The Value Network: Who a Business Depends On

A value network is the set of outside parties a business chose and pays: suppliers, contractors, distributors, the provider that moves its money. Two separate things decide how much any one of them matters. How much of the business's flow runs through that party, and how long replacing it would take. The two properties do not move together, and the second is the one readers skip.

The two properties come apart for a reason worth stating plainly. Size is measured by looking backwards at what was already paid, and a payment record is easy to pull. Replacement time is measured by imagining a stoppage that has not happened, and nobody keeps a report of it. So one of the two properties is sitting in an accounts system waiting to be sorted, and the other has to be gone out and asked about, supplier by supplier. Anjani Stationers Private Limited, an invented printer, and Setu Bazaar, an invented marketplace, carry the illustration below.

What is a value network?

The money going out comes first. Anjani Stationers makes printed stationery and files, and every month it pays a handful of outside parties. A mill sends it paper. A supplier sends ink and coating. A contractor moves the finished cartons. An engineering firm services the folding and gluing machine. A binding workshop takes the overflow when an order is too large for the floor. Anjani picked each of those parties, negotiated with each of them, and can stop paying any of them. Together those parties form the value network: the outside parties a business depends on in order to deliver the thing it sells.

Two words in that sentence do the work, and they are chosen and paid. A counterpartyAny outside party on the other side of an arrangement from the business under examination. The word is neutral: a buyer, a supplier, a lender and a service contractor are all counterparties. that a business selected, and pays, has a name, an invoice, a phone number and a contract termThe length of time an arrangement runs before it has to be renewed or renegotiated, and the notice either side must give to end it early.. A name, an invoice and a term are what make a value network mappable in a way an ecosystem never is. Every party in it leaves a trail in the accounts.

The ecosystem is the other picture. An ecosystem is everything around the business that shapes its options without being chosen or paid by it: the technical institute two streets away that turns out machine operators, the road that was widened last year, the stationery habits of the schools in the district, the other printers whose price lists customers quote back. Anjani did not select any of that and does not pay for it. Value moves through it anyway.

Michael Porter's value chain, set out in Competitive Advantage in 1985, describes the sequence of activities that happen inside one firm's own walls, from goods coming in to the finished thing going out. A value network is the other half of that picture: the parties outside the walls that the firm chose and pays. C. B. Stabell and O. D. Fjeldstad, writing in the Strategic Management Journal in 1998, treated the network as one whole way of arranging value rather than a footnote to the chain, and that is the sense used here.

The same outside party can sit in either picture, and a contract moves it across. Take the technical institute. For years it trains operators, some of whom knock on Anjani's door looking for work, and Anjani takes the ones it likes. Nothing was agreed and nothing was paid, so the institute is part of the surroundings. Then Anjani signs an arrangement to sponsor six operators a year, pays a fee per head, and specifies which machines they must be trained on. Nothing about the institute changed. Anjani now chooses and pays, so the same building has moved inside the network and can be mapped like any supplier.

What separates a dependency from an ordinary relationship?

Here is where a lot of mapping work goes wrong before it starts. Every party in the value network is a relationship. Only some of them are dependencies. Every dependency is a relationship, and most relationships are not dependencies. The difference is not size, not how long the arrangement has run, and not how friendly the last meeting was. The difference is what happens on the day that party stops.

Ask the question in its blunt form. If this counterparty told Anjani tomorrow morning that it was finished, how long before Anjani is producing normally again? If the answer is a few days of nuisance and some phone calls, that party is a relationship. If the answer is that a machine sits idle for months and orders go unmet, that party is a dependency. The test is the recovery, not the invoice.

Feel it at household scale. The same test works there. A household buys rice from a shop at the end of the lane, and has done for eleven years. The shop is a real relationship and the shopkeeper knows every member by name. If that shop shuts on Monday, the household buys rice from the next lane on Tuesday and the whole thing costs an hour of irritation. Now take the one bank account the salary is credited into. The bank is a colder relationship, nobody is on first name terms, and it may cost less in fees than the rice does in a week. But if that account is frozen on Monday, the rent, the school fee and the loan instalment all fail together, and getting a new arrangement running takes weeks of paperwork. The warm relationship is not the dependency and the cold one is.

How much of the business runs through this party?

