Forecasting: Proactive, Informed and Iterative Decision Making

Preface

Many decisions have to made before and during product development. Many factors contribute to a positive or negative outcome when delivering products making it difficult to know if any single decision is right or wrong early enough to make a difference. Most decisions are made through experience or general group “consensus.” This book proposes ways to use historical data or broad estimates of possible input values to re-enforce those decisions and give them a higher chance of being right more often. Most importantly, this book helps you understand and learn why some outcomes were more likely than others.

Statistics and probability are often used terms and the fields of expertise used to understand how numbers help inform decisions. But this book uses a different term: Forecasting. Forecasting is stating with the information you have “now,” how something in the future might unfold using any form of careful analysis (including data if it is available). The purpose of a forecast is to inform and alter a proposed future action. Contrast this to an estimate, which is an approximation of something that could be measured now.

Although the title of this book is “Forecasting using Data,” my definition of data is broader than most. It includes the good old fashioned guess and estimate at times, but does so knowingly and with due care. All information is data, not all data is informative.

I believe the current techniques for making decisions in product development are flawed for the following reasons –

1. Forecasts are overly simplified and don’t capture the assumptions that were made when the forecast was modeled. Often the forecasts assume complete information about the problem and stability of the delivery system which is almost unheard of.

2. Scant use of historical data to re-enforce and inform decisions. There is a preference to ignore everything in the past, and start estimating anew every time.

3. Preference of pre-defining a planned due date and then shoe-horning whatever into that time. And then being surprised when that doesn’t work (consistently anyway).

4. Striving for perfect predictable outcomes when so much needs to be learnt as the product unfolds and more is understood about the actual problems being solved.

5. Belief that metrics are evil and NO MEASURE is superior to even simple measures. Often belief is that metrics will be used to manage individuals disregarding that will occur even less informed without measures.

These flaws are pretty easy to fix as this book will show. Often it’s easier than the existing techniques used which is ironic given the effort expended to get a poor result. Over the years of consulting on these topics I’ve seen surprised look when teams first encounter ways to forecast that cope with uncertainty. Even more rewarding is seeing the conversations that these forecast generate and the pro-active and innovative ways future problems are avoided. Forecasting isn’t just about the end result, it’s about the ongoing journey of making a decision given new information, increasing the chances (often dramatically) of success.

Next chapter: Chapter 1 — Forecasting