It’s Thursday, 4 p.m. An AE is pinging you about a 30% discount they already promised on the call. You approve it, because losing the deal over process is worse. You flag it for the CRO and make a note for next month’s discount review. Then the next ping arrives.

Nobody hired you to run a deal desk. A deal desk is the team that reviews non-standard pricing, discounts and contract terms before a quote goes out, and most companies never staffed one. So the work is falling to whoever sits closest to the CRM, and that’s usually RevOps.

We surveyed 195 VP-level revenue leaders in North America this summer.

That’s most of a business week, spent waiting on an internal chain while the buyer waits on you.

Why RevOps ends up running the deal desk

Most CPQ tools are rulebooks: discount caps, price floors, approval chains, margin thresholds. Each rule answers one question, which is whether a quote is allowed. Inside the lines, the quote goes through. Outside them, the tool stops and asks a person.

That design assumes exceptions are rare. In a company selling multi-year ramps, bundles and usage, exceptions are most of the week. Every rule you add creates one more way for a quote to land in your inbox.

The people running the queue think it’s moving faster

Here’s the finding that surprised us. We asked RevOps and sales leaders the same question: how long does a pricing exception take to clear? 51% of sales leaders said three days or more. Among RevOps, it was 42%.

It’s the only measure in the whole study where sales reported more friction than the team running the process. Both are right about what they can see. RevOps sees an approval that cleared. Sales sees a buyer who stopped replying on day two. (More on that gap in why RevOps and sales disagree about approvals.)

The rules are missing the margin too

Time is the cost you feel. Margin is the cost you don’t see.

A discount cap catches the worst case. Say a rep opens at 25% and the cap is 20%. The rule saves 5 points. But similar deals that won closed at 10%, with white-glove support added and a 5% license uplift in year two. Nothing in a rulebook knows that. The rep sends 20%, the quote approves itself, and 15 points leave with everyone’s sign-off.

Guardrails did their job here. They just weren’t built for this job.

More rules are making it worse

When discounts creep up, the instinct is to tighten the rulebook. Lower the cap, add an approval tier. Each change routes more quotes to you and does nothing about the drift inside the new lines.

The systems thinker Donella Meadows ranked the places where you can intervene in a system. Adjusting parameters like thresholds and caps sat near the bottom; she called it “diddling with the details, arranging the deck chairs on the Titanic.” Changing who sees what information, and when, ranked far higher.

For pricing, that means putting the right information in front of the rep at the moment they choose the number.

Guardrails cap the worst deal. Guidance moves the average one, and the average is where your margin lives.

What a better deal desk process looks like

Four things change.

This is how we built Everstage CPQ, and it works on its own.

Back to the job RevOps was hired for

RevOps was never meant to be the approval queue. The original promise of the role was to design the revenue system: territories, comp plans, routing, the pricing rules every deal runs on. Revenue architects, not maintenance crew.

Most RevOps leaders already think of the job that way. We asked what they’d do with time back, and strategic planning came first. The title moved before the calendar did.

A deal desk sitting in your inbox is the clearest case of architecture turned into maintenance. When the system handles the standard deal, you stop being the queue. You design the pricing policy, and the system applies it on every quote. That’s the part of the revenue system CPQ should give back to you.

If you’re the CRO, the same shift shows up in your number: fewer points going out the door at quarter-end, and quotes going out while the buyer still remembers the call.

It’s the last week of the quarter, 9:47 PM. A rep has a quote open and a 25% discount one click away.

What decides whether they click it?

I opened my session at Gong Celebrate this week with that question, and the room answered the way most rooms do. Pressure. Quota. The buyer pushing. All true, and none of it is the whole answer.

The rep isn’t the villain

When discounts run high, we tend to blame the rep. They’re sandbagging, or panicking, or giving margin away to hit a number.

Look at it from the rep’s chair. They’re about to pick a number, and they have almost nothing to go on. They don’t know what deals like this one actually closed at. They don’t know which terms moved the last three buyers. They remember one painful loss from last quarter, and they know the quarter ends Friday.

