The Offer They Can't Refuse: What the Father of the Theory of Constraints Taught Us About Marketing

Eliyahu Goldratt liked to open his marketing lectures with an apology. The session, he warned one room, would have to begin in a very banal way, because before anything useful could be said, he needed to define what marketing is. Not because the definitions were missing. Every marketing book of the previous fifty years, he conceded, gives a perfectly reasonable one. The trouble was that in most actual companies you will not find marketing at all. You will find sales wearing marketing's name badge, and nobody in the building able to tell you the difference.

So he offered a stronger definition, one designed to be impossible to forget. Marketing, he said, is spreading corn on the ground so the ducks come and sit. Sales is taking a gun and shooting a sitting duck. And if the duck is not sitting but darting about like a mosquito, don't blame the salesperson. Blame the fact that you don't have marketing.

He would let the laughter settle and then push the image one turn further. Shooting a sitting duck sounds easy; try it, and remember that if you miss and the bullet passes near its ear, the duck is not sitting anymore. Good marketing, he concluded, means having a field full of sitting ducks with glue on their feet.

Underneath the farmyard comedy sits a precise distinction. Marketing is the work of bringing a market to desire your product, and while you are doing it you never see an individual prospect, only a section of the market. Sales sees one specific prospect, and has one job, which is to close. Most companies, in Goldratt's estimation, do the second and call it the first. Salespeople are sent out with quotas to chase ducks that were never given a reason to land, and when the ducks keep moving, the company concludes it needs better shooters.

His claim in these lectures was that this failure is neither laziness nor stupidity. Companies skip marketing because they are trapped in a specific, diagnosable conflict, one that ties marketing's hands before any campaign is conceived. He believed the conflict could be dissolved, and that the method for dissolving it was the same one he had spent his career using on factory floors: find the constraint, find the assumptions that keep it in place, and break one of them. The result, when it works, is what he called an offer the market cannot refuse. He had a less polite name for it too, which we will get to.

The man who fixed the factory, then turned around

Some context, for anyone who came to management thinking after the operations classics stopped being required reading. Eliyahu Goldratt was an Israeli physicist who wandered into industry in the 1970s by way of a scheduling algorithm, and stayed to become one of the most quoted business thinkers of his generation. His 1984 book The Goal remains one of the strangest bestsellers in the business canon: a novel, complete with a failing marriage and a plant manager hero, that smuggled a theory of production into airport bookshops and sold millions of copies. It put words like bottleneck and throughput into the ordinary vocabulary of managers who had never read a textbook, and it is still handed to new operations hires today, four decades on. Goldratt died in 2011, at 64.

The theory the novel carried is the Theory of Constraints, and it can be stated in a breath. Every system has one constraint, one weakest link that limits the performance of the whole. Strengthen anything else and you have improved nothing, whatever the local numbers say; find the constraint, get the most out of it, subordinate everything else to it, and the whole system moves. In a factory this is concrete. The plant produces exactly what its bottleneck machine produces, an hour lost there is an hour lost forever, and an hour saved anywhere else is a mirage.

The marketing material comes from what happened when companies took him seriously. Fix production the way Goldratt prescribed, and something awkward emerges: enormous excess capacity. The plant that everyone believed was full turns out to be able to make twenty or thirty per cent more than anyone is selling. The constraint has not disappeared. It has moved, out through the loading dock and into the market. The hard problem stops being how to make the thing and becomes how to sell it, and a company that has spent years perfecting its operations discovers it has no idea how to manufacture demand.

That is the situation the workshop addressed. It was the marketing session of an eight-part program Goldratt taught on running a company by TOC logic, and it was blunter and funnier than his books. What he offered in it was a method rather than a set of slogans, a claim he backed with unusual arithmetic: he said he had personally been involved in well over three thousand of these marketing analyses. We will follow the method end to end, through a real case that doubled a company's market share, and finish with the strategic rule he considered bigger than any single offer.

Two prices for the same thing

The method starts with an observation so simple it feels beneath a physicist's attention. Any product has two values, held by two different parties, and they have nothing to do with each other.

The first is the supplier's perception of value, and it is built from effort. The more money, time and ingenuity it took to bring a product from nothing into a customer's hands, the more the supplier feels it must be worth. A thousand dollars of material went in. New technology had to be developed. A whole distribution network had to be built. Anyone who has sat in a pricing meeting has heard this value talk, usually in the form of a sentence that begins with everything we've put into this.

The second is the customer's perception of value, and it is built from benefits. What will owning this product do for me? How many of my problems does it solve, and how badly did I need them solved? Goldratt kept restating the point because he knew how hard it is to absorb: the customer's perception has nothing to do with your cost. Suppose your efforts were heroic and the product still barely touches the customer's needs. They will pay very little, and they will be right.

The trap is that companies price, plan and argue almost entirely from the first perception while revenue arrives entirely according to the second. Every founder who has justified a price with the cost of the engineering behind it is anchoring on the wrong number. The market never sees the effort. It sees the benefit, and it pays for nothing else.

