The 7 sources of disruption

Find the disruptions before they find you. Disruptions rather than trends, because I have a problem with trends and it is worth being precise about what the problem is.

For a trend to be identified as one, it has to be in use somewhere already. By the time it has a name and a report attached, it has started losing the leverage that came from originality, which was the only thing making it worth chasing.

Following an identified trend is reassuring and it costs you the surprise your customers would have felt and the competitive break you might have made. It also removes risk from the equation, and risk is the entire philosophy of innovation, which leaves you waiting to see what competitors do before acting. That is benchmarking with better vocabulary.

So I look at sources of disruption instead: the places where change originates before anyone has packaged it. Over the years I settled on seven of them, and they have held up well enough that I still use the same list.

The argument I want to make here goes further than the list, and it partly undermines it. Those seven sources are public. Your competitors can look in exactly the same places at exactly the same moment, which means spotting something first buys you very little.

What almost nobody does is follow the consequences three steps out. Advantage lives at step three, and this article is mostly about what happens there.

The inherited reflex	Where it breaks	The shift that counts
Hunt the trendOrganisations look for trends, which by definition are movements already visible and already in use elsewhere. Watching sources of disruption means looking at the places change originates, before anything has been packaged and named.	Detection buys almost nothingThe seven categories where disruption originates are public and visible to every competitor at the same moment. Spotting a signal first confers far less advantage than the foresight industry suggests, because the signal is rarely scarce.	Advantage lives at step threeThe value comes from following consequences several steps beyond the signal, asking what follows from each answer in turn. Run that chain as a stress test on your own assets rather than as a prediction, and it stops being speculation.

Why trends are the wrong thing to hunt

The trend industry sells an object that has already lost most of its value by the time it reaches you. This is a structural feature of what a trend is rather than a criticism of the people producing them, and understanding why decides where you should be looking instead.

A trend identified is a trend already in use

Identification requires evidence, and evidence requires that enough people have already done the thing for a pattern to be visible. A movement cannot be named as a trend until it has stopped being rare, which is the whole difficulty.

By the time a movement is documented, analysed and published, its originality premium has been spent. You are buying a description of behaviour that your competitors’ customers are already exhibiting, which is useful for planning and close to useless for differentiation.

None of that makes trend reports worthless. It makes them a different product from the one they are usually sold as, closer to a market map than to an early warning system, and the two justify very different prices.

The first mover advantage that mostly is not one

The phrase first mover advantage does a great deal of unearned work in strategy conversations. Moving first is only an advantage where the position can be defended, and most of the positions organisations rush to occupy after reading a trend report cannot be.

The pattern is familiar in Australian markets. A movement gets named, everyone in the sector reads about it in the same quarter, and six competitors launch a comparable response within a year, at which point the advantage was never first mover at all.

Being early matters when you are early to a consequence nobody else has worked out. That is a different game from being early to a signal, and it is the one this article is about.

Following trends removes the risk innovation requires

Following a documented trend means waiting to see what competitors do and then doing a version of it. The comfort of that position comes from having removed the risk, and risk is the mechanism by which innovation produces returns in the first place.

There is a reason this appeals to boards. A decision supported by a published trend is defensible in a way that a decision supported by an unusual observation is not, even when the second one is better.

The discipline of deciding which movements to ignore is the other half of this work, and I have set it out separately in the case for discarding more trends than you follow. This article deals with where to look once you have stopped chasing them.

What disruption means, and what Christensen actually said

Disruption is one of the most misused words in business, and the person who coined its technical sense spent years saying so. Being clear about which meaning is in play here protects the argument and, more usefully, sharpens it.

Christensen’s definition and his objection to the misuse

Clayton Christensen, the Harvard Business School professor who introduced the term in the mid-1990s, defined disruptive innovation narrowly. It describes a process where a smaller entrant with fewer resources takes root at the bottom of a market or in a new segment, then moves upmarket until it displaces the incumbents.

The mechanism has a specific cause. Incumbents concentrate on their most profitable customers, improve their products past what other segments need, and leave the neglected ground to someone willing to serve it with something merely good enough.

