A new tool lands on the front page of every trade newsletter in the same week. Half your peers are quietly piloting it; the other half are quietly hoping it goes away. You have no special insight into which camp is right, and that, not the technology itself, is the actual problem to solve.

The quick version

  • Horizon scanning is the habit of looking for early, faint signs of change, "weak signals", before they become obvious (and expensive) trends.
  • Emerging-tech literacy is being able to ask sharp questions about a technology, what it can and can't do, where it's overhyped, without needing to build it yourself.
  • The trap is timing: people overestimate a technology in the short run and underestimate it in the long run. Both errors cost money.
  • The move is to separate watching from betting: track many signals cheaply, but only place a real bet when the evidence, not the noise, justifies it.

The idea in depth

Horizon scanning: looking for weak signals on purpose

Horizon scanning isn't reading the news faster than everyone else. It's a structured practice, formalised in the public sector long before it reached the boardroom. The UK Government Office for Science defines it in its Futures Toolkit as "the systematic collection of insights on emerging trends and weak signals", the small, easy-to-dismiss observations that might grow into something that reshapes how you operate.

The toolkit's working method is deliberately unglamorous: scan across categories so you don't only see what you already worry about. The standard lens is PESTLE, Political, Economic, Social, Technological, Legal and Environmental. A "weak signal" is a data point that doesn't fit your current model of the world yet: a fringe research paper, an odd start-up, a behaviour showing up in a niche community. Most weak signals fade. A few don't.

So the move is: make scanning a routine, not a reaction. Pick three or four PESTLE categories that could genuinely reshape your work, and assign each to a person (or yourself, one category a week). The goal is a short running list of signals, "saw this, here's why it might matter", reviewed monthly. You are not predicting anything yet. You are widening what you'll notice.

Emerging-tech literacy: judging hype with a sober ruler

Scanning surfaces candidates; literacy is how you judge them. Two tools do most of the work here, and they cut in opposite directions, which is exactly why you need both.

The first is the Gartner Hype Cycle, which maps how expectations around a technology tend to move: a technology trigger, a peak of inflated expectations, a trough of disillusionment, a slope of enlightenment, and finally a plateau of productivity (overview; Gartner's own methodology page). The practical use isn't to pinpoint a technology's exact spot, Gartner sells that, and even then it's a judgement. It's a reminder that peak excitement and peak usefulness are different moments, often years apart.

The second is Amara's law, attributed to futurist Roy Amara, longtime president of the Institute for the Future: "We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run" (Roy Amara). Amara himself was reportedly uncomfortable calling it a law, he saw it as an observation, not a rule, and that caution is worth keeping. It's a thinking aid, not a forecast.

flowchart LR
  A("Technology trigger") --> B("Peak of inflated expectations")
  B --> C("Trough of disillusionment")
  C --> D("Slope of enlightenment")
  D --> E("Plateau of productivity")
  B -. "loud here" .-> F(["Press & pitches peak"])
  D -. "useful here" .-> G(["Quiet, durable value"])
The hype cycle's awkward truth: the noise peaks early, the value arrives late. Leaders Loop

Put together, they explain the two ways leaders lose money. Overestimating in the short run looks like buying at the peak, funding a big rollout on a slide deck, before anyone knows what the tool is actually good for. Underestimating in the long run looks like dismissing a technology because the first version was clunky, then scrambling to catch up once it quietly crossed into being useful.

So the move is: before any commitment, ask two questions out loud. If this is overhyped, what would that look like? (Usually: impressive demos, thin production use, vendors talking faster than customers.) And if this is being underrated, what would that look like? (Usually: unglamorous, growing, real users solving real problems off to the side.) Holding both questions stops you anchoring on whichever story reached you first.

Scenarios: planning for more than one future

Scanning and literacy tell you what to watch. Scenario planning tells you what to do when you genuinely can't know which way things break. The canonical example is Pierre Wack's work at Royal Dutch/Shell in the early 1970s, written up in his Harvard Business Review article "Scenarios: Uncharted Waters Ahead" (September 1985). Wack's team didn't try to predict the oil price. They built a small number of plausible, internally coherent stories about how the world might change, including one in which oil-producing nations sharply restricted supply, so that managers had already thought through the shock before it happened.

The key word is plausible, not probable. Wack deliberately steered away from "here's our single best forecast," because a single forecast quietly becomes the only future anyone prepares for. A handful of distinct, believable futures keeps the organisation's options open, and, crucially, changes how decision-makers see the present.

So the move is: for a high-stakes technology bet, write two or three short scenarios, "the tool becomes a commodity everyone has," "it stays a specialist niche," "regulation slams the door", and ask what you'd want to be true in each. The actions that look smart across all your scenarios are your safe early moves. This is the practical handshake between foresight and reversible vs irreversible decisions: scenarios show you which doors stay open and which slam shut.

