Every leader who has ever sat in a "do we bet on this technology yet?" meeting has argued, without knowing it, about a curve. Two curves, in fact, and most of the heat in the room comes from people unknowingly arguing about different ones. Get the two straight and the disagreement usually dissolves into a decision.

The quick version

  • The technology S-curve describes how one technology improves over effort or time: slow, then steep, then flattening as it nears its physical limit. Knowing where you sit on it tells you whether to keep investing or jump to the next curve.
  • The adoption lifecycle describes people, not the technology: innovators, early adopters, an early majority, a late majority and laggards, a roughly bell-shaped spread first measured by Everett Rogers.
  • Geoffrey Moore's contribution was the chasm: a gap between visionary early adopters and the pragmatic early majority that kills many promising products.
  • So the move is to ask two separate questions, how much improvement is left in this technology? and which kind of customer are we actually selling to right now?, and stop conflating them.

The idea in depth

The first curve is about the technology itself. Plot the performance of a technology against the effort (engineering hours, R&D dollars, cumulative attempts) poured into it, and you tend to get an S: a slow, frustrating start while the basics are worked out; a steep middle where each unit of effort buys a large gain; and a flattening top where the technology bumps against a natural ceiling and further effort buys almost nothing. Richard Foster set this out in Innovation: The Attacker's Advantage (1986), and drew the conclusion that still unsettles incumbents: the company furthest up its curve, the market leader, the one with the most to defend, is also the one with the least improvement left to buy. A challenger starting a fresh curve, with a higher ceiling, eventually overtakes it. Foster called this "the attacker's advantage." (For a free summary of the argument, see Farnam Street's write-up of Foster.)

The move: plot, honestly, where your core technology sits on its own S-curve before you reach for its defence. If each engineering quarter delivers a smaller gain than the last, you are near the top, and the right bet may be the awkward second curve that looks worse today but has more headroom left in it. The discomfort is the point. A new curve almost always underperforms the mature one it will eventually replace, right up until it doesn't.

An honest limitation: you can rarely see the ceiling of an S-curve until you are sitting on it. Researchers who have examined real cases, including a much-cited Production and Operations Management study (Christensen, 1992) testing Foster's framework against the disk-drive industry, found the curve a powerful lens but a poor crystal ball: the "limit" often moves as new science arrives, and timing the jump is far harder than the tidy diagram suggests. Treat the S-curve as a way to ask better questions, not as a forecast.

The second curve is about people, not technology

The S-curve says nothing about who buys. For that, Everett Rogers' Diffusion of Innovations (first published 1962, now in its 5th edition) gives the canonical map. Rogers found that adopters of a new idea fall into a roughly normal distribution and split into five groups by how readily they take it on: innovators (the venturesome first ~2.5%), early adopters (the respected opinion-leaders, ~13.5%), the early majority (deliberate pragmatists, ~34%), the late majority (skeptical, ~34%) and laggards (tradition-bound, the final ~16%). The percentages come from standard deviations of a bell curve, not field counts of every product, so treat them as a clean mental model rather than a measured constant.

flowchart LR
  A(["Innovators
~2.5%"]) --> B(["Early adopters
~13.5%"])
  B -.->|the chasm| C(["Early majority
~34%"])
  C --> D(["Late majority
~34%"])
  D --> E(["Laggards
~16%"])
Rogers' five adopter groups, with the gap Geoffrey Moore added between early adopters and the early majority. Leaders Loop

Frank Bass turned the same idea into arithmetic. His 1969 Management Science paper, "A New Product Growth for Model Consumer Durables," modelled adoption as two forces, innovation (people who adopt independently, from advertising or novelty) and imitation (people who adopt because others already have). Fitted to eleven consumer durables, it reproduced the lifecycle's rise and fall well enough that INFORMS members later voted it one of the most influential papers in the journal's first fifty years. The Bass model is why "it'll spread by word of mouth" is more than a hope: imitation is a measurable term in a real equation.

The move: name which group you are actually selling to this quarter, because each one buys for its own reasons. Early adopters buy a vision and will forgive rough edges. The early majority buys a reference, proof that someone like them already succeeded with it. Pitch a pragmatist in visionary language and you lose them, however good the demo.