Christensen himself objected to how the word travelled. Writing with Michael Raynor and Rory McDonald, he argued that the disruptive label had been applied far too carelessly to any newcomer shaking up an established industry, and used Uber as an example of a company frequently described as disruptive that does not fit his theory.

The Christensen Institute, which maintains the theory, is equally blunt about it. Disruptive innovations are not breakthrough technologies that make good products better, they are innovations that make products more accessible and affordable to a larger population.

What this article means by disruption

My use of the word is the broader one, and I would rather declare that than borrow authority I have not earned. A disruption here is any development that breaks an assumption a market has been relying on, whether or not it arrives from below in Christensen’s sense.

A regulatory change can do it. So can a price curve crossing a threshold, a technology arriving from an unrelated field, or a behaviour that was ridiculous last year and ordinary this year.

The test I apply is practical rather than theoretical. If a development would make one of your current assumptions false, it counts, and the seven sources are simply the categories where such developments tend to originate.

Why the two views are complementary

The distinction between the two uses is not a quarrel, and treating it as one wastes something useful. Christensen’s theory explains a mechanism, usually clearest in hindsight, by which incumbents lose to entrants they were right to ignore at the time.

The seven sources answer a different question, which is where to look before anything has happened. One is diagnostic and retrospective, the other is exploratory and prospective, and an organisation needs both at different moments.

Where they meet is worth noting. Several of Christensen’s classic cases would have shown up years earlier in two of my categories, exponentials and inversions, if anyone had been looking there deliberately rather than reading their own market’s trade press.

The seven sources of disruption

These are the seven categories I use, and they are the framework I published under my own name and have kept using since. They are places to look rather than predictions, and their value is that they send you somewhere other than your own industry’s trade press.

They also work hardest in combination. The developments that reshape a market almost always sit in two or three categories at once, which is what makes them large enough to matter, so a signal appearing in only one is usually a product feature rather than a disruption.

Cross-pollination

Cross-pollination is the mixing of ideas, technologies and approaches from different fields to produce something neither field would have reached alone. It is the most reliable of the seven, because the raw material already exists and only the combination is new.

The automotive industry has drawn repeatedly on aerospace, borrowing materials science developed for aircraft to make vehicles lighter and stronger, which improved both fuel efficiency and crash performance. Neither industry set out to help the other.

The practical instruction is simple and almost nobody follows it. Read one publication a month from an industry with no connection to yours, and pay attention to the constraints they have solved rather than the products they have launched.

This is the source I have written about most, and there is a fuller treatment of the method in the piece on cross-pollination and adjacent innovation.

Liberations

A liberation is a moment when a constraint or a limitation is lifted, which creates opportunities that were previously unavailable to anyone. Deregulation is the obvious form, and it is far from the only one.

Telecommunications is the textbook case. Opening those markets to competition brought in new operators, collapsed prices for consumers and reorganised an entire industry that had been stable for decades.

Australia’s Consumer Data Right is a live local example of the same mechanism. Giving customers the right to direct their own banking and energy data to a third party released something that had been locked inside incumbents, and the businesses built on that release did not exist before it.

Liberations are the easiest of the seven to monitor, because they arrive through legislation, regulators and standards bodies with consultation periods attached. Anyone can read a draft instrument, and hardly anyone in a small business ever does.

Exponentials

Exponentials are domains where progress compounds rather than accumulating in a straight line, which makes them systematically underestimated by people planning with spreadsheets. Computing, artificial intelligence, biotechnology, storage and solar generation all behave this way.

The difficulty is cognitive rather than informational. Human intuition reads a curve as a line, so a technology that is irrelevant at one per cent of a market stays mentally irrelevant right up until it is at forty per cent and the incumbents are restructuring.

Additive manufacturing has followed this pattern in manufacturing and construction, moving from prototyping toys to producing custom parts and structures at costs that make small production runs viable for businesses that could never have tooled for them.

The instruction here is to watch cost curves rather than product launches. A price falling by half every few years tells you more about the next decade than any announcement, and it is usually published by someone whose job is measuring it.

