managing their personnel expenses. When I asked to have the tax effect removed, I was told “but then it won’t be a real P&L”. So what! The purpose of my review was not to see whether Microsoft paid its payroll taxes appropriately since that was not actionable at my level. What I wanted was insight into the management of the various groups that reported to me.

Realism in a P&L is great for some purposes, but all too often a misguided sense of realism on the part of the people preparing the P&L effectively ruins it for the purposes of the meetings. In the example above I wanted to understand how various groups were managing their spending and hiring. This meant that I needed to eliminate the tax effect. This wasn’t enough however - groups which had hired less than their approved headcount were all under budget, but that didn’t necessarily mean that they were doing a good job of controlling expenses for the people that we already on board. So, I had spending per head calculated because that would tend to allow a more direct comparison of groups which were fully staffed versus those that were not. I say “tend to” because there are sometimes some subtle effects that do not scale, however the per head numbers were far better than making me do the division in my head during the meeting. By the time I was done, the figures did not look much like a standard P&L, and some people even sniffed that they were not “real” as a result. However they did give me exactly what I wanted out of a group budget review.

This brings us to a general rule:

Any calculation which the audience is likely to want should already be on the page.

Excel is much better at math than your audience is. If there is a metric or calculation which a reasonable percentage of your audience is going to ask then put it on the page. Don’t make them do the math in their heads during the meeting.

The converse is that whenever you see people in the audience doing math in their heads it means that you failed. I can’t tell you how often I have seen a meeting devolve into a bunch of Microsoft managers trying to calculate things in their heads which should have been perfectly obvious as a question to the people creating the spreadsheet in the first place. Sometimes this is unavoidable, because during the meeting someone will invent a new comparison or metric. When this occurs it is usually something that should be noted so that next time around it is already calculated.

There are whole list of other rules of thumb that follow a similar vein - making the numbers more accessible to the audience.

Don’t put extraordinary items into ordinary line item categories!

This is another pet peeve of mine. People who create P&Ls or other quasi-standardized financial forms tend to imbue a lot of significance in the standard line items. They treat them as if they are holy, and cannot be altered. Invariably some unusual item will come along - and all too often that unusual item is forced into a standard category whether it fits or not. This just drives me crazy, because it is a triumph of the sanctity of a spreadsheet format over common sense. When it comes time to review the page where this sin was committed there will be some huge, unlikely number sitting there in an otherwise ordinary category. People will be puzzled. An explanation will be sought. Eventually the stupid mis-categorization will come to light, and somebody will say “well, we put item X in there”. Not only does this waste time, but it often pollutes the rest of the data because it can be hard to accurately subtract the item during the meeting.

Example - when we started Microsoft Research we had a number of expense items, such as funding external research at Universities. This would sometimes get categorized as “purchased code”, sometimes as a capital investment, and (I swear to god) a set of research funding was put into office supplies! When I compared Research to WPG groups this lead to some funny conclusions, including me wondering, “My god, how many paper clips can Rick Rashid be using?” In each case the people preparing the budget were well meaning, and they had some logic for their decisions. Eventually I demanded that we create some specific line items called (surprisingly enough) “external research funding”. It took lots of insistence on my part because it means changing some damn standard spreadsheet, but the result was a lot more sensible - to me at least.

Many businesses have fixed costs that wind up being amortized across multiple users. This leads to another rule:

Be careful about mixing fixed and variable expenses!

I have seen many examples where somebody starts with some simple assumptions about the fixed and variable expenses. Then they take a bunch of other assumptions about how fast the market will take off, what sort of hiring ramp can we achieve and other parameters and blend everything together into a very complicated set of formulas that result in a P&L.

This sounds nice, but as soon as somebody like me or Bill or others who want to understand the business see this P&L, the first thing that they do is to start trying to mentally reverse engineer the damn thing to recover basic fixed and variable expenses. The complicated modeling is a pointless exercise, unless people understand the basics of how it is put to together. It is criminal to have to calculate this stuff in your head during the meeting when it existed cleanly but is not presented. The more “realistic” (i.e. complicated) the other assumptions are, the harder it is to get to the bottom of this and figure out what is going on. It makes a lot more sense to say up front what the fixed and variable components are and clearly illustrate the basic dynamics of the business. Then (and only then) should you treat the audience to the P&L calculated by the model.

