How Multi-Branch Distributors Can Fix Inconsistent Customer Discounts

White paper · Optimise · 14 min read

Sources: Pricing Insight engagement analyses, anonymised where the client has not approved publication; the Pricing Periodic Table, CXO Edition; Tradelink case study, published. Identified and modelled figures are not realised results.

Three prices for one product code. A builder who buys across state lines rings the Sydney branch and is quoted one price, rings Melbourne and is quoted a second, and finds a third on an old special that nobody at either branch knew was still live. Most multi-branch distributors we have worked with have some version of this, and the first instinct is always to blame the branches. The data says otherwise. Branch managers set the prices they were given the authority to set, inside a structure that gave them no other way to win the order.

Inconsistent branch discounts are an architecture problem and an authority problem. Tiers were built without spend thresholds, so very different accounts sit on the same tier and the tier means nothing. Special prices were loaded without an expiry date or an owner, so a deal struck for one job in 2019 is still the price in 2026. Discount authority sits at the counter, so the person facing the customer is also the person deciding what the business will accept. Fix those three things and the branches fall into line, because the system no longer lets them drift. This paper sets out how, using the twelve elements of the Pricing Periodic Table that decide the problem, and what the evidence from distributors with 20 to more than 200 branches says it is worth.

01How branches drift apart

Nobody decides to have inconsistent prices. Many networks arrive there by acquisition. A distributor buys a competitor in another state, keeps its branches, its people and its customers, and keeps its price file because migrating it would upset the customers it just paid for. Three acquisitions later there are four price lists, four discount habits and a head-office list that nobody in the field quotes from. At a building products distributor selling more than 25,000 products through branches, resellers and cut-to-size services, reference prices, overrides, project quotes, rebates, credits and branch variations each weakened the final invoice price in a different way.

Inheritance is the start. Habit does the rest. A branch manager under pressure to hold volume gives a builder a special price on framing pine, loads it against the account, and it works. The special is never reviewed, because reviewing it means a difficult conversation and the system does not prompt one. Across a network those decisions accumulate: at one cabinet and architectural hardware distributor, roughly 384,000 customer and product combinations each carried their own negotiated price. The tiers existed on paper. In practice they were bypassed.

Customers finish the job. A builder who buys in two states compares the two invoices. A national account expects one price and gets three. A trade customer learns which branch gives the better deal and buys there, and the lowest branch price becomes the national price, one account at a time. In one timber group, one branch needed only a 16 per cent margin on framing pine while another in the same state needed 32 per cent, and the same builder could buy from both a day apart. Neither branch was wrong by its own rules. The group had no rule that covered both.

02What the data shows when the branches are lined up

Twelve months of invoices, one question: what did the same type of customer pay for the same item, branch by branch and seller by seller? The answer is wider than management expects. At the hardware distributor, discounts ranged from about 1 per cent to 59 per cent with no consistent link to customer size, and the same item sold at prices up to 60 per cent apart between comparable customers. Half the range traded within a 21 per cent spread. Nearly one product in five traded at prices more than 40 per cent apart, and the widest items approached 100 per cent.

Two tests separate earned differences from unearned ones. The first fits a price-volume relationship for every product; at the hardware distributor, volume explained about a tenth of the price differences within a product, and the other nine tenths was negotiation. The second compares customer margin with customer size; size accounted for about a quarter of the variance, and two customers of the same scale routinely sat 10 or more margin points apart. A business that believes its discounts reward volume should expect both numbers to be high.

Branch patterns sit in the same data. At the building products distributor, discovery found an account spending about A$2,000 a year priced below an account spending about A$100,000 on the same products, and one branch accounting for about half of all price overrides in the network. At Tradelink, the published case study records counter overrides, below-cost lines and inconsistent quote rules across more than 200 branches facing local competition, with no national rule that told them how to respond to it. None of those symptoms is cured by telling the branch to stop.

03Three mechanisms, none of them at the branch

Tiers without thresholds. A discount tier is only a tier if there is a rule for getting onto it and a rule for falling off it. Most distributor price files have tiers named for customer types, with no spend threshold attached to any of them, so a branch manager puts a new account on the tier that wins the order and nothing in the system objects. In one five-tier file, representatives parked most new customers on Tier 2, and an account spending A$500 a year sat on the same tier as one spending A$1.2m.

Special prices without expiry. A special price is a decision about one job, one period and one customer. Loaded without an end date, it becomes a standing entitlement and a precedent other customers ask for. At the hardware distributor, a special price agreed for one project stayed live after the job finished, was used on later orders by the same customer and sometimes by others, and missed the annual increase. Over time the pool of special prices became the real price list for some accounts.

Authority at the counter. When the person facing the customer can change the price without a second person seeing it, the price is set by whoever is under the most pressure that day. At the hardware distributor, territory managers could discount down to a single margin floor with few guardrails above it, and most of what happened between list price and that floor was decided in the car park. At a New Zealand rural supplies co-operative, branch staff overrode prices at the counter so often they had a name for it: the good bugger button. Rebates compound it: head office agrees a rebate with a buying group, branch staff cannot see it and discount again to win the work, and at one steel roofing business the company paid twice for the same sale.

