Saturday, March 21, 2009

Designers and decision-makers

I understand why Douglas Bowman is upset, but he's ultimately wrong: he makes a common error, that of using zero as an approximation for a very small number.

First, let me avoid misunderstandings: design is important and trained graphic designers tend to do it better than other people; experiments don't solve all problems and sometimes mislead managers; judgment and data complement each other. On to why Mr. Bowman is wrong, using an hypothetical based on one of his examples.

We learn that Google tested 41 different shades of blue for some clickthrough application. Given his writing, he appears to think that the idea is ridiculous; I disagree. Suppose his choice of blue is off by a very small amount; to be precise say that his favorite color leads to one in ten thousand fewer clicks than the one that does best in the experiment. (How finely tuned would his color sense have to be in order to predict a difference of 0.0001 clickability? Without the experiment we'd never know.)

The problem is that a small number in day-to-day terms (one in ten thousand) is not used in day-to-day applications (serving millions of search queries per day). Googling the number of links served per day I get about 200 million searches, each with a few sponsored links. Let's say 5 links per search, for a total of 1 billion links. Even if the average payment to Google for a clickthrough is only 5c, the difference in colors is worth $ \$5,000$ a day or 1.8 million a year. (These numbers are for illustration, but management at Google knows the real ones.)

This hypothetical loss of 1.8 million doesn't seem much compared to Google's total revenue but it is a pure opportunity cost of indulging the arrogance of credentialism (meaning: "as a trained designer I should overrule data"). I don't intend this as an attack on Mr Bowman, because I don't think most designers perceive the problem this way. But this is the business way of looking at the decision.

Ok, but what if he is right about the color choice? That is, what if after running the experiment the color that performs best is the one he had chosen?

Then the experiment will waste some clicks on the other colors and there's the added cost of running it and processing the data. Say it costs $\$100$k to do this. That means that if there is more than a 5.56% chance that Mr. Bowman is wrong by at least 0.0001 clickability, the cost of the experiment will pay itself off in one year.

Using numbers lets management ask Mr. Bowman a more precise question: Can you be 95% sure that the maximum error in color choice translates into fewer than 1 in 10,000 clicks lost?

The main problem here is the same as with most experience-based judgements when they encounter lots of data: they are roughly right and precisely wrong. And, while in each instance the error is very small to be noticed, multiplied across many instances it becomes a measurable opportunity cost.

Wednesday, March 4, 2009

How to deliver great presentations

Most presentations are terrible, and that's a choice made by the presenter.

Some may find it hard to believe that anyone would choose to be a bad presenter, so let me explain what I mean. When I finish a exec-ed session, a MBA class, or a seminar, some audience members ask me for advice on how to be better presenters. I ask them a simple question: how long do you prepare for a one hour presentation? Typically the answer is between fifteen and thirty minutes. My answer is about five hours, if it's a presentation for which I already have all the background material.

Let me repeat: If I am ready to start preparing the presentation, meaning I don't have to learn the material or find the information being presented, just create the means of presenting it, it takes about five hours to prepare a one-hour presentation. It gets worse: if the presentation is about twenty-minutes long, it takes more than three hours to prepare it, and it may take one hour to craft the one-minute elevator pitch for any given topic.

Presentation design is very time-consuming. It takes a lot more work than jotting down an outline and turning it into bullet-point slides. Except for people who make presentations for a living, this work is not a good use of the presenter's time. When I tell those who ask me for advice my five-hour per hour rule, they recoil in terror: they can't take that much time away from their other responsibilities. And that is what I mean by their choice to be bad presenters. It is not irrational for them to do so, but it is a choice that they make nevertheless.

Great presentations are not created in the delivery, they are crafted in the preparation. There are ways in which a bad delivery can destroy a great preparation, but there is no way in which a great delivery can salvage mediocre preparation.

I've done this for a while now and I have settled on a method that works for me. I offer it here, free of charge and of guarantees. It is based on three observations: presentation design is work, it takes time and effort, and its results get better with experience and practice.


