The Premise: why margin falls as B2B businesses grow

the signals

Each quarter, the margin has a new explanation.

Freight, stock, disputed invoices, and added headcount get separate meetings, though they often trace back to the same orders and the same terms.

One quarter the bridge blames freight. The next, it is expedite cost or a costlier substitute. By the third, it is credit memos settling last quarter’s orders, a rebate accrual under review, or warehouse overtime for work that was never in the price. Each explanation is true in the spreadsheet. None of them traces the cost back to the orders that produced it, so the next quarter brings another.

Working capital grows faster than sales. Safety stock sits staged against supplier dates nobody trusts. Material bought for a project that slipped sits with no cancellation charge behind it. Product made or bought to one customer’s spec waits on a release the customer has not scheduled, invisible to the P&L until it ages into the obsolescence reserve. DSO stretches as disputed lines and short-paid invoices wait in the queue.

Headcount grows where orders get fixed by hand: credit clerks working short-pays through customer portals, a coordinator keeping offline price files current, an expediter running hot transfers between warehouses, accountants cleaning up the close.

Revenue is rising. Customers are reordering. The board sees a company still moving, and while the top line holds, the margin reads as a bad quarter: freight, the market, a supplier increase.

More of the work behind growth now runs on inventory, exception handling, credit cleanup, and senior time.

The senior team feels it before the reports connect it.

the rules

Built for a smaller, simpler version of itself.

Each shortcut cost less than the fix when it was taken. Most of them are still in use.

The business was not built wrong. It was built around fewer customers, products, suppliers, and locations, and for that company, the price file, the lead-time fields, the customer rules, the freight assumptions, and the billing logic were enough.

Then it kept growing, and each change was cheaper to handle by hand than to build in. When a good customer needed an exception, the rep honored it, and it became precedent: that account’s normal treatment. When a supplier’s lead time slipped, the buyer padded the order and left the old number in the file. An override that saved one renewal was easier to reuse than to replace. A note to billing took a minute; the customer record had no field for a promise like free freight on one product line to job sites, and adding one meant a ticket to the ERP consultant and a test cycle nobody had time for. The acquired business kept its own account habits and supplier terms because the order desk could work around them.

Experienced people absorbed it in the course of the day: the senior CSR fixing an account’s price before release, the warehouse manager walking the row before promising availability, the pricing analyst keeping the spreadsheet that sat beside the price file, the controller applying a credit memo by hand to the right invoice.

It worked because there were two cushions: gross margin wide enough to pay for the extra handling, and enough slack in the day for people to do it. Nothing forced the question. Growth can thin both at once. Larger accounts come on at tighter prices, and more orders use up the spare hours, so the judgment that once closed an occasional gap becomes a requirement for routine business.

The ERP holds the terms the business started with. The business runs on the ones it granted since.

the promise

The order is booked before the business can keep it.

The price, date, freight assumption, spec, and customer treatment are set before the rest of the chain has to deliver on them.

An order is never just demand. By the time it is booked, it carries a full set of promises: the price the rep quoted, the date the customer was promised, the freight built into the price, the spec accepted, the terms attached, and often a documentation requirement, a packaging condition, a ship-complete rule, or a service obligation the business has taken on for that account.

It is accepted against what the system holds at that moment. The price comes from the file as loaded, plus the overrides learned over time. The screen shows on-hand stock that is reserved for no one until the pick ticket prints. The lead time is a default that may no longer hold, the ship-from warehouse is the one the freight was priced around, and the customer’s treatment may sit in order notes or an old renewal rather than where order release can use it.

Nothing has gone wrong yet. The order looks clean at entry: priced, dated, accepted, complete enough to move. Then the rest of the business has to deliver it.

To keep the date, purchasing pays for a spot buy or an expedite from the supplier. The stock the screen showed sits in two buildings. The CSR overrides the ship-complete hold so the order ships on time, and logistics pays a minimum charge on each leg to a dock that asked for one delivery. A packaging rule the pick ticket does not carry is caught at the dock and reworked, or it reaches the customer’s receiving and the pallet comes back. The invoice goes out at the original price; where the pallet came back, the customer deducts it.