Where this breaks down, an honest limit

Foresight has a real failure mode, and it's worth naming. None of these tools predicts the future; they discipline how you think about it. The hype cycle is a stylised pattern, not a measured law, plenty of technologies skip the trough or never reach the plateau, and Gartner's placements are expert judgement, not data. Amara's law is a heuristic that can excuse any position ("it's overhyped now but inevitable later" justifies both buying and waiting). And scenario planning done badly becomes a workshop that produces a glossy report nobody reads. The tools earn their keep only when they change a real decision. If a scan or a scenario exercise doesn't end in a different action, watch this, pilot that, stop funding the other, you've done foresight theatre, not foresight.

A worked example

Imagine Priya, who runs a 40-person customer-operations team at a mid-sized insurer. A new AI tool that drafts claim-response letters is everywhere this quarter; her CFO has seen the demo and wants "an AI strategy by Friday." (Figures and names here are illustrative.)

The panic move is to buy seats for all 40 people and announce a transformation. Priya does the opposite. She runs the literacy questions first: the demos are dazzling, but when she asks the vendor for reference customers running it at scale, the list is short and recent, a short-run-overestimate tell. So she treats it as a signal worth a small, reversible bet, not a big one.

She picks three people, gives them the tool for one category of routine letters for four weeks, and defines in advance what "useful" means: faster drafting and no rise in the error rate her compliance team already tracks. She writes two quick scenarios for the CFO, "this becomes table stakes within a year" and "this stays a niche assistant", and notes that the pilot is a smart first step in both. The judging tool here is plain: drafting is a probabilistic system bolted onto a deterministic, regulated process, so a human still signs every letter.

Four weeks later she has evidence, not vibes: real numbers on speed and errors, three colleagues who can speak to it, and a clear recommendation. She has spent almost nothing, kept every option open, and turned "AI strategy by Friday" into a defensible decision. That is horizon scanning and emerging-tech literacy doing their actual job, not seeing the future, but refusing to be stampeded by it.

flowchart TD
  A("Weak signal spotted") --> B{"Could this materially
change our work?"}
  B -- "No" --> C(["Log it, keep watching"])
  B -- "Maybe" --> D("Run literacy check:
over- or under-hyped?")
  D --> E{"Reversible, cheap
way to learn?"}
  E -- "Yes" --> F(["Small time-boxed pilot
with success defined upfront"])
  E -- "No" --> G(["Write scenarios,
find the no-regret move"])
  F --> H("Decide on evidence")
  G --> H
From signal to decision: watch widely, bet narrowly, judge on evidence. Leaders Loop

Frequently asked questions

Isn't this just keeping up with the news?

No, and the difference matters. News tells you what's loud right now, which is precisely the peak-of-hype moment when a technology is least understood. Horizon scanning is structured (you scan across PESTLE categories, not just your feed), forward-looking (you ask what a signal could become), and it ends in a logged decision about whether to watch, pilot, or ignore. Reading more isn't the skill; deciding well on weak information is.

I'm not technical. Can I be "emerging-tech literate"?

Yes. Literacy isn't the ability to build the thing, it's the ability to interrogate it: What is it genuinely good at? Where does it fail? Is the excitement coming from customers or from people selling it? Who's relying on it in production, not just in demos? Those are leadership questions, not engineering ones. Pair them with one or two trusted technical colleagues who'll tell you when an answer is wrong.

How much time should this take?

Less than you fear. A workable rhythm is one hour a week scanning your assigned categories, a 30-minute monthly review of the signal list, and a half-day scenario session only when a genuinely high-stakes bet is on the table. The point is consistency, not volume, a steady trickle of noticing beats an annual "innovation offsite" that produces a deck and nothing else.

How do I avoid foresight becoming theatre?

Tie every exercise to a decision. A scan that doesn't change what you watch, a scenario that doesn't change what you'd do, or a hype-cycle chart that doesn't change what you'd fund is decoration. End each session with one sentence: "Because of this, we will now ___." If you can't finish the sentence, you've found nothing actionable yet, which is itself a valid, money-saving result.

When is it right to be an early adopter rather than a watcher?

When the bet is cheap and reversible, lean in, small pilots are how you learn, and the cost of being early is low. When the bet is expensive or hard to undo (re-platforming, restructuring a team, a long vendor contract), wait for evidence and let scenarios guide you. The error isn't adopting early or late; it's matching the size of the bet to the strength of the evidence.

Related in the Toolkit

Where to go next

  • "Scenarios: Uncharted Waters Ahead" (Pierre Wack, HBR, 1985), the foundational, still-readable account of scenario planning from the person who proved it at Shell.
  • The Futures Toolkit (UK Government Office for Science), a free, practical manual for horizon scanning, PESTLE driver-mapping and scenario building, with templates you can lift.
  • The Signals Are Talking (Amy Webb, 2016), a working method for telling a real trend from fringe noise, from the founder of the Future Today Institute.
  • Amy Webb's 2024 Emerging Tech Trend Report (SXSW, YouTube), a yearly, data-driven scan in action: watch how a professional foresight analyst reasons from signals to implications.
  • The Gartner Hype Cycle (overview), the five phases, plus the criticisms, so you use it as a thinking aid rather than a crystal ball.