Put the two perceptions side by side and you can watch them tear a company in half. Goldratt drew it as one of his conflict diagrams, but the logic works in plain sentences. A good marketing decision must produce enough sales volume, and to get volume you must act on the customer's perception of value, because they will not buy for any other reason. A good decision must also keep you solvent, which seems to mean earning a reasonable margin on each product, and margin is calculated from your costs, which means acting on the supplier's perception of value. Two requirements, pulling in opposite directions, both non-negotiable.

He offered a test for whether your company lives on this rack. Listen for a particular argument. Don't accept the order, the price this client is willing to pay is ridiculous. We must accept the order, we don't have enough work. Don't accept it. Accept it. Why did you accept that order? Every company has heard some version of this quarrel, often weekly, and Goldratt's point was that it is the conflict performing itself in the corridor. As long as the conflict stands, marketing's hands are tied, because every attempt to act on what the customer values collides with someone defending the margin, and every defence of the margin prices the product out of what the customer will pay. Companies do not resolve the tension. They oscillate on it, order by order, for decades.

Goldratt's habit, faced with any such conflict, was to refuse the compromise and attack the assumptions underneath instead. The first assumption he attacked is the one that looks safest: that protecting the margin on each product protects the company.

Cost accounting is lying to you

The belief runs deep. Every product must carry a positive margin, because any sale below cost drags the company's profit down. It has the feel of iron law, of arithmetic itself. Goldratt spent a career demonstrating that it is false, and in this seminar he did it with an example compact enough to fit on a napkin.

Start with the crack in the foundation. Product margin is calculated from product cost, and product cost depends on how the overhead is spread, which depends on how many units you make and sell. Produce one unit and it carries all the overhead alone. Produce two and its cost falls by nearly half. The margin figure that looks so authoritative on the spreadsheet is therefore a function of a sales forecast, and forecasts at the level of the individual product are, in Goldratt's word, a joke. The number carries three decimal places of confidence and rests on a guess. He noted the corroborating puzzle from the real world: most companies that lose money show a positive margin on every product they sell. If the accounting says every line is profitable while the company bleeds, it is not the company that is lying.

Then the demonstration. Take a company selling $100 million a year and losing $2 million. It has excess capacity, the constraint is in the market, and the spare capacity amounts to twenty per cent more volume than it currently sells. Now suppose it is lucky: there exists a segmented market that will absorb that extra twenty per cent, but only at prices thirty per cent below today's, which is comfortably below the products' calculated cost.

Segmented, in Goldratt's usage, has a strict meaning that will matter through the rest of this article. A section of the market is segmented if, and only if, whatever price and quantity you sell there has no effect on the prices and quantities you sell everywhere else. Sealed compartments. This has nothing to do with finding a niche, and achieving it is not easy, which is a warning we will come back to.

The received wisdom says taking this business deepens the loss, since every additional unit sells below cost. Now do the actual arithmetic. Twenty per cent more volume at the old prices would be $20 million of new sales; at thirty per cent off it is $14 million. What does producing it cost? Only the truly variable costs, the ones that rise with each unit made: raw materials, subcontracting, commissions. In this example those run at forty per cent of the original price, which on that volume is $8 million. So the extra sales bring $14 million in and send $8 million out, adding $6 million of throughput. And operating expense? Nothing changes. The capacity is excess; the people and machines are already paid for whether they run or not. The delta is zero, so the whole $6 million falls through to the bottom line. The company that was losing $2 million now makes $4 million, and every additional unit was sold at a loss on paper.

Each sale below cost, and the total is profit. Cost accounting, Goldratt concluded with some relish, is simply wrong, and he had proven it more than once. The numbers that matter are the ones the example actually used: throughput, the rate at which sales bring in fresh money after truly variable costs; investment; and operating expense. Product cost appears nowhere among them, because outside the spreadsheet it does not exist.

One caveat, and he leaned on it hard. This is a demolition of the accounting, and it is emphatically a pricing strategy for no one. In practice, companies that spot this logic cut prices without first checking that the market is truly segmented, in the strict sealed-compartment sense. The discounted units leak back into the main market, the main market demands the discount, a competitor matches it, and a price war begins, which everybody loses. He was not recommending discounts. He was clearing away a false belief so the real move would be visible, and the real move runs in the opposite direction entirely.

The job is to raise value, and there are rules

Go back to the conflict. Volume demands you act on the customer's perception of value; solvency demands you cover your expenses and investment, which is what the margin rule was clumsily trying to guarantee. Restated properly, the two sides still pull against each other, so cleaning up the accounting has not freed marketing's hands. Something else has to break.

Goldratt found it by asking when the conflict actually bites. If a customer's perception of value is already higher than what you need to charge, there is no conflict at all. No one has to choose between volume and margin when the market is happy to pay more than the price on the list; nobody volunteers to underpay you out of spite. The conflict exists in exactly one case: when the market's perception of value sits below yours. Which means the entire trap hangs on a single hidden assumption, that the customer's perception of value cannot be raised very far.