The order is not wrong.

Every field on it was valid when it was entered. The rep quoted from the price file; the CSR released what the screen allowed. What no screen showed was what the order would cost to keep.

It still ships, and the customer may still pay. But the margin goes to the spot buy, the second freight leg, and the hours spent correcting the invoice, and the cash waits in disputed balances until the correction clears.

What one order absorbs becomes, across the order book, standing work for people who were hired to do something else.

the variety

Growth adds difference, not just volume.

The terms, the inventory depth, and the handling that win the account are the same things the order has to carry.

What one order absorbs, growth multiplies. More of the same orders would make the business better at carrying them: the same bins, the volume tier on the supplier rebate, consolidated freight, invoices that match the contract table. These businesses grow by adding difference instead: new products, customer-specific terms, service attached to the product, new sales channels, and acquired businesses that arrive with their own history.

The unit stays familiar while the work under it changes. A SKU stays a SKU until it carries a cut length, a frozen revision, a heat number, a country of origin the customer restricts, or an approved-manufacturer list. The ERP still sees a flange; the receiving dock sees a missing mill test report and turns the truck away. The customer with one ship-to becomes one with twenty, each with its own barcode labels, receiving window, and split-shipment rule, and an EDI map that rejects an invoice when a unit of measure or a ship-to code does not match. An added warehouse picks up local pricing nobody wrote down, its own units of measure, and supplier lines the central system never mapped.

Most orders that carry these conditions look clean. The parts ship and arrive on time. The cost turns up later, in a carrier bill for a second leg, a held invoice, or an hour at the bench nobody charged for.

Much of that variation is the reason customers buy. They buy because the stock is actually there, because it ships complete and on time, because the company works to their terms: the payment schedule, the EDI mapping, the invoice built to pass their audit. They buy the service attached to the material and the rep who knows the problem before their own buyer does.

They are paying for a business that bends so they do not have to.

That is a legitimate thing to sell, as long as the price carries the stock, the labor, and the paperwork it takes.

The cost of that service belongs in the commercial decision. When it is not there, the business wins the account at a standard price and delivers it with custom work: the warehouse kits on overtime, the back office corrects the invoices, and each new account of that kind brings its own manual handling. Revenue grows, and the cost to serve it can grow faster. That is operating leverage running backward.

The work is not to remove the complexity. It is to decide what each piece is worth and put it where the next order can find it.

the reports

The ledger records every cost, and rarely the order that caused it.

Revenue shows the sale; its economics arrive later, in freight bills, substitute costs, price credits, and balances still awaiting payment.

The effects reach the reports, split across the numbers the business already reviews. Follow one order that needed a spot buy, a split shipment, and a backordered line, and it shows up in five places, none of them labeled with the order.

Revenue reads cleanest. The order shipped and was invoiced in full. By that measure, the company is still moving.

Margin takes the hit in pieces. The spot buy lands in cost of goods, the second freight leg in freight expense, the price credit in sales allowances, the extra handling in warehouse labor. Some of the loss lands below gross margin, where the margin bridge rarely looks. A price held below the file to protect a renewal becomes the account’s new baseline. By the time the controller builds the bridge, the order that produced the difference is several steps removed.

Service metrics point outward. If the CSR moves the backordered line to a new release with its own promise date, a report measured against the latest promise shows the order on time and complete. If not, the miss registers in OTIF, backorders, and escalations, and sends attention to operations, supply, and the warehouse floor. Seeing why takes the account’s terms and the stock record read together, and the standard service reports do not put them side by side.

Cash shows it later. The customer’s AP rejects the invoice over a price or unit-of-measure mismatch. The whole invoice sits in disputed aging, the part that was right with the part that was not, while billing and the account team settle the correction. Elsewhere, buyers who no longer trust the lead-time fields pad their own reorder points, and the stock behind those numbers ties up cash the plan never assumed.

