Forecasting induction on scanty grounds
Sometimes stepping back a bit and taking a broader view, reminds us of the simple, but often forgotten, truths about what were doing and how we might best achieve our goals. And thats exactly the kind of view Nich Guthrie, head of forecasting global marketing for AstraZeneca, gave attendees at eyeforpharmas recent Forecasting Excellence Europe conference in Madrid with his presentation Forecasting Fundamentals: Why forecast?
Guthrie began by taking a step back and extracting a definition of forecasting from the 1949 edition of the Oxford pocket dictionary that his mother gave him as a child. In it, Oxford defined forecasting as: estimate or conjecture beforehand. And conjecture, he says, was defined by Oxford as: induction on scanty grounds a guess.
So, a forecast is a guess, Gutherie says. You look into the future and guess whats going to happen, but no one knows exactly. And as he says, its not really even about the guess, its about the decisions one makes based on the guess.
How many of you have presented a forecast only to have the big guy in the room say, Oh, I dont like the look of that, Im sure we can do better with no view toward the data or assumptions or anything else? Guthrie asks. His view is that your forecast is bad, and his guess is better than yours.
So, why guess? Because those guesses are helpful, he says. The key is making the best guess you can, allowing you to make the best decisions possible.
Guesses for good decisions
Those decisions might be about which of two molecules to develop, which of several options a given molecule might be used for, or which of two or more strategies might be pursued. And all the answers, Guthrie says, lie in the forecast.
Its all about what we think their future value is, how we guess they might be used and what we might get out of them, he reminds us. Making decisions are what all forecasts are about budgets, setting goals, assessing performance, etc.
Ultimately, what makes a forecast helpful or useful is the decision youre trying to make, Guthrie says. So its important to look at that decision before determining what type of forecast you want to do. And the decision youre trying to make probably means different things to different people, he says. So, you must look at who youre talking to, Guthrie advises.
If youre in local marketing, he says, youre going to be interested in short-term volume and sales forecasts and annual budgets. But if youre in operations, youll probably be most interested in both short and long-term volume forecasts and if youre in strategic marketing, youll probably be more concerned with long-term sales forecasts and business cases for incremental investments, Gutherie says.
I keep being asked for THE forecast the definitive one number or forecast, he says. But, again, it depends on the decision youre making. Some of our finance guys want 25 years post launch, but do they really want the same level of detail month by month probably not but it all comes back to the decision youre trying to make.
Keep it simple
In a nutshell, he says, all forecasts are wrong. He urges forecasters to strive to be approximately right, rather than precisely wrong. Stick to the big picture, Guthrie says. And he reminds us of Albert Einsteins suggestion: keep it simple, but no simpler.
Since its about the decision, the forecast has to be something that people believe in it has to have credibility for them to make the decision based on that information, he says. So, be reasonable. Reasonable forecasts tend to become self-fulfilling.
Guthrie urges grounding your forecast as best you can in the current reality, but reminded attendees that theres nothing wrong with aspirational forecasts.
But first, it must be credible and separating forecasts and plans is so important, he says. If you want to set a budget thats different than your forecast, thats not a bad thing. Just do it with knowledge and understanding.
And he says that the status quo almost always wins. By that I mean that trending typically produces a good short-term forecast, Guthrie says. Changes happen, but they tend to take longer than we expect - or are bigger than we anticipate, but take longer to have an effect. In long-term forecasts its those events that disrupt trends that are probably the most important factor.
He stresses, we also must think about the personal and organizational impacts. He recalls a drug that came to the market that was very successful, but didnt meet its ambitious forecasts. These things impact people, Guthrie says.
An ambitious forecast can be as much of a burden as a blessing, he reminds us. And dont forget there are politics involved these are guesses about the future its hard to say whether youre right or wrong so politics will always come in.
Frame of reference
Everyone has a frame of reference, Guthrie says. And usually, theyre limiting and get in the way of peoples thinking, he says, but are important to understand. These are the biases that get in the way when you start talking about real people, he explains.
He recalls a colleague he once had who was thrown off course by a delayed inhaler launch and thereafter had a personal bias against inhaler drugs. In his mind, he says, it became about the drug itself. It was good to know that though when we were doing a forecast, Guthrie explains. You have to remember that everyone has a frame of reference and they bring it with them wherever they go.
And many are hung up on the accuracy of a forecast, he says particularly those in finance.
They always want to know how accurate a forecast is, Guthrie says. Or they think if we do more work, we can find a more accurate forecast as if its out there just waiting to be uncovered.
He believes accuracy can be expected more in short-term forecasts. My US colleagues tell me they can forecast with an accuracy of less than 2%, which they do by means of a weighted absolute percent error, Guthrie says. And they measure people on it, they set targets for their brands and its part of their brands performance measure brand leaders, to get a high performance rating, must be within the bands of tolerance. So, in short-term forecasts , theres a degree of accuracy because over the short-term trending proves to be the best way to deliver a forecast.
But Guthrie says hes not sure accuracy exists as a concept for long-term forecasts.
We use these forecasts to make decisions, so actually our forecasts influence what the future turns out to be, he says. Its hard to go back and compare forecasts with actual events. I was asked recently how accurate we were with a long-term product forecast and I still dont really have an answer. We know actual sales, of course, but Im not sure which forecast to compare to. Is it the one we did when we licensed the product in, the one we did to justify Phase III studies, or the one we did right before launch? A lot happened and a lot changed in that time. Ultimately, its not the forecast that matters, its the decision you make thats important and that changes the future.
Guessing better
Guthrie says we can guess better by, first, looking at the trends in the market, because effectively it is these trends, without anything else happening, that are going to continue. Ground forecasts in reality whats the current trend and whats going on, he says. The status quo nearly always wins.
He and his colleagues, he says, often battle about whether forecasts should be based on market trends or epidemiology. In the US, he says, they tend to look at prescriptions written, while in Europe most forecasters lean toward looking more at prevalence and incidence of disease. Both are important, he says. But they tell you different things give you a different sense. If you go out far enough, prescription trends may predict that everyone eventually ends up on your drug, while epidemiology sets the boundaries of what a market can be.
But if it was all about trends, forecasting would be easy and many of those doing it would be out of a job, Guthrie says. But forecasting is really more about events, he says. Trends are about today carrying into the future; events are about whats going to disrupt that, he explains. If youre sitting in my shoes, looking at products that have yet to launch, the first event you might think about is you own product launch. How is that going to disrupt the trend? The second is your competitors launch and then theres generic entry.
A medical journal study on the effectiveness of a particular class of drugs may be all it takes to vastly shift a product's future market performance, he reminds us. Unexpected shifts in the market are something you definitely have to think about when youre forecasting, Guthrie says.
And remember, he says, the past is not a perfect representation of the future. History repeats itself, but only on the large, Guthrie warns. History is only an approximate prediction of the future. The future is uncertain; dont be afraid to say so. We make a lot of assumptions and its unlikely youll get them all right. Look at ranges and consider alternative assumptions.
Ranges, he says, give more confidence that youve thought about the possibilities. And if you attach probabilities, Guthrie says, you get a sense of how likely the potential outcomes are and you demonstrate how important it is to try to hit the upside predictions.
Forecasting allows us to make decisions take some mitigating actions and think about alternatives, he says. But there is only one future and, in part, it depends on what decision you take. Remember, a good forecast is the one that helps you make the decision.