Break that assumption and the whole structure comes down. And breaking it, Goldratt said, is precisely the job description. The job of marketing is to raise the customer's perception of value until it clears the supplier's. That is the whole job, and everything else the department does is secondary to it. Raise perceived value above your own valuation, and volume and margin stop being enemies; the corridor quarrel about accepting orders goes quiet on its own.

The reflex response is that raising perceived value means innovation, a new product, something for the engineers. Goldratt agreed it helps and denied it is necessary. There is nearly always a way to raise the market's perception of value with the product you already sell. Finding it takes analysis rather than invention, and the analysis obeys three rules.

First, the offer cannot be built on a lower price. A price cut is the one brilliant idea a competitor can copy before lunch. Goldratt sketched the competitor's position with a certain cruelty: he sits back and watches. It didn't work for you? Excellent, you lost money, you're on your way out of the market. It worked? Tomorrow he cuts his prices too, and you are back where you started, poorer. Any offer meant to break a market constraint must raise value at the same price, or a higher one.

Second, the offer must solve the customer's core problem. Perceived value comes from benefits, benefits come from problems solved, and the bigger and more numerous the problems your product dissolves, the higher the value climbs. But a list of the market's assorted complaints will not do. Complaints are symptoms. Somewhere beneath them sits a core problem producing most of them at once, and the offer that solves the core problem delivers more value than any number of offers that pick off symptoms one at a time.

Third, the offer must be hard to imitate, and Goldratt's standard for this was specific: it should require at least one change to a rooted policy, a practice so long established in your industry that competitors treat it as a law of nature. A clever feature buys you weeks. A patent, he noted, buys less than it used to. A rooted policy change buys years, because before a competitor can copy the offer he must first believe, against his own instincts and his own accountants, that the sacred practice was wrong.

There is one more requirement, and it is the uncomfortable one, so it deserves its own sentence before we watch it in action. To solve the market's core problem, you must first prove that you are causing it. The cause is neither the customer's bad management nor the economy; the cause is you and your competitors together, through the rooted policies you all share. Goldratt was unsentimental about why this matters. The exercise is neither altruism nor guilt. If you and your competitors are not causing the problem, then nothing you change about your own behaviour can solve it, and you have no offer. Being the villain of the analysis is the qualification for selling the cure.

How do you prove a thing like that? You build a map. In the workshop Goldratt walked through a real one, built by a real company, and it is worth following in detail, because this map and what was constructed from it form the most complete example of the method he ever showed in public.

The map of the damage you do

The company in the case sold branded goods to independent retail shops, and it had already done its TOC homework. Production was fixed, distribution was fixed, replenishment was fast, and the freshly exposed excess capacity made the next constraint unmissable: not enough sales. So the team sat down to analyse its market, and the tool they used was what Goldratt called a current reality tree, a page or two of cause-and-effect logic connecting the market's miseries to their roots.

He was insistent about what this analysis is not. It is not a survey. Ask the market what is wrong and you will get a heap of complaints with no structure, what he called undesirable effects, and the market itself cannot tell you how they connect. The market experiences the symptoms; it has never had a reason to trace them. The tracing is the supplier's work, and the fact that almost no supplier ever does it is exactly what makes it a competitive weapon. Everyone in your industry stares at the same complaints. The edge belongs to whoever finds the machinery underneath.

Here is the machinery the team found, and it starts from a policy so standard that nobody in the industry could remember questioning it.

The brands, this company and its competitors alike, offered big discounts on large orders. List price for a small order; something like thirty per cent off for an order covering six months of sales. Consider what that policy does to a shop owner. The shop across the street stocks the same brands. If it buys at thirty per cent less than you, you are finished. So buying big stops being an option and becomes a survival requirement, and every shop is pushed into purchasing not for what it expects to sell soon but for the long term.

From that single compulsion, two opposite miseries flow at once. The first is obvious: the shop's inventory swells. Racks of it, months of it, financed at the bank. The second seems to contradict the first, and this is the kind of thing a cause-and-effect map exists to catch: the shop simultaneously runs out of its best products. A shop's forecast of what will sell is poor at the best of times. Three months in, one product turns out to be a hit and its stock is gone. The shop would love to reorder, but a small reorder gets no discount, and with mountains of the slower stock still sitting on the racks, another six-month order is unthinkable. So the shop does not reorder at all, and the shortage lands precisely on the products that were selling.

Both branches drain the same reservoir. Where does a shop get the money to buy merchandise? From the bank. More inventory means more borrowing and more interest; Goldratt passed on the line shop owners used themselves, that they were not working for their own benefit but for the bank's. And every stock-out on a hot item is revenue that walked out the door. Shop profitability sinks from both sides at once.

Now add the second industry policy: no consignment. The brands would not hear of placing goods in a shop and being paid when they sold. You buy it, you own it. So the whole swollen inventory is financed by the shopkeeper, cash needs climb, and when cash runs short the shop starts missing payments to its suppliers. The suppliers respond the way suppliers do, by demanding cash up front, which makes getting the right merchandise harder still.