This brings us to another point

Be suspicious of complex models.

The more complicated the model, the more error prone it is - both in the process of creating the spreadsheet and in having the assumptions turn out to be true. I love the idea of using computers to model complex phenomena - for example mathematical models of epidemiology or population dynamics in an ecosystem. As soon as I see something really complicated in our business, I’ll be the first to support such techniques, but so far the fact is that our business does not hinge on complex formulas or models.

We are not going make major decisions on the basis of complex factors in spreadsheets. Most things are really simple. It is rare that “models” need to be more complex than saying we’ll get $X per customer for N customers. When they do get more complicated it is almost always gratuitous complexity - trying to calculate effects which don’t particularly matter, or trying to take into account transient effects that nobody is smart enough to estimate correctly.

Along the lines of simplicity, the most basic advice I can give about financial models is this:

Proofread the numbers!

I don’t really understand why, but most P&Ls that I see have one or more totally obvious problem areas. I am able to zoom on these and ask questions than cause much head scratching and consternation. Here is my secret technique - I look for oddball numbers - in essence I proofread the spreadsheet.

Any time I see a number than seems really out of place, I ask a question. This is particularly true about variances, but it applies to all sorts of numbers. As an example, when I see a percentage over 100%, I ask. My all time record occurred in a presentation recently where a number was calculated at 450,000%. Maybe it is just me, but that doesn’t seem like a normal percentage. It is almost certainly either a mistake in the spreadsheet, or one of the sins mentioned above (such as mis-categorizing an extraordinary item).

This particular case was extreme - usually the bad numbers are more like 450% (which would still be unusual), but in almost every pile of financial data presented to me I can point at one or more cases of obviously crazy numbers. My conclusion is that people don’t proofread their spreadsheets. I should be clear that I am not condemning anybody in particular for this, and it may well be that a lot of proofreading is being done, but I am really surprised at how much seems not to done.

In some cases the best thing to do with a crazy number is to fix it in a systematic way - either by removing a bug in the spreadsheet, or fixing a mis-categorization. The case of the 450,000% turned out to be “true” in the sense that a product shipped something like one copy in the previous period (just before release) and then shipped 45,000 in the next period. I consider this to be something of a bug - it would have been better to mark the growth as not applicable in this case rather than calculate it, even if the calculation is exact.

Some times unusual number is really true, in which case I have another comment.

Be prepared for obvious questions!

Whenever you have a weird number, be prepared to explain it. In my mind, this means that there should be an explicit footnote on the page which explains it. Don’t wait for your audience to find it and ask! That wastes time, is ineffective (because people are probably calculating things in their heads). The answer is frequently a bunch of changes in the data.

I think that every weird or oddball number should be self explanatory - i.e. a footnote or other calculation on the page shows what it is about. On top of this, the people in the meeting should know the answer cold. I never cease to be amazed at the shuffling of paper and head scratching that goes on when I ask a perfectly obvious question about something that stuck out like a sore thumb. All too often the answer is that somebody has to “go back and look at it” or “run some models”, which is not very satisfying.

Closing Remarks

The title of this section might lead to think that this is where I’m going comment on problems with the closing remarks in presentations. Most Microsoft internal presentations do not finish on time, so we never get to the closing remarks, or at least give them short shrift. Alas, I am going to have to do the same here. The points in this memo are not the only advice I have about presentations, but they are about all I have time to write up at the moment.

I’m sorry I didn’t get around to discussing some of the more positive things about how we conduct internal meetings, and what great good ideas I’ve seen that could be used more widely. I have learned a lot from hearing great presenters explain their projects. Although there are undoubtedly some of these tips that I could pass along, I have given them less priority for the same reason that positive comments don’t get a lot of time during product reviews - namely those things that are going well speak for themselves, and those which are not going well need lots of attention.

Pozdrav,

Zagreb,05.12.2005. (i priznajem da neću ovo raditi više od par puta godišnje ...)