One uncontrolled price is enough. A particleboard flooring maker with a 10 to 20 per cent premium over its main rival and about 60 per cent of the market let one account manager give one independent hardware store a low price. The store advertised it in letterbox campaigns, a national hardware chain cut its shelf price on the standard sheet by 15 per cent, from A$42.24 to A$36.00, in every store in the country, and wholesalers followed. Over the next two months the maker lost about A$2.0 million of operating profit against extra sales of less than A$50,000, a realised loss. The account manager was not the cause. Nothing in the system stood between the floor and the invoice.

04The fix in twelve elements

The Pricing Periodic Table sets out 61 elements of pricing capability in nine domains, across Value, People and Structures. Twelve decide the branch discount problem, in three groups: the elements that define what the price is, the elements that define who may move it, and the elements that make every movement visible. A distributor that builds the first group without the second gets a clean price file that the branches work around. One that builds the second without the third never finds out.

05Structure: one list, one discount, one ceiling

Every price is one national list price less a single discount that deepens with customer size, and no customer pays more than the published trade-counter price. That sentence is the whole architecture, and the hardware distributor adopted it to replace discounts accumulated deal by deal over many years. The list price for each product is derived from the market rather than from cost: at that distributor, 1.5 times the median price paid by accounts spending more than A$25,000 a year. A flat category discount sets the walk-up price, which is the ceiling and the published price. Named accounts hold discounts that deepen with annual spend, with deeper levels opening only above the A$5,000 and A$25,000 thresholds. Discounts never stack. For each order line the system applies exactly one discount.

Branches keep what is real. Freight to a remote site, a competitor with a yard across the road, a service level the customer pays for: those are legitimate reasons for one branch to sit at a different point from another, and the corridor allows for them. The corridor is a band around the structural price, set by product group on the evidence of how widely that group already trades: narrow for price-shopped groups, where the market enforces tightness and key items are fenced from discounting, wider for groups where volume genuinely earns the spread, published as volume tiers so the width is earned rather than negotiated. Anything outside the band is an exception, and an exception has an owner, a reason code and an expiry date.

What this does to the spread can be measured before any price changes. In one anonymised trade distributor, about 220 customers paid anywhere from about A$51 to A$103 for the same item. Under a single national structure the projected range narrowed to about A$55 to A$86, with the median unchanged at about A$76. The median holding still matters: the structure is not a disguised price rise but a redistribution of the same revenue, so that the price each customer pays can be explained by the customer's size and terms.

Price paid for one item across about 220 customers at an anonymised trade distributor. The after range is projected under a single national structure, not realised.

True net price belongs in the structure because a branch cannot hold a price it cannot see. Where a rebate, a settlement discount or free freight sits behind the invoice, the quoting screen shows the net position and the discount the branch may offer is set against that, not against a list price nobody pays. That is what stops the steel roofing problem of paying twice.

06Authority: who may move a price

Four levels. The higher the level, the fewer the prices and the greater the scrutiny. A territory or branch manager prices within a small band around the set price without asking anyone. Above that, a pricing approver at the quote desk handles approvals within a tolerance, and the quote desk also prices complex and multi-branch jobs centrally. The head of sales handles approvals within a wider tolerance. The Pricing Council handles structure, tolerances and any rebate. Each step carries a response-time target, because a slow rule is routed around. At the hardware distributor the recommended design gave territory managers about 1 per cent of room around the set price, and the team's own line became the design brief: the pathway must be faster than working around it.

The approval ladder recommended at the hardware distributor. Each step has a response-time target; monthly meetings during rollout, quarterly after. Recommended, not yet a reported result.

Floors are enforced in the system, not in a policy document. With manual order-line discounting removed, a below-floor price cannot be keyed; it has to be approved into a record with an owner and an end date. The floors at the hardware distributor were plain: no price below its current level in year one bar the deliberate ceiling corrections, no discount beyond the customer's approved rate without an approved, expiring exception, and a quote floor and tolerance on quote-desk deals. Together those replaced the single informal 26 per cent margin floor that representatives had been pricing down to for years. The floor had become the target, which is what authority at the counter always produces.

Price matching gets its own rule because it is the most common route around every other rule. A match is given only on a sighted competitor quote or invoice, and only with something in return: volume, a longer term, faster payment or the rest of the basket. A match, offer or walk grid by product group tells branch staff which response fits which product. The branch manager gains something in return for all of this: a customer price list that explains every relativity, and an answer to the builder who asks why the branch down the road is cheaper, because under the structure it no longer is.

07Visibility: the override report

Two measures, reported monthly. The first is the revenue and gross margin dollars transacted under exception, which should shrink on a schedule as held prices migrate onto their structural discounts. The second is the gross margin dollars conceded, the difference between the structural price and the exception price at the volume sold, which should stay small, be priced knowingly and reach the council as a dollar figure. Both are cut by person, product and customer and reported against a tolerance the council sets. The distinction matters because a business can reduce the number of exceptions while conceding more margin on each one.