At this point in the tree Goldratt stopped to point at something he considered lethal, a structure he told every analyst to hunt for: the negative loop. Profitability falls. A shop with falling profitability finds its credit line shrinking, because banks read the same statements everyone else does. Less credit tightens the cash squeeze, the tighter squeeze worsens the payment problems, the payment problems choke off merchandise, and profitability falls further, which shrinks credit again. Around and around, each turn faster. He offered an empirical test anyone can run: walk through a shopping mall, note the shops, and come back two years later. The ones that vanished were mostly caught in exactly this spiral.

And the tree has a third branch. The brands constantly release new products, and they advertise the releases heavily, in the magazines and on television, at their own expense. Product lifetimes, he observed, were shrinking everywhere; a generation of golf clubs lasted under six months in the market, a suitcase design under nine. Set that against the first branch of the tree. The shops, remember, have been compelled to buy long-term quantities of the current models. The moment the brand's advertising ignites demand for the new model, everything on the shop's racks turns nearly obsolete overnight. The shop cannot wait for fully obsolete, so it does the only rational thing, which is to fire-sale the old stock at forty per cent off before it becomes unsaleable. Picture the scene from the brand's side. It is spending a fortune persuading the public to buy the new model, while its own retail channel spends its own margin persuading the same public to buy the old one. The launch is sabotaged by the launch's own funding mechanism, and the shop's profitability takes the hit from both directions, rebates on the old stock and weak sales of the new.

Two pages of if-then, and every misery on it, the swollen inventory, the shortages on winners, the bank interest, the cash squeeze, the death spiral, the fire sales undermining every launch, traces back to policies the brands themselves wrote: volume discounts, no consignment, advertise the new hard. The suppliers are the cause. Which is the good news, in Goldratt's inverted way of seeing, because the party causing a problem is the only party positioned to sell its solution. The shops cannot fix any of this. The brands can, and the first brand to do it will be offering its market relief no competitor's brochure can match.

One practical note from the workshop on validating such a tree, because the temptation at this stage is to commission research. Don't, he said; a survey here is procrastination with a budget. Instead, take the tree to a friendly customer and read it aloud as if-then. Come without a tree and you will get two hours of unstructured bitching; come with one and every error gets pointed out precisely and the conversation stays on the rails. Correct it, take it to another customer, repeat. The finish line is unmistakable. You are done when the customer asks whether he can keep a copy.

Building the offer: turn the policies inside out

With the tree validated, constructing the offer stops being a creative mystery and becomes something closer to engineering. The harmful policies sit at the bottom of the map. Ask of each one: what should we do instead? The answers become the foundation of a second tree, a future reality tree, in which the new policies flow upward through the same if-then logic until the market's undesirable effects turn desirable. If the logic holds all the way up, you have your offer.

Watch the mechanism work, policy by policy, in the same case.

Start with the volume discount. In this company's version the products were bulky relative to their price, freight was a large share of the total, and the standing offer was free freight on orders above twenty thousand pounds, worth roughly thirty per cent. That incentive is what forces the shops to over-buy, so it goes. But wait, and this is where instinct screams: shops live on that discount; their whole business assumes it. Take it away outright and there is nothing left to sell them. The resolution is a distinction nobody in the industry had thought to draw, between rewarding the size of an order and rewarding the volume of business. Offer the shop an umbrella contract: commit to buying so much over the next six months and the discounted price applies to every order, however small each delivery. The shop keeps the economics it depends on and stops having to warehouse six months of guesswork to get them.

Next, the shortages. The shops keep running out of their best sellers, so the new policy is deliveries at whatever frequency the sales rate demands, daily, weekly, twice a month, so that a shop simply never runs dry. And here the imitation moat appears, the rooted-policy rule paying off in structure. This company could promise rapid replenishment because it had already rebuilt its production and distribution along TOC lines; small frequent deliveries cost it nothing special. For competitors running conventional distribution, with their warehouses full of forecast-driven stock in the wrong places, the same promise was close to a physical impossibility. They could hear about the offer, understand the offer, and still be years away from being able to make it.

Then the cash strangulation. The old policy was no consignment, ever; the new one inverts it completely. The goods a shop needs for proper display, plus a reasonable working stock, are placed on consignment. The shop pays when it sells, and until it sells, the inventory belongs to the brand. How gracious, Goldratt said, enjoying the look of it, and then supplied the honest gloss: as long as you're about to make a fortune from it, why not be gracious?

Those are the new policies. The future reality tree now has to prove they flip the market's miseries, and the proof runs cleanly. A shop that is never punished for ordering small and never at risk of running dry holds goods only for the near term. Holding only for the near term means far fewer slow movers on the racks, and that single change pays three ways at once, in the three currencies Goldratt always counted. New products now arrive in a shop with room for them, so launches go smoothly and throughput rises. Inventory falls, so the shop's investment falls, along with the interest that was feeding the bank. And when stock no longer sits long enough to go stale, obsolescence stops being a routine expense.

