BOOK
REVIEW: ‘Super-Forecasting: The Art and Science of Prediction’, by Philip
Tetlock & Dan Gardner
As a
Market Analyst – having worked in a variety of environments from global
businesses to academia, consultancy to the military – forecasting is a crucial
skill; and, according to this book, "forecasting
is a skill that can be cultivated."
In one of those aforementioned, cliché-ridden, environments I have oft heard
such statements as: “We’re here to
forecast the weather, not read the news”; or, even more self-aggrandisingly
amusing: “We are prophets, not historians.”
Whatever; it remains the case that a central tenet of the role of an
Intelligence Analyst is to make assessments about the future.
But,
fascinatingly, the central premise of this superb book, ‘Super-Forecasting: The Art and Science of Prediction’, is that,
actually, those who are often trumpeted as ‘experts’ at such things – from
economists, to political journalists, to the CIA – are actually no better at it
than laymen or, even, “a dart-throwing
chimp.” OK, it’s a lot more nuanced than that, as the author Philip Tetlock
– whose research coined the infamous anecdote about that talented chimp – would
point out, but the point stands.
Super-Forecasting presents the lessons of Tetlock’s
research, particularly since he established the Good Judgement Project (GJP), and invited volunteers to sign up to
forecast the future – around twenty thousand did so and some, the super-forecasters (top 2%), stood out
from the pack: from bird-watchers to engineers, from film-makers to retirees,
from housewives to factory workers. He entered the ‘team’ into a forecasting
tournament run by the Intelligence Advanced Research Projects Activity (IARPA),
reporting to the Director of National Intelligence in the US, which was
intended to improve the forecasting of the intelligence community.
And his
eclectic team did extremely well: “In
year 1, GJP beat the official control group by 60%. In year 2, we beat the
control group by 78%. GJP also beat its university-affiliated competitors…by
hefty margins, from 30% to 70%, and even outperformed professional intelligence
analysts with access to classified data.”[1]
In
Britain this is timely, as I’m not sure if ‘experts’ have ever had it tougher
than they have of late, particularly in the political sphere. At the time of
writing (August 2016), we have recently seen: pollsters call the 2015 General
Election wrong, then call the EU Referendum wrong, alongside politicians and economists
making all sorts of scatter-gun forecasts about the implications of the EU
Referendum – from plunging consumer-confidence, to an immediate recession, to
collapsing stock markets – that have thus far been wrong.[2] “People in this country have had enough of
experts”, said Michael Gove; or, as Allister Heath put it in The Telegraph, “The ‘experts’ were not just a little bit wrong, but horrendously,
hopelessly so…The forecasters are on another planet.” Tetlock had ‘lunch
with the FT’[3]
soon after, and Brexit was one of the subjects: interestingly, super-forecasters called this one wrong.
I, on the other hand, nailed it at 52% Leave, with my own forecasts switching
from Remain to Leave about three weeks out! (Continuous revision of forecasts
is an important factor, according to the book: “When the facts change, I change my mind. What do you do?”)
Evidence of my forecast, from WhatsApp! |
So it’s
in this context that the key lesson from Super-Forecasting
is so powerful: those who are more intelligent than average (but not “geniuses”), more numerate than average,
well- and widely-read, and who deploy a logical methodology, can be super-forecasters and superior to the
‘experts’ and the talking heads on TV. They are “foxes”, with wide and shallow perspectives; rather than “hedgehogs”, with depth and expertise in
a narrow field[4].
In the words of the author:
“What makes them so good is less
what they are than what they do – the hard work of research, the careful
thought and self-criticism, the gathering and synthesizing of other
perspectives, the granular judgements and relentless updating.”[5]
And this
is where we find the chink of inspirational light for industry Challengers and
SMEs too: there is no reason why these businesses cannot be better at this than
their larger competitors, as a source of competitive advantage. Indeed, for a
couple of reasons, it is imperative that they are! Ultimately, the purpose of
forecasting is to identify opportunities that can be capitalised upon, and
threats to be mitigated – both potentially more impactful for smaller companies.
From Peter Schwartz, in ‘The Art of the
Long View’, “…small businesses are
even more vulnerable to the kinds of surprises and uncertainties that often
overwhelm the plans of giants.”
But,
first, what is the current state of play? Forecasting, as a systematic
exercise, is largely the preserve of big organisations. One global company that
I worked for employs a team of economists, as well as Market and Financial
Analysts, informing company-wide decisions with their macro- and micro-economic
analysis and forecasts. The likes of Shell and AT&T are famed for their
horizon scanning and scenario planning efforts. This isn’t exclusively so, and
Scwartz gives the example of Smith & Hawken in the US, using scenario
planning to build a successful gardening tool business from scratch.
Indeed,
research from the British Standards Institution (BSI) and the Business
Continuity Institute (BCI) showed that, for 2016, large companies (at 74%) are
more likely to undertake horizon scanning and long-term trend analysis than
SMEs (58%).[6]
Even that figure of 58% seems high to me, certainly in any sort of systematic
fashion – it’s much more likely to be informal.[7]
Encouragingly, that figure is up from 48% of SMEs in the 2013 iteration of this
survey. On a similar discipline, Zurich found that, since the financial crisis,
53% of SMEs are spending more time on risk management; 35% are doing more
long-term financial planning, and 33% are looking at their business continuity
plans more frequently.[8]
But
there’s something else in this research that gets to the crux of my argument
about the competitive advantage to be realised by Challengers and SMEs: the
BSI/BCI found that fully one third of organisations don’t use the results of
their horizon scanning; but, crucially, SMEs were more likely to do so, as 71%
said they use their trend analysis results, against 66% of large organisations.