The first of the two properties is called share of flow. Share of flow starts with a flow that matters, and asks what proportion of that flow passes through the counterparty in question. A party can be tiny in one flow and enormous in another, so three flows are worth measuring separately. The inputAnything bought in from outside that goes into what the business makes or sells: raw material, a bought in part, or a service such as transport or servicing. flow is what the business buys. The sales flow is what the business sells and through whom. The cash flow is the money itself, moving in and out.

Share of flow is easy, and the ease is exactly why everybody starts here and many people stop here. The numbers are already sitting in the ledger. Sorting the outside payments by size and expressing each as a proportion of the total produces a ranked list in an afternoon. Here is Anjani's. Anjani reports revenue of Rs 2,70,00,000/- and earnings before interest and tax (EBIT) of Rs 41,50,000/-, and none of the shares below is a proportion of either of those figures. The shares are proportions of what Anjani pays out to the six parties named, and they total one hundred by construction.

Read that table the way most people read it and the paper mill is the headline. The mill takes 46.0 per cent of everything Anjani pays outside. The next two add up to 31.0 per cent between them, 17.0 and 14.0, so the mill is larger than both together. The mill is the largest relationship Anjani has by a wide margin, and there is nothing wrong with saying so. Share of flow is half the picture, so stopping there is the mistake. On its own the share cannot show what would happen if any of these parties walked away.

One warning about the word concentrationHow much of a total sits with a small number of parties. Concentrated sales mean most of the revenue comes from very few buyers, whatever the total., which gets used loosely here. A high share of flow through one party is concentration, and concentration is a real thing to notice. But it describes the shape of the list, not the consequence of a stoppage, and those are separate questions.

How fast could this party be replaced?

The second property is replaceability, and it is the one that does the work. Replaceability asks a single question with a number for an answer: starting from the morning this counterparty stops, how many weeks until the business is running normally on somebody else. Three things go into that number. How many alternatives exist at all. The lead timeThe gap between placing an order or starting an arrangement and actually receiving the thing. A twelve week lead time means twelve weeks of waiting after the decision has been made. before a new party can actually start. And the switching costEverything that has to be spent or redone in order to move from one party to another: retooling, retraining, re-testing, re-certifying, and the work that goes into a new arrangement. of retooling, retraining and re-testing once it does.

Run those three questions over Anjani's six. The paper mill is enormous and utterly replaceable: nine mills within reach make the same weight and finish, quotes come back in a day, and the first delivery lands in about two weeks. The transport contractor is even easier, with twelve options and a week to switch. The machine servicing contractor is the opposite in every respect. The servicing contractor is the only firm within reach certified on that folding and gluing machine, spares come from the machine maker on a long lead, and an engineer has to be trained on that specific model before touching it. If the contractor stops, Anjani is looking at roughly twenty six weeks, or half a year.

A small supplier with no substitute is a larger dependency than a large one with ten substitutes. The servicing contractor takes 4.0 per cent of Anjani's outside payments, which is 11.5 times less than the paper mill's 46.0 per cent. 26 weeks against 2 weeks is a ratio of thirteen, so the contractor also takes 13 times as long to replace. The two ratios point in opposite directions, and only one of them describes what happens when the phone call comes.

Share of flow and replacement time are independent properties. Knowing one reveals nothing about the other. They are not two ways of measuring the same thing and they are not a cross-check on each other. A party can be large and instantly replaceable, small and impossible to replace, large and impossible, or small and easy. All four combinations occur, and any of them can turn up in a business under examination. A list ranked by size is therefore not a ranking of dependencies. The list ranks one property, presented as though it settled both.

Why is the largest supplier not always the biggest dependency?

Because the two rankings are built from two different columns, and there is no rule forcing them to agree. Sort Anjani's six by share of outside payments and the order runs paper mill, ink and coating, transport, board and packaging, binding, machine servicing. Sort exactly the same six by replacement time and the order runs machine servicing, binding, ink and coating, board and packaging, paper mill, transport. Five of the six positions change. The party at the top of the first list falls to fifth on the second, and the party at the bottom of the first list rises to first.

Only one party holds still. The board and packaging supplier sits fourth on both. An illustration showing nothing but disagreement would teach a rule that does not hold, so the match is worth pointing at. The two rankings are not opposites and they are not required to differ. The two rankings are simply built from different information, so sometimes a party lands in the same place on both and sometimes it does not. Assuming they agree is the one move that is never safe.