The display side compounds it. With display stock on consignment, carrying the brand's full range costs the shop nothing, so every shop shows the whole spectrum, and the replenishment keeps it complete. Full range plus constant availability moves sales three separate ways, and Goldratt took care to separate them. Customers choose the shop that always has what they came for, so traffic shifts toward it. Customers who come no longer leave empty-handed, so fewer sales are missed. And satisfaction repeats itself: find what you want three visits running and you are, in his word, conditioned; fail to find it three visits running and you stop coming. The third effect is loyalty, the slowest to build and the hardest for a competitor to claw back.

Set the two trees side by side and every arrow at the top has reversed. Inventory down, cash freed, shortages gone, launches clean, sales and loyalty climbing. That is what Goldratt meant by an unrefusable offer, and he confessed in the same breath that he hated the polite term. He preferred the original name, shared with the room as a confidence: just between you and me, the mafia offer. An offer so good the market cannot refuse it, made without a single price cut.

The results in this case bear repeating slowly. The team estimated their window, the time before competitors would see the offer working and copy it, at a minimum of two years. When Goldratt told the story, seven years had passed, the company had doubled its market share, and the competitors had still not managed to imitate the offer. Not because it was secret; because matching it meant reversing rooted policies and rebuilding distribution, and the required change in culture was simply too big. That is what the third rule buys. A discount is copied by lunchtime. A rooted policy change was, in this industry, still uncopied after seven years.

At which point you might expect the story to end with the sales force sprinting out the door. What actually happens next, Goldratt warned, is that the trouble moves indoors. The offer is now too good, and the first people to refuse it will be your own colleagues.

Yes, but: how an offer gets finished

The reaction inside the company is predictable to the word. Why should we be so generous? We wanted more sales, and you bring this? And then, from every direction, the phrase Goldratt considered the most valuable raw material in the whole process: yes, but.

His instruction at this stage is the counterintuitive one, and skipping it is where most good offers die. You are in love with the solution. You are convinced it is the best thing since sliced bread. Do not take it to the market. Take it inside, to your own company's professional sceptics, and fight the urge to defend it. In the case at hand, the team sent the future reality tree to people in ten different functions and asked for comments. Within half an hour the fax machine was busy, and the comments were not compliments. Which was exactly right, he said. You do it deliberately. You can ask for praise if you like; what you will get is reservations anyhow.

A yes-but has a precise anatomy, worth spelling out because handling it well is a skill. Yes, I understand your offer. But from it, a negative consequence follows. The objector is not pointing at a flaw that exists today; he is predicting one your solution will create. He will almost never state it in clean logic, arriving instead with what Goldratt called transatlantic arrows, vague associations and gut feeling, and at first you will not know what he is talking about. The work is yours: write his objection out in rigorous cause and effect, tie your own solution to the negative outcome he fears, then take it back to him and ask whether that is what he meant. Half the time he meant something else entirely. And when he confirms it, something better usually follows, because the person with intuition sharp enough to smell the problem usually has the fix as well, and along the way you have converted a critic into a co-author.

In this case roughly forty negative branches came back. He walked through three of them, each resolving a different way, and together they show why Goldratt called this stage the most important part of the solution.

The first would have killed the whole offer within a season. The offer frees the shops' cash; that was the point. But the shops are not the brand's shops. They carry competitors' goods too, and a shop with fresh cash surpluses will spend them somewhere. Follow the logic: the competitors' goods are bought with the shop's own money, while the brand's goods now sit on consignment, still belonging to the brand. A shop's scarcest resource is display space. Whose merchandise gets the window, the stock the shop has its own cash sunk into, or the stock it can return without loss? The objection wrote itself into two pages of if-then ending in a grim conclusion: our display share will shrink, and what is not displayed is not sold. Here was a beautiful offer, aimed carefully at the company's own foot. The fix, once the branch was visible, was almost embarrassingly small: make display share a stated condition of the offer, agreed up front. Raised in advance, no shop objected. In the field it went better than the fix required. Salespeople would present the offer, pause, mention that there remained a small question of what happens to display space, and keep their mouths shut. In every single case, Goldratt reported, the shop volunteered more space than the brand had held before. The branch that would have sunk the offer became one more thing it delivered.

The second objection turned out to be worth ninety million dollars. If the shops no longer finance the inventory sitting in their stockrooms, somebody does, and that somebody is the brand. Our cash may be deeper than any one shop's, but it is not unlimited, and on the scale of the whole market the consignment stock could squeeze us into exactly the crisis we just cured downstream. The team wrote the branch out and admitted its authors had a point. Then, staring at it, they noticed what the old system had been hiding. Under the old policies, goods invoiced on shipment came with payment terms of a hundred and twenty days, and plenty of shops stretched even those. The brand's cash had been locked up all along, in receivables. Under the new offer nothing is bought until the shop sells it, and then payment comes within the week; in practice the average came down from a hundred and twenty days to thirty. Money moved out of one pocket, inventory, and a far larger sum moved back in from another, the payables. Ninety million dollars of cash assembled itself in under six months. The objection had not merely survived scrutiny; investigating it surfaced a windfall nobody had noticed, and, just as usefully, warned the company to prepare for a cash surplus instead of being ambushed by one. Even good surprises, Goldratt noted, can put you in trouble if you have not planned for them.