There are five phases in a presentation: 1. Knowledge or information gathering; 2. Presentation design; 3. Presentation delivery; 4. Audience interaction; 5. Postmortem analysis. To be clear, the five-hour to one-hour proportion is for phase 2 alone.

Phase 1, knowledge acquisition and information gathering, is all the work that goes into what the presentation is about. For example, a sales presentation requires having the numbers, the analytics, maybe some what-if simulations. A teaching presentation (not the best way to teach, but that is another post) requires knowledge of the subject matter, examples, exercises, questions. These are all prerequisites to design the specifics of the presentation.

In addition to the specific knowledge and information to present, it also helps to know the audience: what is their purpose here, who they are, what they believe, etc. These allow a presentation designer to tailor the knowledge (the sales simulations or the teaching objectives) to the audience level. For example, two presentations on clustering, one to the R User Group, another to MBA students, necessarily emphasize different aspects of clustering.

Phase 2, design of the presentation after its objectives are defined and the knowledge is collected, will be described below.

Phase 3, delivery, is very important. But what makes it successful is its reliance on a good design. Many presenters and academics rely on cheap theatrics, which may be entertaining but can never make up for mediocre preparation. Obviously a painfully shy person will flop miserably and a monotone delivery will put the audience to sleep, but most presenters can learn to address these shortcomings.

Truly bad presenters are those who have no trouble facing an audience and deliver in well-modulated tones the flaccid prose of bullet points, punctuated by the switching of institutionally-templated slides. Because that presenter is effectively saying to its audience: I didn't spend my time to make this presentation a good use of yours.

Phase 4, interaction with the audience, usually Q&A, is important because it allows for clarification, expansion, and debate. Therefore, Q&A should be prepared as well as the delivered presentation. Few presenters prepare for Q&A specifically; most assume, incorrectly, that being knowledgeable is enough. Like preparing for phase 3 makes for smooth presenting, so preparing for phase 4 makes for smooth Q&A.

Phase 5, the postmortem, is important, uncommon, and -- when it happens -- usually wrong. Phase 5 is important to learn what works and what doesn't: the actual information and knowledge transmitted, the examples or clarifications used, the delivery media, the rhetorical devices, the questions and the answers, indeed all elements. This feedback is an essential part of learning to be a better presenter.

After a presentation I like to take a few minutes to write my impression of the experience, including my perceptions of what worked and what didn't. I also talk to attendees who linger afterwards, and sometimes email them later. This creates networking advantages, but that is not my main reason. My reason is that there is always opportunity for improvement.

The wrong way to do a postmortem, alas, is to circulate, immediately after the presentation, a closed-answer questionnaire cribbed from some fast-food customer satisfaction survey. If a questionnaire is circulated it needs to be designed specifically for the desired outcome of the presentation and administered with an appropriate lag. Therefore, ending a refresher course in marketing models by asking executives to rate the "likability" of the instructor in a 7-point scale (which is a cribbed "how friendly was your McFriedBurger server" question) is preposterous. Instead, asking the executives some questions about models, a few weeks after the course, measures the impact of the instructor in the variables of interest: did they learn and did it stick?


The focus of this post is phase 2, presentation design, and how I do it. I repeat: this has worked well for me; my approach is a lot of work; it has taken me several years to get good at it; and, no, I don't know any shortcuts.

When I begin preparing, I first summarize the objectives of the presentation in concise state descriptions. For example, one such objective for a model-building course is "the student can select among binomial logit, multinomial logit, order logit, and rank-order logit as appropriate to the type of dependent variable." This is the presentation equivalent of Management By Objectives: define the outcomes first, then plan the actions. Note that these objectives also make for testable outcomes.

With the list of objectives for the presentation, I plan the structure. This requires ordering the objectives and their supporting materials, allocating time for each, building bridges between them, and setting review points. Many presenters do this. Then they falter: their plan, in outline form, becomes the end product -- instead of being the foundation on which to build the presentation, their outline is plastered on slides, made into bullet points.

The detailed plan is a tool for the presenter, not for the audience. A well-designed presentation doesn't waste audience attention on the structure of the presentation. It may bring up the structure of a rationale, but these two structures (rationale and presentation) are not the same. Also, I've found that pointing out the rationale as we go is less effective than just building it and then, in review, go over the structure of the rationale in detail. Preferably without bullet points.