Setu Bazaar shows the same finding in a sharper form, and its shape is different enough to be worth walking through. Setu Bazaar is a marketplace where buyers and sellers meet. Goods worth Rs 5,00,00,00,000/- pass across it in a year, its gross merchandise valueEverything buyers spent across a marketplace over a year, added up. Gross merchandise value is not the marketplace's own earnings, and on a platform the two are usually far apart.. Setu keeps 4.00 per cent of that total as its own revenue, or Rs 20,00,00,000/-. The flow across the marketplace is therefore 25 times the marketplace's own revenue.

The payment provider is sixth of six on what Setu pays it, and second of six on how long it would take to replace. But there is a third column that does not fit in the table, and it is the one that matters most. Every rupee of the Rs 5,00,00,00,000/- that crosses Setu Bazaar in a year passes through that one payment provider, so a party taking 9.0 per cent of the outside payments carries 100.0 per cent of the flow. The smallest invoice and the largest dependency are the same row. Share of flow and replaceability agree for once, and both disagree with the payment record.

How is a dependency map actually built?

The procedure is short and it is deliberately in this order. First, list every outside party the business pays, taking the list from the payment records rather than from anybody's memory. Memory drops the small ones, and the small ones are the point. Second, put the share of flow beside each. The ledger already knows it. Third, put a replacement time beside each, in weeks. The ledger does not know that one, and it has to be asked about party by party. Fourth, and this is where most maps go wrong, sort by the third column rather than the second.

The map is read along the replacement axis first. A spending report is read the exact opposite way. A spending report is sorted by size because its job is to show where the money went, and that is a reasonable job for it to have. A dependency map has a different job. Its job is to show what would stop, and the column that answers that question is replacement time. Sorting a dependency map by size produces a spending report with a different title on it.

One practical note on the third step, the step people skip. A replacement time is an estimate and it will be wrong in the second decimal place. The second decimal place does not matter. The order of magnitude is what matters: a week, a month, or half a year. Getting six counterparties roughly ordered on that scale is worth far more than getting one of them precisely right, and it can be done in a morning of phone calls.

What does a dependency map not show?

Quite a lot, and being clear about the gaps is what stops the map being misused. The map does not show whether the dependency will ever bite. A single certified servicing contractor can run for fifteen years without missing a visit, and the map that flagged it would have been flagging a risk that never arrived. A map is a statement about exposure, not about what is going to happen. A map of dependency is not a forecast of disruption, and reading it as one turns a useful list into a prediction nobody made.

The map does not show what a stoppage would cost either. Twenty six weeks of a machine standing idle converts into lost orders, penalty clauses and possibly a customer who does not come back, and none of that is on the map. Working out what an interruption would cost, and deciding what to do about it, is the subject of Strategic and Business Risk.

And it does not show what a counterparty could extract because of the position it sits in. Extraction is a real and separate question, and it has a name: bargaining power. Whether Anjani's sole servicing contractor can push its rates up, and whether the field it operates in lets it, is the subject of Industry Structure and Sector Behaviour and is answered there. The dependency map stops one step earlier, at describing who depends on whom. How the goods actually move once the parties are chosen, including what to hold in stock against a lead time, belongs to Operating Model and Supply Chain.

Which single party, if it stopped tomorrow, would stop the most?

Practitioners actually ask that question, and it is worth noticing how little it resembles a spending question. The question names no amount. It sets a date. And it asks about consequence rather than size. Put it to a business owner cold and the first name out of their mouth is usually the largest invoice. The largest invoice is the party an owner thinks about most. Ask them to sit with it for two minutes and the name very often changes.

A lender uses it before extending a working capital line. A borrower whose production can be halted for half a year by one small contractor has a repayment profile that the profit and loss statement does not show. An analyst uses it when a margin looks stable and wants to know what could break the stability. An investor looking at a business uses it to ask what has to keep going for the model to work at all. And a household uses the same question without calling it anything: which one arrangement, if it failed on Monday, would take the longest to put right. In every one of those cases the answer is very often not the party with the largest invoice, and the mismatch is exactly why the question has to be asked separately.