Contradictions

Contradictions are situations where two things a market treats as incompatible turn out to be combinable, which invalidates a trade-off everyone had accepted as permanent. These are the most profitable of the seven and the hardest to see, because the trade-off feels like physics rather than assumption.

Electric vehicles did this to an assumption the car industry had held for a century, that environmental performance and driving performance pulled in opposite directions. Once one product demonstrated both, the trade-off stopped being a constraint and became a design choice.

Every industry has two or three of these sitting in plain view. Fast and cheap, personal and scalable, sustainable and affordable, all stated confidently by people who have never tested whether the trade-off is real or inherited.

The exercise takes an hour. List the trade-offs your industry treats as laws, then ask for each one what would have to be true for both sides to hold at once, and check whether any of those conditions have become available since. That last question is where the money sits.

Inversions

An inversion is a situation where the traditional model is turned upside down, so the thing customers used to buy becomes the thing they get free and something adjacent becomes the product. It rearranges an industry without inventing any new technology at all.

Music streaming did this. The industry sold ownership of individual recordings, and the streaming model inverted it into unlimited access for a monthly subscription, which changed what an artist earned, what a label did and what a customer valued.

Buy now pay later is an Australian instance of the same move. Consumer credit had been sold to the borrower with interest as the revenue, and the model inverted it so the merchant pays for the conversion while the customer pays nothing extra when they pay on time.

To find inversions, write down who pays, who benefits and who bears the risk in your industry, then swap any two of them and ask whether the resulting business could work. Most cannot. One usually can.

Coincidences

Coincidences are unplanned events that, when someone acts on them, produce lasting change well beyond the event itself. They are the only one of the seven you cannot go looking for, which makes readiness the entire skill.

A public health emergency reorganised where office work happens, and the change outlasted the emergency by a wide margin. The technology enabling it had existed for a decade and the adoption had been blocked by managerial preference rather than capability.

That is the pattern worth learning. A coincidence rarely creates the capability, it removes the objection, and the organisations that benefit are the ones that had already built the capability while everyone was still arguing about whether it was necessary.

The preparation is unglamorous. Keep a short list of changes you would make if the usual objection disappeared, review it occasionally, and you will be the business that moves in the week rather than the quarter.

Strangenesses

Strangenesses are ideas or behaviours that look bizarre, trivial or faintly embarrassing at first sight and later turn out to have been early. They are the source most likely to be dismissed in a meeting, which is precisely why they carry information.

Cryptocurrency was treated as a curiosity for hobbyists for years before it forced financial institutions and regulators to develop positions on it. Whatever one concludes about its eventual value, the dismissal was not based on analysis.

The reason strangenesses get missed is social rather than analytical. Raising one in a leadership meeting costs credibility, so the people who notice them learn to keep quiet, and the organisation concludes that nobody noticed anything.

The fix is procedural. Give someone the explicit job of bringing three odd things to each meeting, with no obligation to defend them, and the social cost disappears because the strangeness was requested rather than volunteered.

Why detection is worth almost nothing

Having given you the seven, I want to undercut them, because the foresight industry sells detection and detection is the cheap part. The signal you find in these categories is almost never scarce, and an advantage built on a non-scarce input does not last.

Everyone sees the same signals at the same time

The sources of disruption listed above are public. Legislation is published, cost curves are measured by people who publish them, odd behaviours appear on the same platforms everyone reads, and your competitors have access to every one of them on the day you do.

The fantasy sold by most foresight services is privileged access, a signal nobody else has seen. That is occasionally true in deep technical fields and it is rarely true in the markets most Australian businesses operate in.

Watch what happens after a genuinely significant signal appears in a sector. Everyone identifies it within about a quarter, most produce a similar first response, and the differences that matter show up two years later in decisions nobody was discussing at the time.

Advantage lives at step three

Take any signal and ask what follows from it. Then ask the same question of the answer, and again of the answer to that. Almost every organisation stops at step one, a few reach step two, and step three is where a strategy appears that nobody else is preparing for.