So they’re guessing. And at quarter-end, the guess is running generous.

That generosity adds up.

And with 76% of sellers missing quota in the first half of 2026, the pressure isn’t easing.

That’s a rational response to the information they have. Discounting is the one lever a rep can feel. If you want them to pull a different one, you have to show them it exists.

The decision is happening before the quote

Here’s what I think most revenue teams get wrong. We treat the quote as the moment of decision, and we put our controls there: caps, floors, approval chains.

By the time anyone opens the quote, the decision that matters is already made. It happened on the call, when the rep hinted at flexibility. It happened in the Slack thread, when the manager said “do what it takes.” The quote is paperwork recording a decision made earlier, with less information than anyone would want.

And the paperwork is slow. Across roughly 3,700 enterprise deals we analyzed, the median time from the last call to the first quote was 56 hours. That’s more than two days where the only number in the buyer’s head is the one the rep hinted at.

Why discount approvals miss the real decision

Most approval processes are built to catch the worst case. That’s useful, and it isn’t enough.

Our survey of 195 revenue leaders found that at 47% of companies, a pricing exception takes three or more days to approve. Days pass and the buyer cools. Now ask how many of those exceptions were approved exactly as the rep asked. Most teams have never checked. If the answer is most of them, the control added time without adding judgment.

Meanwhile the discounts inside the guardrails go straight through. A 20% cap stops a 25% ask. It says nothing about the deal that should have closed at 10%. (Why that’s the bigger leak.)

How to reduce quarter-end discounting: change what the rep sees

If the discount is being decided on the call, that’s where the help has to show up.

That’s the bet behind Everstage CPQ. It reads the Gong call and the CRM record and drafts the quote while the deal is still taking shape, so the rep starts from the deal in front of them instead of a blank line and a hunch.

SAMPLE DATA

Gong call · Discovery42 min · Buyer and AE

12:04Buyer: We’d need it for about 120 reps to start.

18:30Buyer: Two years works if support is premium.

24:12AE: We can look at how year two is priced.

Drafted quotefrom CRM + call

The rep still makes the call. They’re just making it with more than memory and a deadline.

What CROs should ask on Monday

Three questions, for your RevOps lead:

If the answer to the third is “not much,” you’ve found where your margin is going.

The full findings, including where RevOps and sales disagree about how long approvals take, are in our new repor

This summer we surveyed 195 revenue leaders in North America, all VP-level or above. Half run RevOps. Half run sales.

We asked both groups the same question about their pricing approval process: how long does a pricing exception take to clear?

A nine-point gap doesn’t sound dramatic. What makes it interesting is that it’s the only measure in the whole study where sales reported more friction than the team running the process. Everywhere else, RevOps saw at least as much pain as the people relying on it.

On pricing approvals, the people stuck in the queue are feeling something the people running it aren’t.

Both sides are right

This isn’t one group misremembering. They’re measuring different things.

RevOps sees the approval: a request arrives, it gets routed, someone signs off. From inside the queue, that can look like a day and a half.

Sales sees the deal: the buyer asked for a discount on Tuesday, the rep promised an answer, and the revised quote went out the following week. From the deal’s side, the approval is only one step. There’s the time before the request gets logged, and the time after approval when someone rebuilds the quote by hand. 40% of companies in our survey still build quotes and configure pricing manually.

The clock RevOps is watching starts later and stops earlier.

Why the gap matters

When the people running a process think it’s faster than it is, they don’t fix it. There’s no fire to put out. The approval SLA looks healthy on the dashboard.

Meanwhile the CRO is hearing about stalled deals by name. Sales leaders in our survey told us the window to act is closing before the data arrives (24% named it as their top reason decisions can’t wait). They’re judging the process deal by deal. RevOps is judging it in aggregate. Neither view is complete on its own.

That’s how a queue survives for years. Everyone agrees it could be better, and nobody agrees it’s broken.