Headcount is the last place it shows. Every hire meets a visible need. Where the new jobs keep reconstructing the same instructions, the company is buying people to carry what the records and routines cannot.

The reports are built to close the books, not to price an account.

Each function’s report tells a story that points at another function. Someone has to put the order back together from the records and from the people who handled it, and establish which decisions produced the result instead of assigning it to the report where it surfaced.

the carrying cost

It becomes a ceiling before it becomes a crisis.

Paid for in the senior team’s calendar, a price file nobody can fully explain, and working capital that grows faster than revenue.

The business can keep growing while more of its capacity goes to holding the current order book together.

It starts in the systems. The ERP did not break; it bent. Custom fields on the customer and item masters hold treatments the original setup never anticipated. The price file runs on overrides layered on overrides, and the CPQ carries account exceptions whose reasons have outlived the people who entered them. Because no one fully trusts the record, the planner, the rep, and the pricing analyst each keep a spreadsheet beside it.

Desks fix what reaches them, and each fix lands on the next desk. Customer service sets the system to split backordered lines so orders ship on time, and the warehouse packs more cartons and pays more freight. The warehouse stages pre-built kits to clear the dock, and the cycle counts drift because components left their bins without leaving the system. Billing builds a special invoice format for one customer’s AP portal, and every later change to the EDI maps has to work around it.

What the desks cannot settle climbs. What a senior rep or a warehouse manager used to absorb now reaches the senior team. The COO spends the week clearing routine escalations, and the operation waits for decisions that could have been made closer to the work. The CFO’s close runs late on a margin number the team rebuilds line by line. That time is already in payroll; the work it displaces never appears in the margin bridge.

People who cannot trust the record hedge with stock, and capital carries the rest. Inventory turns fall as defensive stock builds. Receivables age as disputed lines take longer to clear. Working capital climbs faster than revenue because both ends of the cycle thicken at once.

The books can also carry income that will never arrive. Finance accrues the supplier’s volume rebate each month as a credit to cost of goods, against the forecast mix. When the spot buys that keep customer dates go to other suppliers and the volume misses the tier, the receivable will never be paid, and the year-end write-off lands in cost of goods all at once, taking back earnings booked months earlier.

On paper, all of it looks routine: support payroll, freight costs, stock, receivables, a rebate true-up.

None of it is labeled what it actually is: the cost of holding the business together one transaction at a time.

Customers do not always leave. They learn. They learn which dates hold and which slip, which invoices to dispute, which terms to tighten at the next renewal. Their AP systems may start short-paying deviations by default. A missed date that once drew a phone call can become a compliance penalty written into the renewal. The next RFQ comes back with a little less margin and a little more service expected. The customer is pricing the patterns they have learned to expect.

Suppliers draw their own conclusions. They see which forecasts hold, which POs change after they are cut, and how much attention the account takes. Discretionary pricing support gets harder to win when the actual mix drifts from the forecast, and when capacity is tight, a supplier may favor accounts with firmer schedules. The supplier base looks the same on paper while the terms behind it move.

Each local call can remain rational: the safety stock protects the date, the credit memo clears the dispute, the overnight freight saves the account. No single failure is large enough to force the question.

The appetite is still there. The capacity is not.

What is missing is not on this year’s P&L. It is the next acquisition the business could have absorbed, and the operating leverage the plan assumed.

the remedies

Every response runs into the same undecided terms.

A new leader, a new system, and a pricing council each come back to one question: what the business will promise, and what it will fund.

The senior team acts. It changes leadership, funds systems, and assigns teams to catch what keeps leaking through. Each can be the right response, and each changes the margin only if its mandate reaches the account terms underneath and someone has the hours to decide those terms and carry them into use.

A new leader moves fast where the role can enforce: territories, compensation, pricing approvals, credit discipline. A new pricing policy still needs the price file rebuilt, the exceptions decided account by account, and the customers told, while the company sells, ships, invoices, and collects every day. Sponsorship does not supply the hours.