Step one is the obvious consequence, which everyone reaches and which is therefore already priced into your competitors’ plans. Step two is the consequence for your immediate market, which the better competitors reach.

Step three is where the second order effects land: on adjacent markets, on cost structures, on regulation and on the assumptions holding your own business model together. That is also where the reasoning stops feeling comfortable, which is why people stop.

Three is a minimum rather than a target. I have run chains that only became interesting at step five, and I have never seen a chain become interesting at step one.

Run the chain as a stress test, not a prediction

The obvious objection to this method is that it is unfalsifiable speculation. Each step multiplies the uncertainty of the one before it, so three links out you are producing fiction with the confidence of arithmetic, and scenario planning exists precisely because linear chains fail.

That objection is correct if you treat the output as a forecast. The chain is not a bet on a future arriving, and reading it as one guarantees the failure the objection describes.

Run it instead as a stress test on your own assets. The question at each step is not how likely is this, it is which thing we currently own stops being valuable if this holds, and that question has a useful answer even when the branch never arrives.

This is Pierre Wack’s original logic at Shell in the 1970s. Scenarios were built to change what managers noticed rather than to predict the oil market, and the preparation survived even when the specific future did not.

A worked example: Australian rooftop solar

Abstract method is easy to agree with and hard to use, so here is the chain run properly on a signal every Australian can see from their own street. It starts in two of the seven sources at once and ends somewhere most energy businesses were slow to reach.

The signal, and where it sits in the seven

The signal is an exponential: the cost of solar generation fell far enough, for long enough, that household generation stopped being a statement and became an economic decision. A liberation sits underneath it, since incentive schemes and connection rules made the decision available to ordinary households.

The scale is no longer marginal. According to the Clean Energy Council, using Clean Energy Regulator data, rooftop solar supplied 12.8 per cent of all electricity generated in Australia in the first half of 2025, across 4.2 million households and businesses.

A second exponential arrived behind it. The same reporting found 85,000 home battery units sold in that half year, an increase of 191 per cent on the same period a year earlier, bringing the cumulative installed total to around 271,000.

Anyone in the sector could read both figures on the day they were published. That is the point of the exercise: the signal is free, and what follows is not.

The chain, step by step

Each step asks the same question of the previous answer. And so what follows from that? The instruction is to keep going past the point where it feels productive, because that is roughly where everyone else has stopped.

1. Solar becomes cheap enough that household generation is an ordinary purchase. And so what?
2. A large share of households generate a material part of their own electricity. And so what?
3. Batteries let them store and consume what they generate instead of exporting it, so both the energy they buy and the energy they sell back decline together. And so what?
4. Network costs are largely fixed and are recovered per unit of energy delivered, so the same poles and wires must be paid for across fewer units. And so what?
5. The per-unit charge rises for whoever is left on the network, and those people are disproportionately renters and apartment residents who cannot install anything on a roof they do not own. And so what?
6. A cost recovery problem becomes an equity problem, an equity problem becomes a political problem, and the regulatory basis for charging changes.

Step five is where the chain stops being an energy discussion. Around a quarter of Australia’s occupied homes are rented, which turns a technical question about tariff design into a question about who is subsidising whom, and that is not a question that stays inside an industry.

Notice also what happened at step three. The battery is what converts a solar household from a customer who buys less into a customer who barely transacts at all, and the battery figures were the fastest moving number in the whole dataset.

The asset that stops being valuable

Run as a stress test rather than a forecast, the chain produces one uncomfortable sentence. The assumption that network revenue scales with delivered volume stops being safe, and every asset and contract built on that assumption inherits the problem.

That is a useful finding whether or not the specific future arrives on schedule. It tells an operator which part of the balance sheet to examine first, and it does so without requiring anybody to be right about a date.

Strategies follow from the finding rather than from the prediction. Charge for availability rather than volume, lease the infrastructure instead of selling through it, sell or lease household generation and storage rather than resisting it, and look hard at electric vehicle charging, which is a load that solar alone will not cover and which keeps a household connected.