Both sides want the same thing

The part that gives me hope: when we asked what they’d do with time back, the two groups gave different answers that point the same way.

RevOps ranked strategic planning first. Sales leaders ranked time with customers first. Sales wants to coach deals instead of chasing approvals.

RevOps wants its original job back. The role was built to design the revenue system, the territories and comp plans and pricing rules every deal runs on. A revenue architect spending the week clearing exceptions is maintaining the building instead of designing it.

The one deal desk KPI that closes the gap

The fix starts with a shared number. Most deal desk KPIs measure the queue: requests in, approvals out, average time to approve. Track the one the buyer feels instead. Get RevOps and the CRO in a room and time one real deal from the buyer’s ask to the revised quote landing in their inbox. Then look at how many of last quarter’s exceptions went through exactly as the rep asked.

If most did, the approval isn’t adding judgment. It’s adding days.

Every pricing exception is leaking revenue twice.

The first leak is the one everyone talks about. The buyer asks for a discount, the request sits in a queue, and three days later the deal has cooled. In our survey of 195 revenue leaders, 47% said a pricing exception takes three or more days to approve.

The second leak is quieter, and I think it’s bigger. It happens when the exception is approved exactly as the rep asked.

The approval that doesn’t approve anything

Walk through one deal. This one is illustrative, but you’ll recognize it.

A rep opens at 25% off. The cap is 20%, so the guardrail catches it and the quote comes back at 20%. The rule saved 5 points and everyone feels good.

But similar deals that won, same size and segment and product mix, closed at 10%. They added white-glove support and a 5% license uplift in year two. The quote went out at 20%, approved itself, and 15 points left with everyone’s sign-off.

Nothing broke. The guardrail did exactly what it was built to do. It caught the worst case. It had no way to know what this deal should have been. The guardrail saved five points. The evidence would have saved fifteen.

Faster isn’t fixing it

The obvious response to the first leak is speed: buy an approval workflow, route requests through Slack, cut the three days to one.

Speed fixes the lag. It doesn’t fix the discount. A faster approval is still an approval, and it’s still approving 20%.

Our data backs this up. Teams that had automated a step or two, an approval workflow here and a quote template there, felt almost no different from teams doing everything by hand. 20% of them said AI was meeting expectations, against 18% of fully manual teams. Time spent on operational work was identical, at 33%. Only the 23% who had connected most of the chain saw a real difference: 44% satisfaction, and ops time down to 14%.

Discount guardrails vs guidance

Guardrails and guidance answer different questions.

You need both. Guardrails are your floor, and no CFO should give them up. But most of your margin isn’t in the worst deal. It’s in the hundreds of ordinary ones, each a few points more generous than it needed to be. Guardrails cap the worst deal. Guidance moves the average one.

Those few points are worth more than they look. Every discount point comes straight off gross profit, and McKinsey found that a 1% improvement in price realization lifts operating profit by 8–9%.

Deal desk best practices: what guidance looks like

Three shifts, in order of how much margin they protect.

1. The rep sees what tends to close before choosing a discount. While they still have room to move on structure: term length, support tier, a year-two uplift.
2. Rules check the quote while it’s being built, not after it’s sent. A margin threshold flags the line as the rep types, not three days later in someone’s inbox.
3. Repeat exceptions become rules. When the same 15% multi-year discount keeps getting approved, that’s policy nobody wrote down. Write it down. This is the design work RevOps was hired for: architecting the pricing policy, not approving it one quote at a time.

In Everstage CPQ, discount caps, price floors and margin thresholds check the quote as it’s built, clean quotes approve themselves, and Eva suggests rule changes from the exceptions you keep approving. A person always switches them on. It works on its own, without any other Everstage product.

The question for your next pricing review

Stop asking how fast approvals are moving. Ask how many of last quarter’s exceptions went through exactly as the rep asked, and what those deals would have closed at with better information.

That number is the second leak. Most companies have never measured it.