ArticlePDF Available

A DEFT Approach to Trend-Based Foresight



Trend projection is a critical element of most forecasting models. Automatic forecasting methods typically glean trends from past data and extrapolate these trends forward, but do so without consideration of the forces that nurture the trend – and that eventually may turn on it. Adam Gordon’s DEFT framework helps you probe behind the trend to identify the underlying Drivers, Enablers, Friction, and Turners. Copyright International Institute of Forecasters, 2010 FORESIGHT 13
sP e c i a l Fe a t u R e
Trend projection is the bread and butter
of forecasting and the most common basis
of prediction among professional and lay
forecasters. Its intuitive logic powerfully
rests on empirical justication: we can see
something growing or dissolving before
our eyes – in the data, at least – and we can
measure the rate at which it is happening.
e obvious next step is to ask, “Where is
this heading?”
Extrapolation of past data is reassuringly
amenable to quantitative calculations. If
there were 165,000 people living in Las Ve-
gas in 1980, 260,000 in 1990, and 480,000
in 2000, we intuit a trend direction and
rate, and feel halfway to a valid forecast.
It is then a simple step to achieve a pre-
diction by extrapolating that growth rate
to 2020 or 2040. To cover ourselves (“Hey,
nobody can predict the future”), we can
do mathematically more complex things
than mere linear projection, perhaps ap-
plying moving averages or weighting part
of the time series; and because we assume
we’re unlikely to get the rate exactly right,
we cover ourselves with “base,” “high,” and
”low” projections so that “somewhere in
there the future is captured. It feels like
solid analysis.
However, trend projection is “agnostic.” It
does not ask why a trend is moving, which
would allow subjectivity and opinion into
the analysis. It only notices that the data
are moving in a certain direction and fol-
lows this movement into the future. In oth-
er words, it allows us to bypass the causes
of events and behavior that may be tricky
to identify or agree on. We don’t have to
debate open-ended questions such as why
people are choosing to settle in Vegas, what
they are doing there, or what that depends
on, etc. We can just record and verify that
the trend exists and follow it forward.
If forecasting were as simple as extrapolat-
ing in this way, it would be an easy game
indeed. e problem with trend projec-
tion is it assumes the future will be a logi-
cal extension of the past. is, as Paul Sao
(2008) put it, is “looking into the future by
taking a ruler from the past and turning it
a r o u n d .”
To turn the ruler around is to make a huge
assumption: that the same conditions that
underpinned the trend until now will con-
tinue to apply. But if the sad and sorry are-
na of failed predictions tells us anything,
it is that, given one incorrect assumption,
a forecast will be out by miles, not inches.
In 1910, a Bell telephone statistician pre-
dicted a massive ramp-up in switchboard
jobs as telephone use grew, until “every
woman in America” would be employed as
a switchboard operator. is correctly pro-
jected rapidly growing telephone service
demand, but missed automated switching.
A mistake in data recognition, collection,
analysis, or projection will typically result
A DEFT Approach to Trend-Based Foresight
by Adam Gordon
prevIew. Trend projection is a critical element of most fore-
casting models. Automatic forecasting methods typically
glean trends from past data and extrapolate these trends
forward, but do so without consideration of the forces that
nurture the trend – and that eventually may turn on it. Adam
Gordon’s DEFT framework helps you probe behind the trend
to identify the underlying Drivers, Enablers, Friction, and
FORESIGHT Spring 2010
in a small error of forecast. A mistake
in assumption will result in a forecast
of such huge error as to be comic or
Mistakes of awed assumption suggest
that extrapolators see the trend they’re
projecting as a kind of “force” of change
that is taking us forward to the future.
But a trend is no more than a pattern
of change in recorded data, evidenced
by a rise or fall of a measurable variable
when compared at two or more points
over time. A trend is, in other words, the
record of underlying forces at work, the
evidence of those forces but it is not
those forces. And it is these underlying
conditions – the “force eld” sustain-
ing the trend – that determine how the
future will play out: whether the trend
will go forward, stall, reverse, or change
e DEFT framework – Drivers, Enablers,
Friction, Turners – is oered here as a ba-
sis for determining the range and type of
force underpinning a trend.
