Order The Persuasive Wizard: How Technical Experts Sell Their Ideas to Non-technical Decision Makers at Amazon.com for a low price of $12.95. However, the publisher has permitted me to offer it to my blog readers, for a limited time, at a special discount. Go to this site, Wizard and enter the code 7PBGMXNC. The book is an excellent gift for anyone who needs to persuade others.
Our present discussion is the requirement to predict research outcomes, how much it will cost and when it will be complete. Most technologists have only a vague notion of how to do this and, subsequently, they do it inadequately, if at all. This series has several parts because I want to keep each entry succinct and blog-readable. If you follow these proven steps, though, you will know how to make good, reliable forecasts. Part IV considers the Fat Cats.
Fat Cats are the absolute hardest part of predicting research because it involves reading people, and people are hard to read sometimes.
In Part I, we created a high-level list of research and technology tasks. In Part II, we prioritized them. In Part III, we assessed the quantity of each one – how many, how long it would take, how many dollars we would have to spend, etc.
Now, the difficult part is that all this data, or a significant part of it, comes from someone else, as it should. If you try to do everything yourself, as many technologists do, disaster is certain. What do you know about how long it takes to open a requisition and hire someone? What do you know about the company’s supply chain? What do you know about capital budgets and allocation policies? What do you know about the suppliers’ and vendors’ inventories and deliveries? You have to ask questions and get answers from other people.
And, as our dear Hamlet would said, “Ah, there’s the rub.”
The estimates you get from other people are spongy, at best. For example, a vendor wants to ensure he can deliver on time. You told him it was “critical.” So he gives you a longer time period knowing he can deliver in a shorter time period and get in your good graces. He adds unwarranted fat to his estimate. Or, another vendor gives you a totally optimistic delivery time, one that no company on earth can meet because she wants your business and she’ll worry about the late delivery after you’ve signed the contract. Both have errors but in the opposite direction. Can you spot these two estimates when they come in?
Another example can be one in your own chain of command. Management wants you to complete the project by a certain date so you make the assumption that “management” will do what it takes, hire people, and provide capital. But, “management” is not a person, it’s a conceptual entity. When reality sets in, you find that you will indeed get the people, perhaps, but the president has delayed his signature for three weeks because of budget concerns. You didn’t think of that. Or, the capital equipment you need has to go through the company cycle. Your idea of getting “fast track” capital processing is mostly wishful thinking.
Another example might be within your own research group. You ask a team member how long it will take her to do her part of the work. She pads her answer by about 100% thinking that you’ll probably cut her estimate in half anyway, so that way she’ll wind up with the amount she really needs. She doesn’t inform you of that, of course.
Now, all these factors get added, and often, multiplied together. So, if you take everyone at their word, you will find that what you expect to be a nine month project, when you massage all the numbers you’ve been given, will take 27 months. No wonder no one can predict research.
So, what do you submit to management? You want to pad your answer by 100% in anticipate they cut your budget in half? Now, we’re up to four years for what, in reality, is a nine-month project.
So, what is the remedy?
First of all, let’s start with the counterproductive notion that a person is supposed to double everything in anticipation of a 50% cut. I know this is done and I see it all the time. Do you? And, that’s the important part. You have to learn to read people. You probe and pry, you question and query, you parry and thrust. You keep them from lunch, you make them stay over. You make them come in early. You sit in their office. You follow them down the hall. You wait outside the toilets. Whatever it takes to get straight talk and cut out all the nonsense. If they will not cut out the fat, then you do it for them. Not later, but right then and there, in front of them. Show them where they are loitering. Show them that you will not play games. You want good data and valid information and you expect them to provide it if they want to keep their jobs.
There is no substitute for looking a person in the eye and talking one on one. But that does not make their answer totally honest. There is also no substitute for comparison to what they have done in the past. There is also no substitute for comparing different estimates and querying as to why the differences. There is also no substitute for making people accountable and rewarding them accordingly; kudos where earned and stingers where needed. It will not make you popular, but technology persuasion is not about popularity. It is about advancing technology. Your popularity and respect will be achieved through keeping jobs and guiding technology. Employees and peers will see this as a greater, much longer lasting, much more desired reward, than fickle popularity.
Learn to read people. I am not an advocate of reading body language, but I am an advocate of past performance, commitment, dedication, and skin-in-the-game. Make those your tools for evaluating performance and predictions.
So, now that you have all your estimates and they are correct, what do you do? Do you pad them, yourself, in anticipation of a management cut? Not by any means.
But, first, let us make sure we understand what is meant by padding and fat. Nothing is certain. Nothing is for sure. Every wise manager knows that a certain amount of leeway is necessary for contingencies. Murphy’s Law is sure to be at work. Things happen. All data have error bars. All statistics have standard deviations. Therefore, you should include realistic, quantified uncertainties in your estimates. That is in no way the same thing as padding. The former is part of good forecasting and adequate data analysis, the latter is playing games.
Inform your own management that your estimate has contingencies, of course, but no budget-games padding. “But,” you say. “I know they will cut my budget. They always do.” Perhaps, but I will show you later how to deal with that. The first part is to build credibility with management that you are on their side. You are working with them. You, too, want the company to succeed. Your estimates are correct, sustainable, quantifiable, and validated.
So, get rid of those Fat Cats and do the job right. If you want to persuade decision makers, then this is the place to start.