"Do you need a deal desk" is a deal-shape question, not a revenue question. Most guides tell founders to wait until $10 to $15 million in ARR before adding one, then file it under a problem for later. That advice mixes up two different things: hiring a deal desk analyst and having a deal desk process. You can run the second one, a documented set of thresholds and a single approval owner, long before you can afford the first. Skipping that gap is exactly what turns every discount request into a Slack thread nobody can find six months later, when a customer wants to renew on the same terms and nobody remembers what you agreed to. Here's the three-signal test that tells you if you're already there, and what a deal desk looks like before it's a department.
What a deal desk actually is (not a department)
A deal desk is a documented set of rules for handling non-standard deals: pricing exceptions, custom contract terms, and payment schedules that fall outside what you'd normally quote. It is not a job title. At most startups, it starts as a fraction of one person's existing role, usually whoever already owns pricing, plus a short written list of what needs a second look before it goes out.
The trigger for "non-standard" is anything that steps outside your published terms: a discount below your normal floor, a contract length or payment schedule you don't usually offer, or custom legal language that needs a redline. Route those through a single documented decision point and you have a deal desk, even if nobody has that title on a business card.
GitLab publishes its full deal desk handbook publicly, with named roles, SLAs, and escalation paths. It's a useful reference for what a mature process looks like, not a template to copy at ten people. Your version should fit on one page.
The three-signal test
Revenue doesn't tell you whether your deals have started misbehaving. Deal shape does. Run this against your last quarter of closed deals:
- Deal shape. Roughly 1 in 5 of your deals needed a non-standard call on pricing, terms, or payment schedule. That's the point where hand-reviewing each one starts to overwhelm whoever owns approvals.
- Time cost. You or your head of sales spend more than two hours a week on ad hoc exception requests, or deals sit three or more business days waiting on one person's sign-off.
- Consistency. You can't recall, without checking Slack history, what you decided on the last five exception requests and why.
If two of these three are true, build the lightweight process below. If none are, rep discretion is still fine, and adding process now would just slow your reps down for a problem you don't have yet.
The pattern shows up the same way almost every time deal shape and consistency both fail: two reps give different multi-year discount depths to similar-sized customers, each defensible on its own. The real cost isn't the discount. It's eight months later, at renewal, when the customer who got the deeper discount forwards the original quote and asks why this year's increase looks so much steeper.
Why "wait for $10M ARR" is the wrong instinct
Revenue-based thresholds feel objective, which is why they get repeated so often. But they measure the wrong thing. A Series A company selling five-figure deals to mid-market buyers can hit the 20% non-standard-deal mark long before a larger company selling simple, low-touch subscriptions ever does. Waiting for an ARR number means waiting for a symptom that has nothing to do with your actual deal complexity.
The cost of waiting compounds. Sales cycles have been getting longer across B2B: 57% of sales professionals told Salesforce their deal cycles increased year over year. A well-run deal desk works against that trend instead of adding to it. Centralizing non-standard deal review can cut sales cycle time by up to 40% and lift sales productivity by up to 20%, largely because reviewers look at a deal in parallel instead of passing it sequentially from finance to legal and back. Wait for a revenue trigger, and you're choosing to keep the slower version for longer than you need to.
How to build the lightweight version
Skip the department. Start with three documents and one owner.
- Discount bands. The maximum a rep can approve alone, and who signs off above it.
- Contract term tiers. Which lengths and payment schedules are standard, and which need a second look.
- An exception log. One shared doc where every non-standard approval gets written down, even briefly. This is what makes the consistency signal above checkable without digging through Slack.
Put one person in charge of the first twenty or so approvals, usually the founder or head of sales, so you see what actually comes up before writing permanent rules around guesses. Handle a month of real requests and the thresholds mostly write themselves.
This is the same lightweight discipline behind a tiered discount approval process once you're ready to formalize it. The test above tells you when. That process tells you how, in detail: three tiers, sign-off owners at each level, and how to roll it out to a team that's used to deciding on its own. It's part of a broader set of sales ops breakdowns built for teams operating without a dedicated ops hire.
When rep discretion is still the right call
Not every early-stage team needs this yet. Discretion works fine when your pricing bands are narrow, deal volume is under roughly ten a month, and your reps sit close enough to you that a surprising term would get flagged in a standup before it reached a signature.
The tell that discretion is still working: contracts don't surprise you when they land. If finance or the founder regularly finds out about a custom term after the ink is dry, that's the consistency signal from the test above showing up as a symptom instead of a checklist item.
Frequently asked questions
How many non-standard deals before I need a deal desk?
Once roughly 1 in 5 deals needs a pricing, term, or payment exception, hand-reviewing each one usually starts to overwhelm whoever owns approvals. That ratio matters more than any revenue number.
Is a deal desk the same as RevOps?
No. RevOps runs the systems and processes across your whole revenue team. A deal desk is narrower and only governs non-standard deals. Early on, it's often a fraction of one RevOps or sales-ops person's role.
Who should own deal desk approvals at a ten-person startup?
Usually the founder or head of sales, at least for the first month of requests. That's how you learn what your actual thresholds should be before writing permanent rules.
Do I need software to run a deal desk?
No. A shared doc with discount bands, term tiers, and an exception log covers the first few dozen deals. Move to dedicated tooling only once volume outgrows manual tracking.
What's the fastest way to check if I need one?
Pull last quarter's closed deals and count how many needed a non-standard call. If it's near 20%, or you can't recall what you decided on the last five exceptions, start the lightweight process this week.
A deal desk isn't a milestone tied to a revenue number. It's a documented answer to a question your reps are already asking you informally, one Slack thread at a time. Run the three-signal test against last quarter's deals before deciding it's premature. If you want a second set of eyes on the thresholds before you roll them out to the team, that's worth a conversation.