Drivers: e forces
that cause a trend to
move (that is, cause
the emergence of
a pattern in the data) and sustain it. e
trend to hybrid automobiles is driven by
higher gas prices, environmental concerns,
and automotive technology advancements.
e trend to just-in-time production is
driven by globalization and improvements
in supply-chain-management soware
and communications, and so on.
A driver may itself be a trend. at is, the
globalization trend driving “just-in-time”
production is itself the product of many
driving forces: WTO duty relaxation,
global media, spread of English as a world
language, and so on. In another example,
miniaturization is a driver of many inno-
vations in electronics, but is itself a trend
with antecedent technological drivers. In
this way, we oen have “driver-trend layer-
ing,where a driving force creates a trend
that is the driving force of another trend.
We can therefore understand drivers as
being “nested,” and some drivers to appear
as trends, which are further reducible to
their more essential drivers.
Further, a driver can underpin more than
one trend. e “aging” trend is driven by
advances in medicine, better access to
health care on average, healthier lifestyles,
improved education levels, etc. But im-
proved education also drives trends such
as lower infant mortal-
ity, a rising global middle
class, the empowerment of
women, and so on.
Enablers: Any factor that
promotes, facilitates, or
catalyzes a driver. For ex-
ample, government biomedical laboratory
Key Points
DEFT does for trend projections what SWOT
analysis provides for strategic planning.
It oers a framework for organizing and ana-
lyzing factors that will promote and retard the
success of our endeavors.
The four DEFT components are:
Drivers: forces that create and sustain a trend
Enablers: catalysts that support the Drivers
Friction: resistance that impedes a trend
Turners: events that actively block a trend
Understanding a trend as a reection of
underlying driving and blocking forces
reveals why trends are capable of their
sudden surprises and reversals. Although
actively looking behind the numbers makes
for less immediately quantiable foresight, it
is a necessary price to pay to avoid gross fore-
cast error. FORESIGHT 15
funding would be an enabler of biotech-
nology breakthroughs (which would
drive trends in drug development and
medical practice). Venture-capital back-
ing and other forms of investment are
enablers of small business entrepreneur-
ship, which drives economic growth and
employment. Changes in regulation to
facilitate and desired outcome (foreign
investment for example) is another com-
mon type of enabler.
Enablers are not always easily distin-
guishable from drivers. For example,
intellectual property protection (pat-
ents) enables and facilitates a more rapid
spread of ideas that would otherwise
remain proprietary, which spurs further
research. One might be tempted to say
that “intellectual property is the driver of
research.” But to do this is to overlook an
important distinction and assert a causal
relationship where there is merely a fa-
cilitating one. e distinction between
a causal relationship and a facilitating
or catalyzing relationship becomes im-
portant in working out what sustains a
trend and how so, and therefore what
might happen if and when a sustaining
element disappears or the mix of sus-
taining elements changes. A driver may
work without a catalyzer, but a catalyst
has no eect without a driver.
Friction: e resistance
to change that occurs
naturally and inevitably
in human and other sys-
tems. Socially and cul-
turally, most people are
invested in traditional ways and patterns
that are not easily overcome. We judge
innovation against the way we currently
do things, and any new approach must
overcome educational and nancial ob-
stacles to its adoption. It is oen more
convenient, less risky, and less costly
initially – to do things the old way. is
causes inertia, a “stickiness of the status
quo that provides ongoing, low-grade re-
sistance against change drivers and their
enablers, slowing down a trend.
Other common sources of friction are leg-
acy systems in industry, legacy products
owned by consumers, embedded adminis-
trative procedures and habits, and existing
legislation. ese will stand in the way of
change, not out of ideological, political, or
similar motivational stance, but because it
just takes time for people to alter the way
they do things and the systems they are
imbedded in.
Turners (and Blockers):
ese are forces where there
is intention and agency at
work to oppose a trend,
actively seeking to delay it,
stop it, or turn it back in another direction.
Trend-turning forces – think of them as
counter-drivers” – are enacted when peo-
ple or organizations disapprove of a trend
or it runs counter to their interest.
