The advice industry has settled on one instruction about price: raise it. That advice is usually right and it is completely useless in the situation most people actually face, which is a good customer, a soft quarter, and a request for a discount that you have about four minutes to answer.
Discounting is not a character flaw. It is a tool with a very unforgiving failure mode: the damage compounds silently, arrives one renewal later, and by the time you see it in the numbers it is written into every contract you signed in between. This playbook is the arithmetic, the four discount structures that survive contact with reality, the three that quietly kill margin, the scripts, and the way out of a discount you already gave.
Step 1: Do the volume math before you do anything else
The single number that matters is how much extra volume a discount requires just to stand still. It is not intuitive and it is not small.
The formula: required volume increase = discount % ÷ (gross margin % − discount %).
Worked example. You sell a service at $1,000 with a 50% gross margin, so $500 of contribution per sale. You offer 20% off. The new price is $800 and the cost is unchanged at $500, so contribution falls to $300. To make the same total contribution you now need 20 ÷ (50 − 20) = 67% more sales.
Run that across a few margin levels and the picture gets stark:
- 50% margin, 10% off → need 25% more volume.
- 50% margin, 20% off → need 67% more volume.
- 30% margin, 10% off → need 50% more volume.
- 30% margin, 20% off → need 200% more volume.
- 20% margin, 15% off → need 300% more volume.
Write your own version of that table on one page and keep it where you take calls. Most discount requests die on contact with it. If a 20% concession requires you to triple the account, and the account is not going to triple, you are not negotiating — you are agreeing to earn less for the same work.
The gate: if the required volume increase is not plausibly achievable from this specific customer, the discount must buy something other than volume. Which brings us to the only defensible reason to give one.
Step 2: A discount must buy something. Name the something.
A price cut given for nothing teaches the buyer that your price is fiction. A price cut given in exchange for a concrete concession is a trade, and trades are repeatable without damage.
Four things worth trading for, in rough order of value:
- Term. Twelve months prepaid, or a 24-month commitment. You are buying certainty and cash, and both are worth real money.
- Cash timing. Payment up front instead of net-60. A 3–5% discount for prepayment is cheaper than a line of credit and improves the thing that actually kills small businesses.
- Scope reduction. The cleanest trade of all. Fewer deliverables, fewer revisions, standard onboarding instead of white-glove. The price falls because the work fell.
- Proof and access. A named case study, a logo, a reference call, a recorded testimonial, three warm introductions. Cheap for them, genuinely valuable to you, and unlike price it does not recur.
Three structures to refuse:
- The permanent across-the-board cut disguised as a promotion. If it never ends, it is not a discount, it is your new price and you have not updated your model to match it.
- The loyalty discount for existing customers. You are paying your most committed buyers to keep doing what they were already doing, and every one of them tells the others. Give existing customers more value, not less price.
- The match-the-competitor cut. You are letting a competitor with an unknown cost base set your margin. If they are cheaper because they are worse, sell that. If they are cheaper because they are structurally more efficient, price competition is a fight you will lose slowly.
Step 3: Discount the structure, not the number
When you cut the headline rate, you set a new anchor and it is very hard to move it back. When you change the structure, the rate survives intact.
Five structures that preserve the anchor:
- The ramp. Month one at 50%, month two at 75%, full price from month three. Total year-one discount: about 6%. Feels like a large gesture.
- The unbundle. Remove the two deliverables they use least and drop the price by exactly their listed value. The rate per unit of work does not move.
- The volume tier. A published table where the lower price is earned at a stated quantity. Anyone can have it — at that quantity.
- The prepay. Pay for twelve, get eleven. That is 8.3% off and it lands as cash today.
- The pilot. A short paid engagement at a small fixed fee with a defined end date and an agreed full-price conversion. The discount has an expiry written into the contract, which is the part that matters.
Step 4: The scripts
Most discounts are given because nobody had a sentence ready. Have the sentences ready.
When asked flatly for a lower price: “I can get you to that number. To do it I need to take out X and Y — those are what the difference pays for. Do you want the version without them, or the full scope at the original figure?”
When they cite a cheaper competitor: “That is a real quote and it is a different scope. Here is what is in mine that is not in theirs. If those three things do not matter for this project, take theirs — genuinely. If they do, this is what they cost.”
When budget is the stated blocker: “What is the number that gets approved? … Fine. I can build something real for that. It will not be this. Let me show you what it is.”
When you decide to trade: “I can do 12% — on a twelve-month term, invoiced annually up front. Standard terms stay at list. Which do you want?”
When you are holding firm: “This is the price. I would rather lose the project than deliver it badly at a number that forces me to cut corners on you.” Then stop talking. The silence after that sentence is doing the work; filling it is how the discount gets given.
Step 5: Put an expiry date on everything
Every discount you grant gets three things written into the agreement, without exception:
- An end date. Not “introductory pricing” — a date.
- The stated reason. One line: “12% reflects a 24-month commitment.” It makes removal automatic when the reason expires.
- The list price, printed. Show the full rate with the discount as a visible line item. A buyer who never sees the real number cannot be moved back to it.
Then diarise the renewal conversation for 60 days before the end date, not 7. Discounts get extended because someone noticed too late to have the conversation properly.
The exit ramp: unwinding a discount you already gave
This is the situation most people are actually in. A twelve-month sequence that works:
- Day 1 — measure. List every discounted account, the effective rate, and contribution margin per account. Some will be negative. You need to know which.
- Day 30 — stop the bleeding. New customers only at list, no exceptions, starting today. Nothing else works until this holds.
- Day 60 — add value first. Ship something real to the discounted accounts. You are building the case you will make in ninety days.
- Day 90 — the reset conversation. “Your rate was set when we were smaller. From [date] it moves to $X, which is still under list. Here is what has been added since we started.” Give 90 days’ notice. Never do this by email alone.
- Day 180 — step, do not jump. Move in two increments if the gap is more than 15%. A cliff triggers churn that a staircase does not.
- Day 365 — accept the losses. Some accounts leave. If your lowest-margin customers churn and revenue falls while profit rises, the exercise worked. Judge it on contribution, not on logo count.
Five failure modes, named
- Discounting to close a bad-fit customer. Price was never the objection. You have now bought a support burden at a reduced rate.
- The salesperson’s standing authority. If anyone can approve 15% unilaterally, 15% is your price. Set approval thresholds and require a written trade above them.
- Discounting the wrong line. Cutting the high-margin item to protect the low-margin one. Know which is which before the call, not during it.
- The perpetual promotion. Four sales a year is a sale. Twelve is a price list, and customers learn to wait for it.
- No floor. Every product needs a walk-away number computed from fully loaded cost, written down before negotiation starts. A floor decided during a call is not a floor.
The one-page version
- Compute the required volume increase. Usually that ends it.
- If you proceed, the discount buys term, cash, reduced scope or proof. Never nothing.
- Change the structure, not the rate. Ramp, unbundle, tier, prepay, pilot.
- Every discount carries an end date, a stated reason, and the visible list price.
- Know your floor before the call. Say the sentence. Then stop talking.
The goal is not to never discount. It is to never discount by accident — and to be able to say, of every reduced price on your books, exactly what it bought and when it ends.