Where the two curves meet, and where products die

Geoffrey Moore's Crossing the Chasm (1991; revised 2014, HarperBusiness) is the bridge between the two ideas, and it is the one most leaders half-remember. Moore's insight was that the move from early adopters to the early majority is not a smooth handover but a chasm, because the two groups want opposite things. Early adopters (visionaries) want to be first and to leapfrog competitors; the early majority (pragmatists) want a safe, complete, proven product and explicitly look to peers, not visionaries, for their cue. A product can be a darling of the early market and still fall into the gap, because the references that delighted the visionaries mean nothing to the pragmatists.

The S-curve tells you whether the technology is still worth improving. The adoption lifecycle tells you which customer you are improving it for. Confuse the two and you optimise the wrong thing.

The move is Moore's own prescription, and it is counter-intuitive: to cross, narrow rather than broaden. Pick one beachhead, a single, specific use case for one specific segment, and deliver the "whole product" that completely solves it, so a pragmatist can adopt with no missing pieces. A dominant niche generates the peer references that let you spread to the next niche, and the next. An honest limitation: Moore's evidence is illustrative case studies, not controlled trials, and some critics argue that in modern software, distribution and viral loops can blur or shrink the chasm. Use it as a strategic lens; don't treat the bell curve as a literal timetable.

A worked example

Picture a mid-size logistics firm, call it illustrative, because the numbers below are invented to show the mechanics, not reported from a real company. Its routing software runs on a fifteen-year-old optimisation engine. For years, each upgrade shaved meaningful time off deliveries; lately, a full quarter of engineering effort improves route efficiency by a rounding error. That is the top of an S-curve. A small internal team has prototyped a machine-learning router that today is worse on average, but improves noticeably every month. That is the bottom of a second curve with more headroom. Foster's lens says the uncomfortable thing: keep a foot on the old engine for reliability, but the real investment belongs in the curve that still has room to climb.

Now overlay the adoption lifecycle, internally this time. The first depots to take the new router are the two run by tinkerers who love new tools (innovators and early adopters); they tolerate the rough edges because they buy the vision. The leadership team's instinct is to roll it out company-wide on that strength. The chasm is exactly here. The other twenty depot managers are pragmatists: they will not switch on the say-so of the company's two known gadget enthusiasts. The move is not a big-bang rollout; it is to pick one beachhead, say, all same-day depots in one region, make the new router complete enough that a cautious manager has nothing left to worry about, and turn that region into the peer reference the rest of the network will actually believe. Two curves, two decisions, one coherent plan.

flowchart TD
  Q(["A technology bet"]) --> S(["Q1: Where on the S-curve?
How much improvement is left?"])
  Q --> P(["Q2: Which adopter group
are we selling to now?"])
  S --> S1(["Near the top → fund the next curve"])
  P --> P1(["Crossing to the majority?
→ pick a beachhead, build the whole product"])
  S1 --> R(["One coherent plan"])
  P1 --> R
Keep the two questions separate, then recombine them into a single decision. Leaders Loop

Frequently asked questions

Are the S-curve and the adoption curve the same thing?

No, and conflating them is the most common error. The S-curve plots a single technology's performance against effort. The adoption curve (and its bell-shaped cousin, the diffusion curve) plots how many people have adopted over time. A technology can be improving fast (mid S-curve) while almost nobody has adopted it yet, and vice versa.

Where do those 2.5% / 13.5% / 34% numbers come from?

From Rogers' decision to slice a normal distribution at fixed standard deviations from the mean. They are a clean way to think about relative group sizes, not a measured law of nature, real products vary widely. Use the shape and the order of the groups; don't quote the percentages as if they were counted.

Is "crossing the chasm" still relevant for software that spreads virally?

Partly. Moore's core point, that pragmatists demand peer references and a complete product, holds up well. What has changed is speed: modern distribution, free tiers and network effects can compress the timeline and let some products reach the majority faster. The chasm is better read as a difference in customer psychology than as a fixed gap on a chart.

How do I know when I'm near the top of an S-curve?

The tell is diminishing returns: each comparable unit of effort (an engineering quarter, a research dollar) buys a smaller performance gain than the last, consistently, over several periods. One flat quarter is noise; a clear downward trend in the marginal gain is the signal to start funding the next curve, before a challenger forces the issue.

Doesn't betting on the next curve mean cannibalising what works today?

Often, yes, and that discomfort is why incumbents under-invest in it, which is precisely Foster's "attacker's advantage." The answer is rarely to abandon the cash-generating mature technology; it is to fund the next curve in parallel, protected from the mature business's metrics, so it isn't strangled for looking worse on day one. (See Three Horizons & organisational ambidexterity for how to structure that.)

Related in the Toolkit

Where to go next