Regarding tools, outliners were designed for outlines, so why not use them? Two reasons: first, outliners over-emphasize hierarchy; second, it's too tempting to use their entries, usually sentence fragments, as presentation materials -- "why, they export right there to Powerpoint templates, see?"

Extreme hierarchical thinking brought on by outliners is easily ridiculed; instead of the cheap laugh, I just like to point out that Richard Feynman wrote his Lectures On Physics, covering all of Physics, with only three hierarchy levels: chapter, section, and text. If all of Physics needs only three levels, why does each slide on quarterly sales figures for Duluth need five?

I like to outline on a text editor, as it allows for easy insertion, deletion, and movement of topics. Some people use outliners with immense discipline, working within two levels only and never exporting anything, but I don't have that kind of self-control.


Having the plan, do I start Keynote? No. Not by a long shot. A presentation lives and dies mostly by two of its moments: the opener and the closer. These require a lot of thinking in text. Now, in some cases, the opener or the closer are actually other media: pictures, videos, sounds. But I decide to use these media only after I figured out what the message is. And for that I need to think; I do so better in text.

A strong opening statement, a memorable closing statement, clear bridges between topics, and a few key sentences in the presentation (what in media circles are called sound-bites) take a disproportionate fraction of the effort. It is very important to get these right, not just because they are what the audience will remember most, but also because they punctuate the delivery.

These sentences don't just come to me. They are crafted. Most good sentences are: Ted Sorensen's "Ask not what your country can do for you, ask what you can do for your country" (no, JFK didn't write his own speeches; presidents have speechwriters); Winston Churchill "We shall fight" speech (Churchill did write his speeches); Steven Pinker's "I know you believe you understand what you think I said, but I'm afraid what you heard is not what I meant"; and Sir Humphrey Appleby's unintelligible answers in the Yes Minister series. They all share one thing: they were not improvised.

I found useful crafting advice in many sources: The Elements Of Style, by Strunk and White, On Writing Well by William Zinsser, Speak Like Churchill, Stand like Lincoln by James Humes, and Artful Sentences by Virginia Tufte. All are worth rereading once in a while. Being well-read helps, of course, if by well-read one means versed in the works of authors whose wordsmith abilities are singular, as opposed to buying and scanning the fashionable -- and somewhat illiterate -- authors du jour.

Once the key sentences are crafted, there's the rest of the plan to fill in. Here, depending on the audience and the delivery, I decide on how much detail to write down. (When I write a speech for someone else, I write the entire speech, of course.) I'm partial to outline plus key sentences for teaching and to full text for speeches and professional presentations. It doesn't mean I read the text, but it means I invest some time in polishing the words.

In addition to this polish, there are other advantages to the speechwriting approach: I can check for rhetorical pitfalls and use my knowledge of cognitive psychology to minimize communication failure. A good source for those who would rather not get a PhD in decision sciences just for speechwriting is Made To Stick by the Heath brothers.

Knowing some limitations of the human mind helps identify possible problem points in a presentation. Take the case of very large or very small numbers. If I have the number one billion in my presentation, I will come up with some way to make it clear how big it is. Say we're talking about one billion web transactions per week for some transaction system; here's my approach:

"What's the average age here, about thirty? If you made one transaction every second of every day of every year since you were born, you'd only reach one billion transactions two years from now, at around 32. Imagine all those seconds. Now, that is the volume of transactions this system manages, every week: more than the number of seconds since two years before you were born. Per week." (The redundant "per week" is on purpose.)

Another advantage of writing the speech is that it can be edited for clarity and simplified. There are three common types of failed academic presenter: first, those who don't know the subject. Writing the speech ahead would raise their awareness of that fact and they would either learn or cancel the presentation. Second, those who obfuscate on purpose to make the trivial appear deep. Obviously this is a matter of presenter objectives, but a good copy editor -- here preferably not the presenter -- would cut through the obfuscation. Third, those who have new, complex, and difficult material to present and cannot make up good clarifying examples on the spot. Writing the material ahead would allow them to reflect on where sticking points might appear and to prepare alternative explanations or clear examples for these places.