The third objection solved a problem the team had been dreading all along. Shops had run for decades on the habits of ownership: buy big, hold mountains. Habits have inertia. Offer consignment to a shopkeeper trained by twenty years of volume discounts and he may go on holding mountains, only now the mountains are financed by the brand, which would find itself funding vast stocks of slow movers across an entire market. The fix the team devised was a replenishment rule with teeth: consignment stock is replenished under exactly two conditions, either the shop has reported the sale, or the shop commits to enlarging the brand's display space. Otherwise, nothing ships. The rule forces the transition to lean stock at the pace of actual sales. And it carried a bonus that dissolved a much older headache: policing. Consignment schemes classically require monitors, because a shop that sells without reporting keeps the cash and the secret. Under the replenishment rule the incentive flips by itself. A shop that hides its sales gets no replenishment and dries up on the spot, punished not by an auditor but by its own empty racks. No policemen anywhere, Goldratt said, and it works beautifully.

Follow the procedure to its end and the sequence is short enough to memorise: spread the tree to your sceptics and solicit reservations; document each one in cause and effect; verify it with its originator; get the fix from him if he has it, and break your head until you find it if he doesn't; fold every fix into the offer. What emerges is not the offer you fell in love with. It is that offer with its display-share condition attached, its cash plan prepared, its replenishment rule installed, ten times more powerful and, for the first time, finished. The corn is finally worth spreading.

Goldratt closed this part of the seminar with the claim that gives the method its nerve, and he grounded it in his three thousand cases. There is always a mafia offer. Always, even for a pure commodity, even for salt, even for matches. Many times an engagement began looking like the exception, a total impossibility, and then, after a day, two days, and if you're really dumb three days, out it came. The scarce ingredient is not opportunity. It is the willingness to sit down and do the thinking.

Segment the market, not the resources

A finished mafia offer, Goldratt then told his room, is only the first step. Something bigger had been masked all along by the same local-optima thinking, and he introduced it by rescuing a word from its common usage. Segmentation, in most marketing conversations, has become a synonym for finding a niche. That is not what it means, and the difference is worth a fortune.

Go back once more to the customer's perception of value, and this time plot it. The graph runs across your whole market: along the bottom, the value a customer places on your product; on the vertical, what share of the market holds that valuation. Without a single survey, Goldratt argued, you already know the shape. The curve starts at the origin, because anyone who values your product at zero is by definition not your prospect. It must come back down to zero somewhere on the right, because no meaningful share of any market thinks your product is worth a billion dollars. And since you are currently selling the thing, it is above zero somewhere in between. A hump, in other words: rising, peaking, falling. Every market has one, and the wider the spread of needs, the wider the hump.

Now draw a single vertical line through it, and understand that this is what a price is. One price, anywhere on that curve, dissects your market into three parts, and every one of them costs you.

To the right of your price sit the customers whose perception of value is higher than what you charge. They would gladly pay more; they will never volunteer to; you are leaving their surplus on the table permanently. In the middle sit the customers whose valuation falls just below your price but who buy anyway, because your competitors offer no escape. Goldratt had a tender name for this group: the bitching and moaning section. They feel overcharged, they complain about everything, they nurse zero loyalty, and the day any competitor undercuts you they are gone, which is uncomfortable arithmetic given that this group is usually the bulk of your sales. If your salespeople keep reporting that prices feel a little high, he noted, that is the sound of this section talking. And to the left of the line sits the largest group of all in many markets, the people whose perception of value sits so far below your price that they simply do not buy. One number on a price list has amputated them from your market entirely.

The standard conclusion is that markets are like this, uniform things with one clearing price, and some customers are simply not yours. Goldratt inverted the causality. When customers differ in their needs, the market is inherently segmented already; wide spreads of need mean enormous inherent segmentation. What makes a market look uniform is your own behaviour. One product, one price: it was the supplier's offer, and nothing else, that flattened all that diversity into a single line and then blamed the market for it. His phrasing was blunt. We make a dispersed market into a uniform market, and then we cry.

Real segmentation, then, is the art of unflattening. It means finding a way for the customer with high needs to pay a high price while the customer with modest needs pays a modest price, for what is, from your side of the counter, the same product. And it carries one test that separates the craft from the con: when the customer who paid a lot meets the customer who paid a little, and they will meet, he insisted, don't fool yourself, the one who paid more must be able to shrug. What you bought for less money was not what I needed. If the conversation ends in a shrug, the segmentation holds. If it ends in a lawsuit or a cancelled contract, what you built was resentment on an instalment plan, and the market will collect.

At first hearing this sounds impossible. Same product, honestly different prices, no resentment. Goldratt's response was a tour of companies that had done it at civilisation scale, and his first exhibit was chosen for its familiarity. Buy a return flight and the airline demands to know how long you are staying before it will quote a price. His telling of it had him at a Tel Aviv ticket counter, bristling: what business is it of yours? If I stay more days will you fly me in a different aeroplane, feed me better food? But the question is not curiosity, it is instrumentation. A passenger flying to New York for three days, jet lag and all, is travelling because he must; his need is high, and he will pay. A passenger staying three weeks might be weighing a holiday against staying home; price him gently or lose him entirely. Same aircraft, same seat, prices apart by half or more, and the length-of-stay question exists to read each passenger's position on the value curve. The scheme worked for decades, though Goldratt appended a warning we will return to: the airlines kept slicing until, in his words, they segmented themselves almost to death. Ask the passengers around you what they paid, and no two answers will match. The people who understood the logic are gone, and what remains is a nightmare wearing the logic's clothes.