Trend blocking and turning, therefore,
goes hand in hand with agency and power.
e tobacco industry was able to counter
the trend against smoking for more than
30 years. e medical and social evidence
against tobacco was strong, but any fore-
cast that ignored industry power to block
and shape trend outcomes would have
grossly underestimated how long it took
before smoking would become socially
and legally unacceptable. Business has
many ways of turning trends in industry –
by lobbying, for example, or by buying out
Trend blocking and turning is also a staple
of politics and the judiciary, accomplished
through legislation, and of public-interest
organizations (media campaigns, protest-
petitions, street marches, etc.). Forecasts
that run a trend forward without antici-
pating such countering forces are inevita-
bly wrong on timing and oen wrong on
FORESIGHT Spring 2010
substance too. Real, lasting change in the
direction of a projected trend will only
occur if and when drivers overpower
countering forces.
Culture and values are another key
source of trend turning. In the trend to-
ward cloning humans, for example, the
key question in determining the future
is not whether it can be
done, but whether most
people want to see it done.
If the answer is no, coun-
ter-trend forces will be
certain to emerge to head
o the trend. While these
forces are strong, the trend
is going nowhere, no mat-
ter what techno- bio- or
nano-experts can see in the lab. Obvi-
ously dierent social groups or dierent
societies may hold dierent values, and
in a dierent context the trend could
move rapidly.
We can thus conceive of the future as
what will happen as forces for change
battle forces that inhibit or redirect
change – drivers and enablers versus
friction and turners/blockers. Any mo-
ment in the future will be the net eect of
these forces at that point in time. Good
forecasting will investigate each of the
DEFT categories and provide a sense of
the balance of power for and against the
trend. In a static situation, by denition,
the total force of drivers and enablers is
equal to the countering force of friction
and blockers-turners. Where we have
a steady transition – a trend – change
drivers and enablers are slowly overcom-
ing countering forces. Where drivers
and enablers are strong and blockers are
weak, we can expect the pace of change
to be rapid and, under extreme condi-
tions, exponential. Where blockers and
turners are strong, we should expect them
to deect or nullify the trend, or reverse it
It is worth noting that to scrutinize the
DEFT categories is consciously to restore
the “why, how, who, and what” questions
to trend analysis that modeling typically
overlooks. Why is this trend moving?
What will keep it moving, or how will it be
dissipated and who may stop it? Actively
looking behind the numbers makes for
less immediately quantiable foresight, but
this is a necessary price to pay in avoiding
gross forecast error.
Understanding a trend as a reection of
underlying driving and blocking forces
also clears up another apparent puzzle:
why trends are capable of such surprising
shis and reversals. is is not a puzzle if
we see that a trend has no life of its own.
It is a candy wrapper in the wind: when
the wind reverses, the wrapper reverses;
when the wind stops, the wrapper stops.
It is only as dependable as its underlying
and countercurrent forces. e DEFT view
of trends makes us watch the wind not the
wrapper, and so helps foresight analysts in
the following three vexing situations:
Anticipating Systemic
Interaction and Complexity
A key problem in simple trend projection
is the implication that we can validly sepa-
rate out the variable we are interested in
and roll it forward, without allowing for
the complexity of how other variables af-
fect it and each other, and therefore the
whole emerging picture.
A famous example of extrapolating a
single trend while failing to account for
broader systemic forces was various 1970s
forecasts asserting that the world was run-
ning out of oil. e rising oil-consumption
trend of the day was put against known
reserves, leaving a simple calculation as to
what year oil would run dry. ese fore-
casts did not take into account advances in
Actively looking behind the
numbers makes for less
immediately quantiable
foresight, but this is a neces-
sary price to pay in avoiding
gross forecast error. FORESIGHT 17
computing, geophysics, materials science,
and engineering that favored improved
discovery and recovery of oil, as well as
more ecient rening processes, all of
which changed the trend line.
e DEFT perspective helps us unmask
interactive driving and enabling eects
such as these, allowing us to see unexpect-
ed consequences. It also primes the ana-
lyst more broadly for a “systems view” of
change, one that factors in how variables
are linked in ways that are self-reinforcing
or self-limiting. If a driver systemically re-
inforces itself – for example, wage pressure
leading to ination, leading to yet more
wage pressure that driver becomes pro-
gressively stronger, creating greater and
faster change than a simple trend projec-
tion would allow. Self-limiting systems, on
the other hand, exist when variables are
linked in a way that is intrinsically cor-
rective of change. Systemic action against
change will stall the trend or lead to side ef-
fects or “blowback” situations, all of which
will be invisible to the trend extrapolator.
