Discount approval gates: who signs off on the price?

Friday, 5:20 p.m. An account executive is about to send the last quote of the quarter. The buyer called that morning and said the decision would be made today, but only if the price came in well below list. The rep has a three-year relationship on one side and a quota that will not close on the other. She opens the quote, pushes the line discount to a number nobody has approved, generates the PDF and hits send. The manager who should have signed off is on a plane. On Monday morning finance opens the record and asks one question: who approved this discount? There is no answer, because nobody approved it. There was only a sender.

This article is about closing the gap between the send button and the approve button. We start with why discount authority has to be written down rather than passed around as folklore. Then we look at whether the gate should trigger on deal value or on margin, how to calculate a margin floor you can defend, how many levels an approval matrix really needs, how fast an approval has to come back, where a pending quote should sit so it does not rot in an inbox, how a rejected quote finds its way back, and where the decision itself is recorded. We close with the standard objections to approval gates and a thirty-day rollout order.

Who actually holds discount authority, and why write it down?

In most teams discount authority is unwritten. It circulates as custom: everyone knows the rep can go a little way on their own and has to ask the manager beyond that, but nobody has ever read it anywhere. The problem with custom is that it erodes through exceptions. One deep discount goes out at quarter end, the next month somebody says it was allowed last time, and the floor quietly moves. Six months later your average discount has settled somewhere nobody chose.

Unwritten authority has a second cost: it turns a pricing decision into a personal negotiation. Asking for approval becomes a favour, and granting it becomes a favour returned. The manager who says yes is being generous rather than applying a rule, which means the next rep learns to ask harder rather than to build a better case. Once the rule is written, the person disappears from the equation. The quote goes to review because it crossed a threshold, not because the rep pushed.

A workable rule defines three things at once: what opens the gate, who decides, and how long the decision may take. Miss any one of them and the rule does nothing. Saying that anything past a certain discount requires approval, without naming an approver or a deadline, changes nothing in practice. It helps to see where this rule sits inside the wider flow, which is what a proper quoting process that leaves nothing to chance is built to do.

Should the gate trigger on deal value or on margin?

These are two different thresholds catching two different risks. A value threshold catches the large commitment: every quote above a certain total goes to a second pair of eyes whether or not a discount was applied. The reasoning has nothing to do with price. Large deals carry scope that may be mis-specified, payment terms that may be aggressive, and delivery dates somebody promised optimistically.

A margin threshold catches the small deal that makes no money. A modest total can still fall below cost once onboarding hours, a custom integration and three months of heavy support are loaded onto it. A value gate never sees that. Mature teams therefore run both: a value ceiling and a margin floor. Rocketly's Manager Approval Gate evaluates exactly this pair of conditions — a quote that crosses the value threshold or drops below the margin floor is routed for approval automatically and cannot be sent to the customer until someone approves it.

There is a third threshold that most teams skip: line-level deviation. If a single item is priced well away from its standard rate, that quote sets a precedent even when the total and the overall margin look fine. The same customer will expect that rate at renewal, and comparable accounts hear about it. Adding per-line deviation to the gate is what protects the price list over a multi-year horizon rather than a single quarter.

How do you calculate a margin floor you can defend?

The margin floor is the line below which the deal stops working in your favour. The most common mistake in setting it is counting only direct cost. In a software subscription, infrastructure is usually the smallest number on the sheet. Implementation hours, integration work, training, the support intensity of the first quarter and the account manager's time typically add up to more than the hosting bill.

A practical method: pull the actual delivery cost of every deal closed in the last twelve months, group it by customer segment, and take the upper-quartile cost per segment as your floor rather than the average. Using the average hides half of your expensive deals by construction. The upper quartile gives you a cautious floor — the gate opens a little more often, but you stop being wrong in the direction that costs money.

Review the floor quarterly, not annually. Labour cost, currency and product scope all move. For teams quoting in more than one currency this is a problem in its own right: the same discount percentage lands on a different margin once the rate has drifted. Because the Multi-Currency and Live Rate feature seals the exchange rate at the moment the quote is issued, a later margin review does not turn into an argument about which rate applied.

How many levels should the approval matrix have?