His second exhibit had, in his estimation, contributed more to marketing than most textbooks: Xerox. When the copier arrived, Xerox faced the value curve in its purest form. One office needs four copies a day, another fifty-five thousand, and no conceivable machine price serves both. Sell the machine cheap and the heavy users get the bargain of the century; price it for the heavy users and the light users vanish. Xerox's answer was to stop selling the machine. The machine arrives, is installed and maintained, and costs nothing. You pay per copy. The light user pays trivially, the heavy user pays tens of thousands a month, and when the two meet, the shrug test passes effortlessly, because the question how many copies do you make settles everything. The thing the customer wanted all along was copies, and Xerox finally priced the thing the customer wanted. Whenever the spread of needs is wide, Goldratt generalised, ask whether your customer wants the product or the use of the product, and if it is the use, stop shoving the product down his throat and sell the use. The segmentation follows by itself. He practised the sermon, too: the very seminar on these tapes was sold not as tapes but as viewing rights per person, so that an individual paid modestly, a company training fifteen per cent of its staff paid more, and a company remaking its whole culture paid the most, at a discount he was happy to extend because by then its interests and his were identical.

The third exhibit was IBM in its imperial decades, and it answers the objection that all this is fine for services but a manufacturer must build different products for different segments. Picture the situation, which Goldratt insisted was generic: a high end of the market wanting a full-strength machine at a million dollars, a middle wanting a decent one at half, a low end, the largest by far, paying two hundred thousand at most. Development is the expensive part; suppose the top machine costs eighty million to develop, the middle sixty, the small forty, and your development budget stops at a hundred million. Every room he posed this to fell into the same debate: build the Rolls-Royce, or the Volkswagen for the masses? IBM built the top machine only, and then created the middle and bottom models by sabotaging it. An extra wire, a delay circuit, and the million-dollar computer became the half-million-dollar computer. A client who outgrew his machine ordered a field upgrade, which meant a technician arrived with cutters and snipped the wire. Absurd on its face, and Goldratt piled up the sense of it with visible pleasure. Development budget: one machine instead of three, with a million or two for the degradation circuits. Engineers: all working on the top of the line, which is where the best engineers insist on being. Production, purchasing, distribution, spares: one product flowing through everything, with the inventory savings that implies. Service: technicians learn one machine. Even the customer wins on the day he upgrades, because it is the same computer, fully compatible, nothing to relearn and nothing to convert. And the market, which had seemed to hold three price points, was eventually served at seven, because once you are selling capabilities rather than boxes you can slice as finely as the value curve warrants. The customer is not buying your hardware, he is buying what the hardware is permitted to do, and what it is permitted to do is a dial you control.

Hold the three exhibits together and the pattern Goldratt wanted extracted comes into focus. The airlines segmented prices without multiplying planes. Xerox segmented without multiplying machines. IBM segmented seven ways while developing, building, stocking and servicing one computer. The rule, and he asked his audience to etch it in their minds, is segment the market, not the resources. Externally, as many segments as the diversity of needs will honestly support. Internally, one product, one development effort, one distribution system, one service organisation. The airlines' later misery shows the rule's other edge: segmentation that multiplies internal complexity, fare classes breeding fare classes until nobody remembers why, is self-inflicted damage wearing strategy's clothes. And the rule outranks even the mafia offer in his hierarchy, for a reason he would develop in his session on strategy: nobody knows how the market will behave in the future, and the company that has segmented its market without segmenting its resources is the company that can be wrong about the future and still flourish.

The offer in 2026

Goldratt delivered these lectures before the word software meant subscription, which makes it easy to measure how well the ideas travel: look at what the best-run companies charge for now, and notice how much of it he would recognise.

He would recognise the modern software tier sheet instantly, because it is IBM's wire drawn in a pricing table. One codebase serves every customer; the difference between the starter plan and the enterprise plan is which capabilities have been switched on, and the upgrade is a field upgrade in the original sense, a flag flipped, the wire cut remotely. The companies that do this well obey his rule without knowing its name, segmenting the market into five tiers while running one product, one engineering team, one deployment. The companies that do it badly, drowning in bespoke contracts and per-customer forks, are the airlines, segmenting their resources along with their market and paying for it in complexity nobody can price.

He would recognise usage-based pricing as Xerox's copy meter, generalised. Paying per API call, per compute-hour, per seat actually active in the month: each is a mechanism for reading a customer's position on the value curve and charging accordingly, and each passes the shrug test for the same reason the copier did. The startup metering three hundred dollars a month and the enterprise metering three hundred thousand are buying the same product and will never feel cheated at the comparison, because how many copies did you make answers everything now as then.