Seeing Trend Breaks and
Inection Points
Nassim Taleb (2008) memorably encapsu-
lates the demerits of extrapolating trends
in e Turkey Problem. Imagine you’re
a turkey. Every day a nice man comes to
feed you. Every day you get bigger. Your
feedings get bigger, too. If you extrapolate
these trends, you will condently predict
your own continued growth, happily en-
joying ever-larger meals.
But your future is anksgiving.
is is hard reality for those who predict
the future by extrapolating trends. Even if
our turkey has excellent data points, care-
fully observed and diligently recorded,
and even if our turkey is mathematically
sophisticated, applying all the latest mod-
eling techniques from moving averages to
compound regression — he is still going to
be wrong about
his future.
Dead wrong.
All the data
analysis in the
world, all that
fancy comput-
er soware, all
that pricey con-
sulting time bought
and paid for, and he’s
not just slightly under or
over in his projection. He’s
plucked, stued, and roasted.
e lesson is there is oen some-
thing outside a trend a framing condi-
tion that is set to cause a break, inection
point, or “discontinuity. (But, vexingly,
there may not be.) A future-determining
framing condition is not easy to see, but it
will be always be invisible to trend-based
extrapolation. A DEFT-based analysis at
least points at the framing and determin-
ing conditions behind a trend, and thus
has a ghting chance of seeing external
Getting a Better Grip on the Pace of
Change and Its “Lumpiness”
Trend tracking gives the illusion of a
dependable rate of change, sometimes ac-
knowledging past accelerations or decel-
erations, but still aggregating them into
an overall steady growth or waning to the
present. is greatly increases the risk
of mistakenly implying a constant pace
of change in the future. But, even where
trends do evolve as expected where in-
teractive eects or trend breaks do not oc-
cur they seldom evolve at the expected
rate. ey speed up and slow down, which
is, of course, a function of the changing
balance between DEFT elements.
More misery for trend projectors lies in
the fact that the forces behind a trend sel-
dom have a linear eect on it. If the DEFT
FORESIGHT Spring 2010
Adam Gordon is the author of
Future Savvy: Identifying Trends to
Make Better Decisions, Manage Un-
certainty, and Prot from Change
(Ed. note: See David Orrell’s review in
the Spring 2009 issue of Foresight).
Adam runs foresight and scenario-
planning workshops and teaches
industry foresight at various promi-
nent business schools and executive-
education courses. See our interview with Adam in this issue’s
“Forecaster in the Field” feature on page 50.
eld remains unstudied, and the analyst
has at best a vague sense that “X leads to Y,
she or he would likely assume that more X
leads to greater Y (i.e., Y coming into being
faster). But linearity of cause and eect ap-
pears only in textbooks, not the real world
of lags and thresholds. Oen the applica-
tion of a force results in “no change” as the
system absorbs and compensates for its
impact, but only up to a threshold or “tip-
ping point,whereupon the trend kicks in
and sometimes runs exponentially, caus-
ing fundamental change or system col-
lapse. Lag and threshold eects are notori-
ously hard to anticipate or model, but lack
of insight into a trend’s sustaining condi-
tions makes it impossible. Roy Batchelor
(2009) provides an interesting discussion
on how threshold eects might be incor-
porated into our forecasting models. See
his article, Forecasting Sharp Changes, in
the Spring 2009 issue of Foresight.
Sometimes change is held back due to one
specic blocker or, as scientists term it, the
“rate-limiting factor. It may take an out-
side event to overcome this factor and so
shake loose an even balance of forces for
and against a trend. New funding sources,
an assassination, a scientic breakthrough
are the kinds of jolts that can release a
trend or reverse a running trend. e im-
plications a surprise event will have on a
trend are invisible unless we consider the
trend in its DEFT components.