Four price realisation KPIs sit alongside, each defined once and reported the same way every month on a fixed set of customers and products: price realisation (pocket price achieved divided by target price), the list-to-pocket difference in dollars and basis points of revenue, override share (revenue sold through overrides and exceptions as a share of all revenue) and increase realisation (increase achieved divided by increase planned). Each is measured like for like so volume and mix do not distort it, reconciled to reported revenue and gross margin so finance trusts it, and for a branch network cut by branch. At the hardware distributor the structure was modelled to bring price dispersion from 26.4 per cent to 22.5 per cent in year one. A branch whose dispersion is not narrowing is working around the structure, and the report shows which products and which people before the quarter's margin does.

Commercial numeracy makes the other eleven elements stick. Authority is safely given to a branch manager who can read a margin and withheld from one who cannot; the co-operative's recommended rule that tied discount authority to demonstrated understanding of mark-up and margin is the right principle, even without a reported result yet. Incentives follow the same logic. A branch paid on sales will discount to the floor. A branch paid on gross margin dollars against a structural baseline has a reason to hold.

08Moving the accounts, and what it is worth

Nobody moves to the new structure in one step. Bringing branches into line means moving customers to new tiers one account at a time and managing the conversations that follow, and the design decision that makes it survivable is a cap on how far any account moves. The hardware distributor ran two versions of its first-year strategy. The early version let any line move by as much as 15 per cent, and about 30 per cent of walk-up customers would have seen a decrease. The adopted version capped any line at about 7.5 per cent in the first year, gave named accounts increases only, set the walk-up increase at about 9 per cent and left the typical line moving about 0.1 per cent. Seven in ten customers were projected to see a net rise of about 1.6 to 2.0 per cent. The team's summary was three words long: prices move gently. The gentle version gives up some first-year margin to win acceptance, and that is a CEO decision.

Each branch manager receives the accounts to move and the conversation to have. Walk-up customers see a published price notice at the counter and online; named accounts receive a letter with their customer price list, issued with the annual increase. The sales team gets the rule book, the price lists, the quote desk and the exception process before go-live, because the interviews at the hardware distributor were unambiguous: this change succeeds or fails with the sales team, and it has to be carried by proper tools rather than by instruction.

What it is worth depends on the book, and each figure below carries its status. At the hardware distributor, the adopted structure was modelled to capture approximately A$4.00m of gross margin in year one, 2.0 per cent of revenue, lifting gross margin from 33.6 per cent to 34.9 per cent through capped price movements alone; the fair-value analysis separately identified A$4.16m of pricing sitting more than 10 per cent below fair value after volume is credited, a reading that overlaps the structural target rather than adding to it. At Tradelink, the published case study records A$10.8m of annualised margin opportunity identified across more than 200 branches, with a 40 to 60 per cent reduction in exceptions projected; the realised reconciliation is not public. At Stramit, where everyone priced differently, a diagnostic, playbooks, override limits and training identified an A$4m to A$7.5m annual EBIT opportunity. Super Retail Group reduced ad hoc discounting by up to half across four retail brands, a realised result. None of these is a benchmark for another business. Each is what one book of transactions showed when the branches were lined up.

09The sequence

Diagnose before designing. The first piece of work is the twelve-month invoice comparison, branch by branch and seller by seller, with the price-volume and size-margin tests that separate earned differences from unearned ones, a count of every special price without an end date, and a list of every way a price can be set or changed without approval, including direct loads and counter overrides. Finance agrees the baseline before anything moves, so that every later figure reconciles to it.

Design on the evidence. The national list, the tiers and their thresholds, the corridor widths by product group, the fenced items, the floors and the four authority levels are set from the dispersion analysis, not from a template. The council is constituted and owns the rule book from this point.

Pilot on a category, then roll out by band. One or two product categories with a clear hierarchy go live first, with the ceiling validated in the system, the special-price data corrected and the override report running from the first day. The remaining categories follow in sequence, accounts move inside the cap and dormant specials expire on a schedule. The annual refresh re-tiers every account on the last twelve months of spend and retrieves the next tranche of the structural target. From then on the branches are consistent because the system keeps them consistent, and the council's review of the override report is the standing check that it stays that way.

10Frequently asked questions

11How Pricing Insight helps

The starting point is the invoice comparison across branches, not a new price list. Our Strategic EBIT Opportunity & Risk Diagnostic lines the branches up on twelve months of transactions, tests whether discounts are earned, counts the specials and the override routes, and sizes the opportunity against a baseline finance accepts. The work then runs in the three stages every Pricing Insight engagement follows.

The hardware distributor, the building products distributor, the timber group, the steel roofing business, the particleboard flooring maker and the rural supplies co-operative in this paper are Pricing Insight clients described without identification. Their figures are modelled, identified or recommended unless stated as realised, and none should be read as typical or guaranteed. Tradelink, Stramit and Super Retail Group figures are as published in Pricing Insight case studies. The 220-customer example is anonymised and its after range is projected.