Being approximately right
Decision making in business is 80% gut and 20% fact. Facts and data are important. Business plans, forecasts, ROIs, etc. are all critical parts of decision making. But, breakthroughs, by definition, cannot happen based on facts. If the future is predictable and can be modelled, then we can’t have breakthroughs.
In our pursuit of these factual information, often to present to the management or other decision makers, we tend to miss this simple fact – the output is dependent on the input. Or simply put – Garbage in, Garbage out.
We all know how underestimated the market potential of computers or mobile phones were. I worked for a company in Singapore that was bidding for a mobile licence. We had a great foreign partner, the leading mobile operator then. After going through all the numbers and the business case, our partner decided that the company would not be viable. Today, we have three mobile operators in Singapore and doing reasonably well, thank you.
While heading a product program a few years back, I needed to include a business modeling tool in my sales pack. I looked amongst other units in the company and found that we had an expert in this field. In fact, he had been working on a business modelling tool for a while. I found he had a wonderful excel tool with over 10 sheets, 100s of input parameters and complex calculations. He had used the model in a number of cases and was proud of the level of accuracy in the predicted outputs.
I picked it up and played around with it for a while. And decided to apply it to my own needs. That is when I found that the majority of inputs needed to be estimated. What would be the take up of the service ? How long will consumers use the service, etc. I could make a guess based on other reference services, but it would remain exactly that – a guess !!
So, I had this wonderful tool that churned out some great outputs based on rather shady inputs. Essentially, I was getting accurately wrong outputs. And for my customers to make their decision, these outputs contributed only to the 20% part of the decision making. What they needed were sensitivities – the effect of changes in input assumptions, not necessarily the exact single point solution.
In short, they needed information that was approximately right rather than accurately wrong !
And this might be useful to remember when in an heated conversation with spouse. You don’t want to lose an argument by being accurately wrong 🙂
March 2nd, 2009 at 8:40 pm
In the digital world, especially mobile, there are so many levels of complexities and uncertainties that it’s very hard to be “accurately right” modeling all the variables. Too often, all the data just isn’t available to begin with, so you end up making decisions based on estimates x guesses x hypotheticals… each variable making the answer less accurate. Unfortunately, some managers drill on this stuff far too much because they want to feel good about making a decision they know is risky, and so there’s a spreadsheet to point at if it doesn’t go as planned.
It is better to have use proxy variables to generate approximate and directionally accurate answers. Growing a business is about taking educated risks, and good managers are good at using that gut feeling (which is really just pattern recognition from deep experience and understanding) and using that to make these tough decisions.
Unfortunately, in many workplaces, this is not rewarded. Risk aversion is rewarded instead, and managers that are good at not taking any risks, can stay in power until the ship sinks. They push for the numbers to be exactly right, clearly misunderstanding the complexity involved. Even worse, this spreads to other executives that recognize the best way to keep their position is not to rock the boat too much.
Question back to you: 20% of decision making is fact-based, and 80% is gut — is that the way it should be? and what does that say about how resources should be allocated when confronted with a big decision?
March 3rd, 2009 at 11:35 am
I believe the problem you are referring to is called ‘modeling risk’ – the more the parameters/ inputs, the greater the estimation required and the higher the risk from wrong estimates, which is why, the best models are often simple and yet most people tend to think otherwise – probably as a crutch for not using common sense??
March 4th, 2009 at 11:55 am
Is “approximately right” an oxymoron ?
Notwithstanding your use of marketing speak I beg to disagree with your argument.
First of all the assumption that decisions are 80% gut and 20% fact. Just take a look at Warren Buffet’s letter to shareholders http://www.berkshirehathaway.com/letters/letters.html. I get the impression there is more than 20% fact involved in the decisions Warren makes !
I have a feeling I know the modelling tool you refer to. One of the clear learnings from my time at the company we both worked for was that the variables required to build a model for mobile products and services were massively misunderstood.
As a counter point I’d like to share my time working on radioactive waste monitoring programs in the 70’s. Our work was designed to predict the potential for radioactive waste leakage into the environment about 100 years into the future in a number of different scenarios. The variables included weather (for the next 100 years), including catastrophic events (like earthquakes that could disrupt the geological layers we were modelling). My point is that we developed models that would accept highly variable inputs, and indicate the sensitivity of the model to these inputs.
I believe that the problem you refer to is the lack of thorough analytic techniques (Excel does not support a very thorough set of statistical techniques, try doing multivariate Bayesian analysis in Excel).
Now that I have introduced statistics let me back-down on my introductory point. In fact being “approximately right” is the basis of most forecasting models. The critical thing is to know what degree of approximation you have made to arrive at a particular forecast.
In summary I disagree with your argument but agree with your conclusion ! Perhaps proving that your gut instinct does work after all …
June 29th, 2009 at 12:17 pm
Very interesting read.
I agree with the primary point of the article ‘guts’. But your substantiation is completely based on predictive analysis (I wouldnt blame it on Excel like in the post above, but would agree to his remark on statistical accuracy or the lack of it in predicting trends based on past data). What about the actual market feed back and human factors in the recent past indicating the poor reception of a service or product? Wouldnt it then be inexpedient to still carry on with guts?
April 23rd, 2011 at 1:36 pm
[…] I wrote in a blog post a while back, in business it is all about being approximately right. Given the uncertainty of the parameters on which decisions are to made, one can be assured that […]