Batchelor, R. (2009), Forecasting sharp changes,
Foresight: e International Journal of Applied
Forecasting, Issue 13 (Spring 2009), 7-12.
Sao, P. (2008). Keynote address, Convergence
‘08 Conference, Silicon Valley, California, No-
vember 15-16.
Taleb, N. (2008). e fourth quadrant: A map
of the limits of statistics, Edge Foundation,
Available at
... As a first step, corporate foresight scholars generally advocate conducting an environmental analysis by, for example, directing a search towards environmental areas along the PEST (Political, Environmental, Social and Technological environment) model or its variations (Lesca, 2004;Godet and Durance, 2011). Identified drivers of change are then used to develop scenarios or create alternative pictures of the future (Fink et al., 2005;Gordon, 2010;Lehr et al., 2017). Another goal is to explore and support the development of new business fields (Heger and Rohrbeck, 2011). ...
... • Trend audit, through which the firm identifies drivers of change in the PEST categories and analyses the forces sustaining and inhibiting the change driver in the future (Gordon, 2010), and ...
Full-text available
The deliberate exploration of how the future competitive landscape may evolve is critical to uncovering threats and opportunities for firms that seek to improve their core businesses and advance to a superior position in the markets of the future. While techniques such as business wargaming can effectively support this process, such techniques can benefit from placing competitive considerations within a broader future landscape shaped by geopolitical, social, technological and economic forces. Scenario planning allows for the exploration of interactions across multiple external forces to create a rich set of narratives on how the future may unfold. This paper will discuss the potential of combining scenario planning and business wargaming to enable strategists to anticipate moves and countermoves and foresee their consequences. We use a real-life case study to illustrate how a scenario-planning exercise can guide the crucial stage of selecting relevant future competitors of a firm prior to engaging in a business-war-gaming exercise. We then introduce what we term the prospective competitive strategy process to guide the analysis of potential competitive dynamics, emphasizing the synergies between scenario planning and business wargaming.
... The problem is that the method assumes the future to be a logical continuation of the past [54]. Disruptive developments such as the COVID-19 pandemic, the war in Ukraine, or the current energy crisis render trend analyses difficult [60]. ...
Full-text available
Rural areas in Europe have been undergoing structural change for years, resulting in strong functional differentiation and fragmentation into prosperous and structurally weak areas. Rural areas are as diverse as the challenges they face. Not only various megatrends but also social, economic, technological, ecological, and political trends influence and shape the development of rural areas. In order to understand the problems and causes of the current development of rural areas, this paper provides a differentiated trend analysis for rural areas at the European level. A content analysis of more than 70 EU-funded project reports on rural areas and rural development is carried out. The aim is to identify trends and their drivers over the last decade and to summarize opportunities and challenges for successful rural regeneration. These external and overarching trends, particularly in the context of a sustainable transition, renewable energies, the emergence of new technologies and growing awareness of environmental impacts, offer rural areas more opportunities than ever before. The interplay between external incentives (e.g., European Union policies) and endogenous regional development (e.g., local actors) is crucial for the successful exploitation of these opportunities for rural regeneration.
... The first practice is achieved in part by assembling the data over time to the present time. This provides trend recognition, as well as understanding of the presence or absence of future validity in these trends (Gordon, 2010). Prospecting is also achieved by investigating the systemic forces and feedback loops that structure and limit change in the situation under study (Meadows, 2008), and further by exploring the deeply held beliefs, myths and metaphors (Inayatullah, 1998) that underpin contemporary societal representations. ...
Full-text available
Design thinking is inherently and invariably oriented towards the future in that all design is for products, services or events that will exist in the future, and be used by people in the future. This creates an overlap between the domains of design thinking and strategic foresight. A small but significant literature has grown up in the strategic foresight field as to how design thinking may be used to improve its processes. This paper considers the other side of the relationship: how methods from the strategic foresight field may advance design thinking, improving insight into the needs and preferences of users of tomorrow, including how contextual change may suddenly and fundamentally reshape these. A side- by-side comparison of representative models from each field is presented, and it is shown how these may be assembled together to create foresight-informed design-based innovation.