A new system maps the business, documents the workflows, reviews master data, and designs approval paths. Then the blueprint meets the real business. The standard configuration needs units of measure that match, customer terms somebody has settled, and exception paths with named authority. Custom fields, middleware, and exception tables may be the right way to preserve what customers rely on; their value depends on whether the business has decided what it is preserving and who will maintain it.

Management still has to authorize the commercial choices: which terms survive, which stock policy supports the promise, and what changes for the customer. The integrator needs an answer; the account team still has an order to release.

Commercial operations sits closest to the decisions. A deal desk, a pricing council, forecasting routines, and exception tracking can expose the pattern and give it an owner. But the council meets weekly and the quote is due within the hour, and the analyst is measured on how fast quotes go out, not on what the margin looks like two quarters later. The side spreadsheet stays alive. A council can approve new freight terms, but someone still has to agree them with the account, change the customer file, and check that billing uses them.

In a live business, the three run into each other. The new leader announces price discipline. The integrator builds it into the new system as hard validation. The exceptions pile up at the deal desk while customers wait for confirmations, and at month-end, with the revenue target in view, the leader signs a blanket override.

The system bought to enforce the rule now runs on an override table.

The question is who can settle the terms underneath. The decisions cross functions none of them owns alone, and the people who could settle them are running today’s orders. The company’s own team can sometimes carry it. When it cannot, the decisions need a bounded assignment with the authority to settle across functions and the time to see them into use.

The next order has to run on the decision.

the decision

What has to be settled before the date.

The move puts a deadline on commitments the business has honored for years, many of them written down nowhere a new rule could read.

Growth eventually brings a move, and the move comes with a date. An acquisition, a new system’s go-live, a portal, a second site, or an owner’s plan each takes away some of the slack that has been paying for the precedents, or asks for a rule where there has been a judgment call.

The work is to establish what the move actually takes. The price file, the lead-time records, the order notes, and the credit memos show parts of the history. The CSR, the buyer, and the warehouse lead can explain why a treatment exists and what would happen if it stopped. Some treatments were never recorded, some sit in records that disagree, some are in the system with no one left who knows why, and some still work but no longer earn their cost. The answer has to stay open long enough to find the real constraint. Sometimes it is demand, sometimes a supplier or an incentive, and sometimes simply too little stock.

Where the arrangements change, the work comes down to three things the next order depends on.

What it is priced at: a price file that carries the account’s terms rather than leaving them in order notes, and item and customer masters reconciled across acquired businesses.

What it can promise: lead times and units of measure with owners, and stocking rules that match what sales is quoting.

Who can change it: substitution authority that separates an approved alternate from a commercial concession, and a named owner for every shared account.

None of this is novel. All of it is hard, because the precedents are now load-bearing. The override that protected a renewal, the side spreadsheet that keeps quotes moving, the custom field that prevents a service miss: each had a reason the day it was made, and together they hold up the business it has become. Pull one out without deciding what replaces it, and something a customer relied on fails.

So management decides, accommodation by accommodation, what the changed business will carry. Some decisions can be final before the move; others need only a treatment the next order can run on, an owner, and a date to come back to them.

Each one is kept, priced, or stopped.

Kept means funded on purpose, or delivered a cheaper way that keeps what the customer values, with its cost known and the accommodation written into the customer record and workflow.

Priced means carried into the account’s terms once the customer accepts them, or charged where the existing terms already allow it.

Stopped means ended on a date the customer agrees to, or with the notice the account’s terms allow, before the invoice changes.

The people handling the next order get terms they can use, and the cases that still need judgment get a route to someone who can give it.

That settlement, and carrying it into use, is the work I take on through Davello & Co. It is done when the next orders run on what was decided, and the people who had been absorbing the difference by hand can go back to the work they were hired to do.

Where the business already looks like this, the first conversation is about what keeps coming back: consult@davelloandco.com

Why customer exceptions become the way a growing business operates.