A product can be admired, benchmarked and, in time, copied. A business model, the way an organisation creates value, delivers it and is paid for it, is much harder to replicate, because it is woven into cost structures, relationships, incentives and data that a competitor cannot simply photograph. This is why the sharpest competition now happens at the level of the model. An incumbent can often match a challenger's features within a year; matching the economics that let the challenger offer those features is another matter entirely.

Digitisation is what moved the contest here. When coordination and distribution were expensive, most viable models looked alike within an industry, and competition was a race on product and price. As the cost of coordinating through interfaces falls towards zero, models that were previously impossible become ordinary: access instead of ownership, platforms instead of pipelines, outcomes instead of effort. Each of these rearranges who bears which cost and who captures which value, and an industry can be reshaped by a firm that never builds a better product at all.

There is a second force at work. Every product is now also a stream of information and a standing relationship. A machine reports its own condition; a contract renews itself; a customer leaves a trail that can be understood and served. That instrumentation opens capture mechanisms that a purely physical product could not support, and it rewards the firm that treats the relationship, not the transaction, as the thing it is really selling.

The uncomfortable implication for incumbents is that competence at the old game offers little protection in the new one. A firm can be excellent at making and selling a product, hold the best engineers and the strongest brand, and still be outflanked by a rival whose product is merely adequate but whose model changes what the customer has to buy, when they pay for it, and what they are locked into afterwards. The threat rarely announces itself as competition, because it does not compete on the terms the incumbent recognises. By the time it is legible as a rival, the economics have usually already moved.

Beneath the vocabulary, a model answers three questions honestly: what value is created, for whom, and how a durable share of it is captured. The three are easy to conflate and expensive to confuse. It is entirely possible to create enormous value and capture almost none of it, and plenty of admired innovations have done exactly that, handing the surplus to customers or complementors while starving themselves.

Two disciplines keep the analysis honest. The first is unit economics: a model has to make sense one customer, one transaction, one subscription at a time, at the scale it aspires to, not merely in a spreadsheet that assumes the good years. Growth that loses a little more with every unit sold is not a business model; it is a countdown. The second is the question of the moat: what makes the captured value durable rather than transient. The sources that endure are few, network effects that make the offering more valuable as it grows, switching costs that make leaving painful, genuine economies of scale, and proprietary data with a learning loop that improves the offering faster than a rival can catch up.

It also helps to separate value from price, and price from cost. The value a customer receives sets a ceiling on what they might pay; the intensity of competition and the customer's alternatives set the floor; and where the price actually lands between the two is a question of the model, not of the product's merit alone. Firms that fixate on building ever more value while neglecting the mechanism that converts value into price tend to be among the most admired and the least profitable in their market.

A model that creates value, captures a defensible share of it, and stands up as unit economics is worth building. One that satisfies only one or two of the three is a hypothesis in need of more work, however elegant the proposition sounds.

A concrete case sharpens the distinction. A manufacturer that instruments its machines can create real value for customers by predicting failures before they happen. Whether it captures any of that value depends entirely on the model wrapped around the capability. Sold as a one-off feature, the surplus flows to the buyer and little returns; offered as an uptime guarantee priced against the downtime it prevents, a durable share comes back to the manufacturer, and a data advantage compounds with every machine in the field. Same capability, same value created, radically different value captured, and the difference is a choice of model rather than a matter of engineering.

New models are rarely invented from nothing. They are more often a known pattern applied to an industry that had not yet met it, and the useful skill is recognising which pattern fits the value at hand and what it demands in return.

Access over ownership. Subscription and as-a-service models trade a one-off sale for a continuing relationship. They smooth revenue and deepen the data relationship, but they move the centre of gravity from winning the sale to earning the renewal, which is a different discipline and a different cost base. A subscription without a retention capability is a discount with extra steps.

Platforms and marketplaces. Rather than making the product, the firm orchestrates others who do, and captures a share of the exchange. The economics can be extraordinary because of network effects, but they are unforgiving at the start: a marketplace without liquidity is an empty room, and solving the initial chicken-and-egg problem is most of the work.

Outcomes over effort. Charging for a result rather than for hours or units aligns the supplier with the customer and can command a premium, but it only works where the outcome can be measured cleanly and the risk can be shared without either party gaming it. It presumes instrumentation the older model never needed.

Data and ecosystem models. Some value is captured not from the immediate transaction but from the aggregate understanding it produces, or from a position within a wider ecosystem whose participants each contribute and draw value. These are powerful and easily overreached; they demand clarity about consent, trust and the fair division of the surplus, or they collapse under their own cleverness.

Bundling and unbundling. Value can be repackaged as readily as it is created. Bundling raises willingness to pay by combining goods whose value is complementary, and it can defend a position by making the whole harder to leave than any of its parts; unbundling attacks an incumbent by isolating the one component a customer actually wants and pricing it on its own. A good deal of what looks like radical disruption is, on inspection, a disciplined act of unbundling followed by a re-bundling on terms that favour the newcomer.

These patterns are rarely used in isolation. A durable model often layers them, a platform with a subscription tier, an outcome guarantee wrapped inside an access model, and the art is in choosing a combination whose parts reinforce rather than quietly undermine one another. The failure is seldom a bad pattern in itself; it is a combination assembled for novelty rather than for the specific economics of the customer being served.