The chain can also be wrong in instructive ways, and saying so protects the method. Battery uptake could plateau, regulators could redesign tariffs before the equity problem becomes political, and a load nobody has modelled could arrive and reverse the volume decline entirely.

Each of those possibilities changes the timing rather than the exposure. The assumption under examination remains exposed in every branch, which is the difference between a stress test and a forecast, and the reason the exercise survives being wrong about dates.

None of those requires special access to information. Every input above was published by an Australian body whose job is publishing it, which is exactly why detection was never the scarce part.

How to choose where to spend your foresight effort

Three ways of doing this work are available to an Australian business and they cost very different amounts of money and attention. They also produce different things, and choosing badly is how organisations end up with an annual report nobody acts on.

Three approaches compared

Set them side by side and the trade-off is between breadth and depth. Buying breadth is fast and produces material everyone else has. Building depth is slow and produces the only output competitors cannot replicate by writing the same cheque.

Approach	What it actually buys	The signal you need it	The main risk
Buy a trend report	Breadth, vocabulary and a defensible reference for a board paper	You need to brief a board or a bank quickly and have no internal view	Your competitors bought the same document and reached the same conclusions
Build an internal watch routine	Continuous awareness owned by people who know your customers	You keep being surprised by things your own staff had already noticed	It confirms what the business already believed unless outsiders are involved
Run consequence chains on a few signals	Second and third order implications for your specific assets	You already know the signals and cannot say what they mean for you	Uncomfortable findings, and the discipline collapses without someone senior in the room

Which signals point to which approach

Answer one question honestly before spending anything. Can your leadership team already name the three developments most likely to break an assumption in your business? If they can, buying more detection is a waste and the money belongs in derivation.

If they cannot name three, the problem is a watch routine rather than a report, because a report bought once will not build the habit that produces the answer next year.

The sequencing mistake I see most often is buying the report first. It feels like progress, it produces a document, and it lets an organisation postpone the harder conversation about which of its own assumptions is exposed.

Most organisations need the third approach and buy the first. The third costs less and is harder, which is a combination that reliably loses to a procurement process.

How I can help you find your disruptions

Almost nobody I work with needs more information about the future. They need someone who will keep asking what follows from that after the room would rather stop, and who has no stake in the answer being comfortable. That is where my work concentrates.

Keynotes and workshops

Keynotes work best where a leadership group has been treating this as a research budget question. Running one live chain on their own market, in front of everyone, does more than an hour on methodology because the discomfort arrives in public.

Workshops go further by taking three signals the group already knows about and pushing each one past step three. The output is a short list of exposed assumptions rather than a report, which is harder to file and ignore.

Diagnostics and innovation programs

The diagnostic looks at how far your organisation currently reasons before it stops, which is usually further in private conversation than in any meeting. The gap between those two distances is the finding.

Longer programs then work on the conditions that let the reasoning continue, and there is more on what that requires of a leader in the eight essentials for leading innovation. I have also applied the same argument to my own trade in the piece on disruption in consulting.

If you want to test whether this fits your situation, a conversation beats a proposal. You can tell me one signal your industry is arguing about and we will take it three steps out before the call ends.

Conclusion: the signal is free, the reasoning is not

Identifying a potential disruption is the beginning of the work rather than the result of it. That sentence sounds obvious and the entire foresight market is built on organisations behaving as though the opposite were true.

The seven sources of disruption send you to better places than a trend report will: cross-pollination, liberations, exponentials, contradictions, inversions, coincidences and strangenesses. Use them, and hold on to the fact that your competitors can use them equally well.

What separates businesses is what happens after the signal. Ask what follows, then ask it again, and keep going past the step where the reasoning starts landing on your own assets and the room finds reasons to stop.

Disruptions also cut both ways, and pretending otherwise would be dishonest. They create openings and they destroy livelihoods, which puts a real obligation on anyone acting early to think about the consequences of the thing they are building as carefully as they thought about the opportunity.

Find the disruptions before they find you, and remember that finding them was never the hard part.

Frequently asked questions about sources of disruption