Level count follows the nature of the decision, not the size of the org chart. Two levels serve most teams: the band a rep owns outright, and the band a manager reviews. A third level earns its place only for quotes that break the margin floor or alter standard contract terms, because at that point the decision is no longer a sales decision but a commercial-risk decision, and finance needs to see it.

Anything past four levels stops functioning. Each level adds waiting time, and in a four-link chain the competitor has finished their presentation before your quote reaches the buyer. Widening the band usually beats adding a level: give the rep a little more room and route everything outside that room to one manager.

The table below is an example authority matrix for a mid-sized B2B sales team. The bands and response times depend on your own margin structure; the point is the shape, not the numbers.

Discount band	Margin position	Approver	Target response
List price through the rep's own band	Above floor	No approval needed	Immediate
Middle band	Above floor	Sales manager	4 hours
Upper band	Close to floor	Sales director	1 business day
Beyond the upper band	Below floor	Director and finance jointly	2 business days

Set the rep's band wide enough that it is genuinely usable. A matrix that sends nearly every quote to review is not a gate but a barrier, and teams route around barriers rather than through them.

How fast should an approval come back, and what does delay cost?

A quote sitting in review is a quote going stale. Buyers open their decision window on their own calendar, and when you miss it you do not lose an argument, you lose your place in the queue. Approval time should therefore be written as a commitment rather than an aspiration: if the rule says four hours, it means four hours.

The cleanest way to make that commitment real is to define approval time as an internal service level. The same reciprocal obligations teams write between sales and marketing work just as well inside the sales organisation; the logic behind using an SLA to make obligations mutual transfers directly. The manager commits to a response window, and the rep commits to sending a complete request.

If you want to price the cost of delay, watch two numbers: average time in the approval queue, and the win-rate gap between quotes that went to review and quotes that did not. The second number tends to sting, and it usually argues for making the gate faster rather than removing it. Reading that alongside the broader work of accelerating a deal in practical ways puts approval time in the right frame: it is one more queue in a cycle full of queues.

Where should a pending quote sit? Why an inbox is not a queue

If the approval request travels by email, the quote's fate now depends on the randomness of somebody's inbox. Maybe the manager spotted it among a hundred and fifty messages; maybe it sat for three days and nobody noticed. The rep chases, the manager says it never arrived, and the conversation moves from the substance of the quote to the fate of a message.

A queue removes that ambiguity. The Approval Queue lists every quote that crossed a threshold in one place, together with how long each has been waiting. The manager opens the list in the morning, sees the oldest item first, and either approves from the queue or sends it back with a reason. Nothing disappears into an inbox, and "it never reached me" stops being an available explanation.

The second benefit is measurement. Once waiting time is visible in a queue it becomes a number you can manage: which band waits longest, whose queue is growing, which days of the week create a backlog. Reading that data next to the rest of the funnel is what turns it into action, and it is easier when the pipeline stages and their conversion are tracked on the same screen.

How does a rejected quote find its way back?

The neglected half of every approval gate is the return path. Approval is easy — the quote goes out. What happens on a rejection is usually undefined, and the rep is left guessing. In a well-built gate a rejection is not a full stop but a reasoned instruction: not "I'm not accepting that band" but "drop the discount on the implementation line, stretch the term to two years, and I'll approve it."

Making the reason field mandatory is the cheapest improvement available to any gate. A rejection without a reason forces the rep to guess, and the guess produces a second wrong quote. A rejection with a reason is effectively a revision brief, and it usually produces an approved quote on the second pass. Watch the tone as well: anything you would not want the customer to read should not be typed into the record, because quote records are read by more people than you expect.

On the revision pass, narrowing scope is often a better answer than cutting price. The Package Selection and Partial Acceptance feature lets the buyer approve only the lines they want, which keeps "buy less" on the table as an alternative to "pay less". That lands inside the buyer's budget without touching your margin floor. Framing that conversation well starts with understanding what "it's too expensive" is really telling you, because most price objections are value or timing objections in disguise.

Where does the decision live afterwards?

An approval gate does not exist only to cap discounts; it exists to leave a trace. Six months later, when the same customer comes up for renewal, the account manager needs an answer to "why did we give this band last time?" If the answer lives in one person's memory, it leaves when they do.