And the mafia offer itself is easiest to see in the companies that identified a rooted policy their whole industry treated as physics, and broke it. When Zappos decided returns would be free in both directions and the try-on period absurd, it left the price of shoes alone and rewrote the policy, held by every shoe retailer at once, that the customer bears the risk of a bad fit. The undesirable effect that kept people from buying shoes online turned desirable, buying became risk-free browsing, and the years competitors spent unable to stomach the returns bill were the window Goldratt promised. He would want us to notice the anatomy, though, more than the example: the offer solved the customer's core problem, it contained no discount, and it rested on a policy change that rivals found culturally almost impossible to copy. Any market where every supplier maintains the same customer-hostile practice because that is how the industry works is a market with a mafia offer lying in it, unbuilt.

The deeper transfer is the diagnosis underneath all three. A company whose product, fulfilment and infrastructure have stopped being the limit on its growth has a constraint in the market, and most such companies respond by optimising the things that are no longer the problem, shipping features and shaving costs while sales stays stuck. Goldratt's marketing method is, in the end, a single instruction for that moment: the constraint has moved, so move your thinking to where it went.

Trying it on your own market

The method compresses into a sequence you can start this week, though not, as we will admit in a moment, finish this week.

Begin with a sheet of paper and write down three undesirable effects your customers suffer that connect in some way to what you and your competitors sell. The discipline is in the pronoun: their miseries, in their operations and their lives, and not your miseries in dealing with them, which is what everyone writes first. Slow payers is your complaint; a cash squeeze that makes them slow payers is theirs. Goldratt was strict about the number for a reason. If you cannot produce three, then either you do not know your market, in which case go and visit it rather than surveying it, or your product's real value lands one link further down the chain, with your customer's customer, and the offer you eventually build will be the one that helps your customer sell more. He was equally blunt about the special case where no undesirable effects exist at all, which he met most often in startups: a beautiful solution for no problem is better discovered before the company is founded than after it folds.

The route from those three effects down to the guilty policies has intermediate steps, and the workshop was specific about them, because each one guards against a particular way of fooling yourself. Take each undesirable effect and write out the dilemma that produces it, the conflict the customer is caught in that keeps the misery alive. Goldratt's shops were caught between buying big to survive on price and buying small to survive on cash; some such trap sits under every persistent complaint, since a problem with no dilemma underneath would already have been solved. Writing the dilemmas forces you into the customer's shoes properly, and it exposes the standard temptation, which is to declare at this point that you don't know and had better commission an expert or another survey. You do know, was Goldratt's answer. You have been in this industry for years. Force yourself to sit and write, and check that what you wrote sounds true.

With three dilemmas on paper, generalise them into one, the generic conflict your whole market is stuck in. Then hunt for the assumptions holding that conflict together, and here comes the constraint that gives the method its commercial teeth. You are not looking for just any assumption, because many assumptions, once surfaced, could be broken by the customer changing his own behaviour, which would solve his problem and sell you nothing. You are looking specifically for assumptions that are stated policies of yours and your competitors, the volume discounts and payment terms and no-consignment rules of your own industry, because a conflict resting on your policies is a conflict you can dissolve, profitably, by yourself. Goldratt admitted this restriction is mentally hard to hold; it also does the aiming. Connect policies to effects with rigorous cause and effect and keep the whole map to two pages, in large letters. If you need more than two pages, he warned, the connections are not good enough yet. Read it to a friendly customer as if-then, correct what he corrects, and repeat with another until one of them asks to keep a copy. Invert the guilty policies and draft the offer, holding it to the three rules: no price cut anywhere in it, the core problem solved, at least one rooted policy broken. Then circulate it internally and harvest the yes-buts, treating each objector as the co-author of a stronger offer, until the negatives have been trimmed and the offer is finished. And once it is, ask the last question: whether the needs in your market are spread widely enough that the same product should be sold by use, or by capability, at honestly different prices, without adding one item to your internal complexity.

Honesty about the price of all this kept Goldratt's version of the pitch from sounding like one. The analysis takes a few days for novices, the full offer considerably longer, and the reason most companies will never do it is that it demands the one resource they refuse to spend, which is sustained thinking. Nor does the method hand you a template, and he was cheerful about the contrast with his production work, where every analysis converges on the same drum-buffer-rope answer. Of the seven hundred marketing implementations he was personally aware of at the time, what came out was seven hundred different solutions. That, he suspected, is simply the nature of marketing: the method is general, and every offer it produces is yours alone.

Which leaves his closing challenge, and it deserves to land the way he aimed it. You can keep complaining that you don't have enough sales. You can keep complaining that the competition is unfair. Or you can sit down, map the damage your industry's policies do to your own customers, and build the offer they can't refuse. Everyone he had watched actually attempt it had succeeded. The ball, he told his room, is in your court.

This article draws on recordings of Eliyahu Goldratt's marketing workshop. The same ideas, the market constraint, the unrefusable offer, segmenting the market rather than the resources, run through his business novels The Goal and its sequel It's Not Luck, which remain the most enjoyable way into his thinking.