... The first practice is achieved in part by assembling the data over time to the present time. This provides trend recognition, as well as understanding of the presence or absence of future validity in these trends (Gordon, 2010). Prospecting is also achieved by investigating the systemic forces and feedback loops that structure and limit change in the situation under study (Meadows, 2008), and further by exploring the deeply held beliefs, myths and metaphors (Inayatullah, 1998) that underpin contemporary societal representations. ...
Conference Paper
Design thinking has grown into a considerable field of management theory and practical application, helping organizations renew products, services, or even business models and strategy by way of the processes which designers have traditionally used to create and innovate. Design thinking orients itself towards the future in the sense that all design is for products, services or events that will exist in the future, and be used by people in the future. This has resulted in an "overlap" between design thinking and the domain and concerns of strategic foresight, and a small literature has grown up around how design thinking may be used to enhance future studies. This paper is informed by that literature, but particularly considers the relationship from the other point of view: how methods from the strategic foresight field may sharpen design thinkers' perception of the users of tomorrow they are designing-thinking for, and help design thinkers create a point of view ahead of time as to how changing contextual circumstances may reshape end-users' needs and preferences. The study rests on research into activities of 300 design thinking practitioners, investigating where and how they use future thinking currently, and suggesting how this need may be fulfilled in a richer and more theoretically grounded way.
Full-text available
Scenario planning has steadily grown to become a significant part of business and organisational foresight processes, particularly where planning situations demand approaches beyond traditional forecasting, due to extent of uncertainty variables or length of future time under consideration. However, despite general consensus as to the importance of the scenario approach in general, and rapid growth in both theory and practice in the field, fundamental questions remain over which situations are most tractable to scenario planning and why; and, in the face of uneven success in application, which among an apparent myriad scenario planning approaches best serves different planning situations, or organisations holding different goals. This dissertation makes an intervention into this problem, investigating to what extent scenario planning projects can be separated by underlying project purpose, and, based on original primary case studies and case-­‐based structured interviews, finds that two meta-­‐categories of purpose exist, which are here referred to as " adaptive " and " visionary-­‐advocacy " purposes. It is argued that a purpose-­‐based distinction of scenario modes provides part-­‐explanation of the effective basis, or absence thereof, of scenario work for different situations—a basis which is achieved via congruence of scenario project purpose with (a) underlying organisational planning purpose, and (b) the extent of organisational influence over external conditions, including macro-­‐variables of change, that constrain it. These findings suggest additions to scenario method as currently understood, particularly pre-­‐project analysis (audits) of both an organisation's planning purpose and its external constraint conditions, to ascertain the presence of absence of necessary congruencies, so as to inform adoption of the purpose platform (and allied methodology) more likely to produce successful outcomes in application. iv This dissertation is dedicated to the memory of Professor Colin Firer, Graduate School of Business, University of Cape Town. v ACKNOWLEDGEMENTS
The Spring 2010 issue of Foresight featured Adam Gordon's presentation of "A DEFT Approach to Trend-Based Foresight." DEFT is the Drivers, Enablers, Friction, and Turners that underpin a trend and determine its course and longevity. In this article, Roy Pearson offers his perspectives on implementing the DEFT approach and discusses how DEFT can help augment other takes on trend projections. Copyright International Institute of Forecasters, 2010
Batchelor concludes we have learned yet again just how bad we are at forecasting sharp and hard-to-reverse changes in economic conditions. The linear models that perform well most of the time aren’t cut out for this job. Waiting in the wings are some nonlinear models – of regime switches, catastrophes, and critical network events – that explain how normally stable, complex systems can become unstable, and that might actually provide us with dependable early warnings of bubbles and crashes. Copyright International Institute of Forecasters, 2009
Keynote address, Convergence '08 Conference
  • P Saffo
Saffo, P. (2008). Keynote address, Convergence '08 Conference, Silicon Valley, California, November 15-16.
The fourth quadrant: A map of the limits of statistics, Edge Foundation
  • N Taleb
Taleb, N. (2008). The fourth quadrant: A map of the limits of statistics, Edge Foundation, Available at taleb08/taleb08_index.html