That is why the decision has to be stored with the quote itself rather than beside it. Version History and Comparison keeps a snapshot of every revision, so which band was granted in which version, by whom and when, can be read back line by line. Putting two versions side by side shortens the "that isn't what we agreed" conversation dramatically, because the difference is visible rather than remembered.

The same trail carries forward when the quote becomes a contract. Only an accepted quote can be converted, and the total, currency and parties move across unchanged, so an unapproved band cannot leak into a signed agreement. That record is still there at renewal time, and having last period's pricing rationale intact is the first requirement for running renewals as a process rather than a hope.

Does an approval gate really slow sales down?

This objection comes up on every team and it is partly fair: a badly built gate does slow sales down. What slows things is not the gate's existence but how often it opens and how long it stays shut. A gate with narrow bands, deep levels and no time commitment is an obstacle. A gate with one level, a sensible band and a response window measured in hours often speeds the cycle up, because it removes the vague hours a rep spends wandering around asking whether a number is allowed.

The second objection is that a gate signals distrust. In practice the opposite holds. A written band tells the rep exactly what territory is theirs, and knowing you can decide inside that territory without asking anyone is more autonomy, not less. A rep who does not know where the line sits either discounts too little and loses the deal or discounts too much and burns the margin.

The third objection is that the gate spoils the negotiation. Good sellers use it as a negotiating instrument instead: "I have to take this number to approval — can you commit to two years in return?" turns a unilateral concession into a trade. Among the methods that actually close deals, structured trades of this kind are the most durable, and an approval gate hands you a legitimate reason to ask for one.

How do you roll the gate out in thirty days?

Week one is measurement. Map the distribution of discounts given over the last two quarters: how many quotes sat in each band, how win rate moved across bands, and above which band deals fell through the margin floor. Any threshold set without that distribution is a guess, and guesses land either far too tight or far too loose.

Week two is writing the rule. One page is enough: the bands, the approver for each, the target response time, and what a rejection reason must contain. Write it with the sales team rather than announcing it to them — whoever is in the room when the rule is written is the person who applies it later. It is worth revisiting how a price quote is actually constructed the same week, so the pricing rule and the document that carries it are designed together.

Week three is configuration. Define the thresholds, open the queue, wire up approver notifications, and run the whole thing in advisory mode for a fortnight: quotes are routed for approval, but sending is not blocked. That period tells you whether the thresholds are in the right place before anyone is inconvenienced by them. In week four make the gate binding, and at the end of the first month measure three things: the share of quotes routed to approval, the average approval time, and how many rejected quotes were revised and won. Quote Analytics and Export lets you pull acceptance rates and the amounts you get stuck at into CSV for your own reporting; reading those figures next to the follow-up discipline that turns sent quotes into revenue shows what the bands are doing to win rate.

Expect the first three months to be a tuning period. If too many quotes are routed, widen the band; if almost none are, lower the threshold. A gate is not judged by how many quotes it stops but by how many go out in the right band, on time. If you want to try this arrangement with your own numbers, you can open a Rocketly account and set the thresholds against your own margin structure.

FAQ

Is an approval gate overkill for a small team?

On a three-person team everyone may already see every quote, but you still need the record. Once the team passes five people, or the first renewal cycle arrives, nobody remembers why a past band was granted. Keep the gate single-level and wide in a small team; the goal is to record the decision, not to slow anything down.

What happens when the approver is on leave?

Nothing should wait. Name a primary and a backup approver for every band. Without a backup the gate jams during the first holiday and the team invents unofficial ways around it, which is worse than never having built one.

Must every quote over the threshold really be blocked?

Yes, otherwise the threshold degrades into a suggestion. A quote that cannot be sent until it is approved is what makes the rule real. In exchange, make sure the threshold sits in the right place: a gate that stops every second quote discredits itself.

What is the fastest way to shorten approval time?

Send a complete request. If the approver's screen already shows the account history, the quote's margin, the rationale for the band and what was obtained in return, the decision takes minutes. Without that, the approver has to ask, and the number of rounds doubles. Standardising the request is the quickest route to getting approvals back in hours rather than days.