Separate the proposition from its delivery. The way value is created and paid for should be able to change without rebuilding the machinery beneath it. When the model and its implementation are fused, every pricing experiment becomes a systems project, and the organisation stops experimenting long before it has found the right model.

Instrument the model. Unit economics that cannot be measured cannot be managed. The metrics that matter, acquisition cost, contribution per unit, retention, the shape of the cohort over time, have to be first-class citizens of the architecture, not a quarterly reconstruction from exports and guesswork.

Treat the model as a hypothesis. A new model is a claim about the world that may be wrong. Design it so that claims can be tested cheaply and reversed without drama, small experiments with real customers and real money, rather than a single large bet defended long after the evidence has turned.

Fit the architecture to the economics. A subscription model needs billing, entitlement and retention capabilities a transaction model never required; a platform needs trust, settlement and matching; an outcome model needs measurement it can defend to a sceptical customer. The architecture is not neutral to the model, and pretending otherwise is how promising models die in operations.

Shape the organisation to run the model. A model is not only a set of systems but a set of behaviours, and Conway's law applies here as everywhere: an organisation built to win one-off sales will quietly resist a model that rewards retention, whatever the strategy deck says. The incentives, the metrics people are judged on, and the teams that own the customer have to be redesigned alongside the model, or the model will be adopted in name and defeated in practice.

Mind the timing. A model can be right and still fail for being early, arriving before the enabling technology is cheap enough or the customer's habits have shifted, or for being late, after a rival has already claimed the network. Reading where a market sits on that curve is part of the design rather than a matter of luck, and it is why the same model can be a triumph in one year and a cautionary tale a few years either side of it.

Product dressed as model. A better feature is launched and described as a new business model. Nothing about who pays, for what, or how value is captured has actually changed, and the economics are the old economics with fresh vocabulary.

Growth without unit economics. The model scales enthusiastically while losing a little more with each unit, on the faith that volume will eventually fix the arithmetic. It rarely does, because the thing that needed fixing was structural, not a matter of scale.

Cannibalisation paralysis. The new model threatens the margin of the old one, so it is quietly under-resourced and never allowed to compete, until a challenger with nothing to protect does the disrupting instead.

The borrowed pattern. A pattern that worked elsewhere is copied without its preconditions, a marketplace launched without liquidity, a subscription without a reason to stay, an outcome model without measurable outcomes. The pattern was never the point; the preconditions were.

Outsourcing the core of the model. In the rush to move quickly, the very capability that would have been the moat, the data, the matching engine, the direct customer relationship, is handed to a supplier who then holds the leverage. Speed is bought at the price of the defensibility the model existed to create, and the surplus flows to whoever owns the part that was given away.

We start with the model you already have, made explicit. Much of the difficulty is that the current value logic and its true unit economics are only half-known, spread across finance, product and operations, and never assembled into a single honest picture. We assemble it, and it usually reveals both where value is quietly leaking and where a better model is hiding in plain sight.

From there we generate candidate models deliberately rather than hoping for inspiration, using the patterns above as a structured search rather than a menu. Each candidate is pressed against the same three questions and the same unit economics, so that the shortlist is made of models that could actually pay rather than models that merely sound modern.

We then design the cheapest credible test for the most promising, a small slice of real customers, real pricing and real delivery, instrumented so that the economics are visible from the first week. And we sequence the shift so that the existing business funds the transition rather than being destabilised by it, protecting the core while the new model earns the right to grow.

Throughout, we keep the arithmetic in front of the room. It is easy for a compelling narrative to outrun its economics, so each candidate carries its unit economics as plainly as its story, and a model is advanced not because it is exciting but because the numbers, tested rather than assumed, hold up under pressure. That discipline is unglamorous, and it is most of the difference between a model that scales and one that merely launches.

None of this works as a document handed over at the end. Model change touches finance, product, operations, sales and legal at once, and a model designed without them is a model that will be quietly declined by the very people asked to run it. We do the work with those people in the room, so that what emerges is a model the organisation has helped shape, and is therefore willing to defend when it meets its first difficult quarter.

We combine three literacies that are usually found in different rooms: the strategic reading of where value is moving, the financial discipline to model the unit economics without flattering them, and the architectural knowledge to build the instrumentation and the delivery a new model requires. That combination is what lets us take a model from a workshop sketch to something running, measured and honest about its own numbers.

There is also a practical corollary that changes what the work is permitted to assume. When an engagement calls for a capability that does not yet exist, it need not wait on a procurement cycle or a vendor's roadmap. The Nashua 360 Enterprise Platform is built to accommodate almost any feature at pace, through extreme vibe coding: what is needed is described in plain language and generated quickly, but always within firm architecture principles and under stringent quality assurance, so that speed never comes at the cost of coherence, security or control. The effect is strategic rather than merely convenient. It moves the make-or-buy line, keeps optionality cheap, and lets the architecture follow the strategy rather than the strategy bending to whatever happened to be on a shelf.

A new model almost always needs a capability the old business never built, a way to meter usage, to settle between parties, to prove an outcome, to keep a subscriber. Being able to stand those capabilities up quickly, and safely, is often the difference between a model that is tested this quarter and one that is discussed for a year. The result we aim for is not a clever slide, but a model that survives contact with real customers, and an organisation equipped to run it.