With the plan (and the speech) I now turn to presentation materials. Most presentation materials fall into two categories: chorus for the speech (as when Steve Jobs says "We sold 300,000 iPods" and the screen behind him has the number 300,000) or object of the speech (as when Steve Jobs says "This is the new iPod" and the screen shows a picture of the iPod; note that the object is not the slide: he doesn't say "this slide shows the new iPod"). Presentation materials can be many things, not just slides.

What is not a presentation material: the outline of the talk. Sometimes it is necessary to have agenda milestones in the presentation, though I try to avoid them as much as possible. Instead of the dreaded agenda slide, I prefer a bridge-or-review statement, maybe with a chorus slide restating the main point. The agenda can -- no, should -- be on paper, with contact information for the speaker and other administrative items. Paper, I said: dead tree with cotton or linen.

The materials to prepare for a presentation fall into four classes. First there's the objectives list, the plan, the speech, and the ancillary examples, prepared responses to questions, and backup talking points. These are speaker-eyes-only materials. I have no problem with audiences knowing that I speak from notes, but my notes are for me not for them: they are designed for presenting, not for audience reading -- they include speaker prompts such as "long pause."

Second, there are presenter-support materials, like projected images and video, sound files, props, walk-through handouts, posters. These are materials designed to help the presenter get a point across, not for further elaboration by the audience. The walk-through handouts may share a physical handout with the agenda and other administrative materials and with reference and self-guided study materials, but it is important to keep in mind their use as presenter supports when making them.

Presenter-support materials are there to help the presenter. In contrast, take-away self-study and reference materials, for instance an article or a list of web references for self-guided study, are there to help the audience learn on their own. In my opinion, unless it poses insurmountable logistical problems, all presentations should have these take-away further exploration documents. In some cases mine have been a single sheet of paper: administrative minutiae and speaker information on one side, references for further exploration on the other.

Note that for teaching engagements there are many occasions in which the students are supposed to prepare materials prior to the class; this is a type of take-away material too. Many rookie teachers have unreasonable expectations of how many audience members will read these materials and how much those who do read actually understand. It is best for most presenters to assume a worst-case scenario of zero on both questions and prepare alternatives that do not depend on the expectations. More so on non-degree engagements.

The teachings of Edward Tufte are essential for making materials used for audience reflection and self-guided learning. Tufte goes a little too far in his dislike of the presenter-support materials (and Donald Norman takes him to task here), but reading his books is essential to learn how to present complex technical materials, and how to receive and consume presentations of complex technical materials.

A fourth type of material is required for instruction engagements: testing materials. Whether they are for grading purposes or for self-testing purposes, a teaching engagement must have an assessment tool. This can also be extended to non-teaching engagements, though I've noticed that some audiences are reluctant to do things that feel too close to being in school. Post-Tutelage Scholarly Disorder, possibly.

(At this point it is probably clear why five hours to prepare a presentation that lasts one hour is positively optimistic.)

The materials themselves vary a lot: from business cases to analytics worksheets to multi-media presentations to post-prandial podium perorations. What is important here is that the materials are created to fit the objectives, the plan, and the words, not the opposite.

Many people like Powerpoint because it allows them to create their slides, their handouts, and their notes in a single program. I don't believe either Powerpoint or Keynote do any these three things well enough. I use Keynote to manage the projector, but I make all the materials with other programs. Powerpoint and Keynote may be the Swiss Army knifes of presentations, but professionals don't use SAKs, they use professional-grade specialized tools.

Putting your slides on paper two, three, six, or more per page is not the same as making a handout. A good handout combines some basic administrative information, some presenter-support materials that can use the high-resolution of paper to their advantage, and all take-away materials and references (including self-test materials if applicable). This task is better done with a word processor and a layout program than by Powerpoint.