In the 2014 iteration of the survey, it asked whether respondents had access to
the final output, finding that only 16% did not in firms with up to 250
employees; but that nearly doubled to an average of 29.6% in firms with more
than 250 staff. It is because owners, directors, and senior managers
(decision-makers) in smaller firms will be involved in the horizon scanning
themselves.
SMEs are
less likely to do it, but more likely to use it when they do, and take greater
advantage of it. My central point is one that I have made before: Challengers
and SMEs can gain competitive advantage from forecasting, because they are more
agile, with shorter lead-times between foreseeing an opportunity (or threat)
and going after it (or evading it).
WHAT TO
DO IN YOUR CHALLENGER BUSINESS
So, how
can Challengers exploit this potential competitive advantage: firstly, by being
better than their large competitors at forecasting; and, secondly, by
capitalising on the results more effectively?
Taking
the guidance from the book, the first thing to do is to identify your potential
super-forecasters and assemble them
in a team, given that Tetlock recommends using teams to enhance accuracy[9] –
remember that those in the team should display some characteristics outlined
previously, and be multi-disciplinary. This group of people could have meetings
dedicated to this, or you could incorporate the whole process into existing
meetings.
Accuracy factors, from HBR article |
Individual traits of super-forecasters, from Credit Suisse article |
Approach in general, from Credit Suisse article |
Then set
your questions. In the book’s ‘Ten
Commandments’, the first is “triage”:
prioritise, focus on questions that you can answer and action. Start with a
handful of questions that are central to the success, or otherwise, of the
business, and the kinds of questions that could drive changes in behaviour.
Perhaps using Porter’s 5 Forces or the PESTLE framework will help to identify
questions; and these may include some of the following:
One of my forecasts, similar to the above, on the GJP |
The Super-forecasting mantra then outlines
how to proceed: Forecast, Measure, Revise, Repeat – an ongoing, incremental
process. Make your forecasts, monitor them and measure their accuracy, revise
your methodology, and then repeat.
In terms
of how to actually make your forecasts, the book is full of practical advice –
from Fermi estimation, to Bayes’ theorem, from Bayesian question clustering to the
use of probabilities, and add in things like Analysis of Competing Hypotheses
(ACH) and the Delphi technique – and requires reading in full to improve
individual forecasting. But, in general, of course you will have selected a
range of people who are inclined to gather data from various sources (sales
data, competitor analysis, market research, macroeconomic data) – get each to
make a forecast on the question, discuss and debate that and come up with a
company-consensus forecast. Enable the group to continue collecting data for
each question, and re-forecast the next time they meet. All the time throughout
this, flag up anything that requires acting upon by the company.
I did
something similar to this with a university – a Challenger in its sector – last
year, where we addressed the question: what will be the size of our market next
year, in three years, and in 2020? Underlying this is the fact that the number
of 18-year-olds in the UK is declining up to 2021, and particularly so for this
university’s geographical market and discipline specialism. We considered this
alongside higher education participation rates, market share of their courses,
and other factors; with participants working in groups to make their forecasts,
providing us with an organisational consensus.
It
forecast a marginal decline in the size of the market. As a result,
practically, it sharpened people’s focus as they realised that, to grow, they
need to take a greater market share of a smaller market; hence everyone is
looking that much harder for opportunities. The university has done very well
since, including new innovations, product launches, and marketing initiatives.
I don’t think any other organisation in that sector has given as much focus to
this, nor made as comprehensive forecasts, and hence it’s a source of
competitive advantage.
And of
course, getting things like this right can matter – the answers to all of the
questions outlined above need to be planned for. Consider the forecast of
Microsoft CEO Steve Bullmer in 2007 that, “There’s
no chance that the iPhone is going to get any significant market share. No
chance.” Before it went on to hold about 42% of the US smartphone market,
and 13% worldwide. A level of complacent forecasting that contributed to
Microsoft being almost nowhere in the smartphone market. What might be the
equivalent in your industry, and will you see it coming?
In
summary, this is an excellent and, importantly, a very practical book for
improving organisational forecasting. Embedding its lessons in your Challenger
business could be a source of competitive advantage, bearing in mind that there
is no reason that Challengers and SMEs cannot be better than their bigger
rivals – that they can become super-forecasters.
In this, the book is a revelation. And, by doing so, the very least you will
have achieved is to instil a mind-set in your staff of critical analysis about
their external environment – a good thing in and of itself.
If you
want to know more, then read the book. If you want to set up a super-forecasting programme in your
organisation – including team selection, training, question-setting and
management – then contact me here. Meanwhile, if you think you’re up to the
task and want to hone your skills, then you, too, can join the Good Judgement Project here.
My most recent GJP forecast |
[1]
Super-Forecasting, pp.17-18
[4]
Originally derived from the work of philosopher Isaiah Berlin.
[5]
Super-Forecasting, p.231
[7] The sample for this survey is
skewed by the fact that it is carried out by business continuity specialists:
two thirds of respondents were in BC roles, where forecasting is important
(“continuity” is the operative word). It’s also skewed by industry, as well
over half of respondents were in financial/insurance, professional services,
IT, public administration, defence or health. Only 6% in manufacturing or
engineering, only 4% in retail.
No comments:
Post a Comment