Rather than focusing on production tools, it is better to begin by choosing the appropriate presenter-support material. Chorus slides are clearly projection materials. High-resolution engineering diagrams are clearly paper materials. With other things, it depends: short quotes can act like chorus for the presenter, and be on slides (put the quote up after saying it, and most definitely don't read it off the projection screen), zooming in and out of pictures can be done in computer screens, but to compare high-resolution images (say two brand logos) color prints are better. Mathematical proofs should either be in the take-away material (i.e. not in the presentation) or -- if really necessary to have them in the presentation -- done by hand on a whiteboard or flipchart pad.

I make my diagrams in Illustrator, my graphs in Stata, R, or Mathematica, my math typesetting in LaTeX, and my text editing in a simple editor (TextWrangler or Pages used just for editing). Handouts are assembled in InDesign and all graphic post-processing for slides is done in Photoshop. I use many pictures but never clipart: professional-quality photos and graphics only. Yes, I bought them for this purpose; professional work costs money. Animations and video are usually created and processed by people who really know how to do it. (Professionals know their boundaries and therefore subcontract when appropriate.) Sequencing of projection materials is done in Keynote, since I use a Mac.

All these materials, on which so many people fixate, are only the supporting cast for a much more significant actor: the speech, the delivery, and -- I'm not immune to it -- some theatre. Namely props, story telling/acting, and physical demonstrations.

Most people ignore the power of physical props and demonstrations. In contrast, Walter Lewin, a famous MIT Physics professor, uses almost no projected images: writing on the board slows his presentation of mathematical Physics down, giving the audience time to absorb it; this he complements with actual physical demonstrations. Lewin links the equations on the board to the physical demonstrations with the mastery his many years of practice afford him. But even so, as he mentions in the video introduction to his course, he rehearses each class twice every time before delivery.

A final remark on materials: I try to be as robust to material failure as possible: the show will go on even if the TSA confiscates my props, if the main character of my exemplar story of heroic management is indicted for fraud that morning, if the computer projection system in the room is stolen, if there are no whiteboards or flipchart pads, if the handouts... actually the handouts are my tool of last resort, so I tend to bring them myself.

As long as I know what I want to say, have my key sentences memorized and well rehearsed, and distribute my handouts, I can get the job done. Maybe not as perfectly as if everything went well, but surely better than the speaker who had a nervous breakdown in front of 500 colleagues at a conference when he/she couldn't get his/her slides off the internet. (Sex hidden to protect the guilty.)

One time I gave a talk to a large audience. I was promised a computer projection system, but the system didn't work. The room was too big to use any props that fit in a carry-on (nobody would see them) and too dim to use handouts, yet I made no significant changes to the speech, just adapted to the situation: all slides were chorus or illustration; you can live without the chorus and you can rely on the audience for illustration. (Instead of showing a picture of a crowded supermarket, I said "imagine being inside BigBoxStore on the friday after Thanksgiving.") The handouts had the references so the audience could explore the ideas on their own later.


The best advice about public speaking I ever received was "the audience is there to listen to you." Which means I must have something that is worth their time -- "must" in the sense that it is my job to make sure of it. It's like eating out: I'm the restaurateur, they're the diners. The objectives, the plan, the speech, that's the prime rib. The slides, the props, the handout, that's the side dishes. That's why so many presentations are bad: all the effort goes on the side dishes. In the end, there's no meat in them; I make sure there's plenty of meat in mine. So if there's an issue with the side dishes the clients are still well fed.

A final point on preparation: always rehearse. Out loud. In the presentation room and with a camera if possible. Often a rehearsal will suggest some changes, from words that don't work well aloud (avoid rhymes; don't over-alliterate), to projection materials that are wrong for the room (small type on medium screen in large room is a common problem), to demonstrations that won't work in the space provided or break safety rules (burning a copy of a Business Week article to emphasize a point requires talking to the fire marshall beforehand).

Some people confuse memorizing a speech with rehearsing it. If you want to memorize a speech, more power to you, but it does not replace a full scale, all-materials rehearsal. I don't memorize speeches, since I can read a teleprompter or my notes discreetly. (President Obama looks like he's watching a tennis game when he reads from the teleprompters.)

So what is the secret? No secret, just hard work, practice, and experience. The books I referenced throughout helped me. They will help you.