Why Borrowed Benchmarks Fail
The standard playbook says an enterprise AE should carry somewhere between four and six times their on-target earnings. That maths rests on assumptions that simply don’t hold at a Series A data infrastructure company: a known ICP, an established budget line, referenceable customers, a sales cycle you can predict.
Selling a new storage layer or orchestration platform into engineering teams has none of those. The buyer often doesn’t know the category exists. The deal cycle is whatever the POC says it is. The ACV is whatever the last Founder-led negotiation happened to land on.
The problem isn’t ambition. It’s that the number was imported from a market that already exists.
Quota Is a Hypothesis, Not a Target
In year one, quota isn’t really a performance contract. It’s a test of the motion.
Build it bottoms-up from the only real data you have – the founder’s closed deals. Cycle length, ACV range, conversion from POC to contract, how many qualified conversations produced each win. Extend that honestly across a rep’s capacity and you get a defensible number. Small, probably. Real, definitely.
Then treat it as a hypothesis. Revisit it quarterly, in the open, with the rep in the conversation. If the POC-to-close rate turns out to be double what the founder managed, raise it together. If enterprise security reviews are adding a quarter to every cycle, adjust for reality rather than pretending the plan predicted it.
A missed quota should teach you something about the market. If it only teaches you something about the rep, you set it wrong.
Territory When Every Account Is Greenfield
Geographic territories make sense when demand is distributed and known. In a new category, neither is true.
Carve territory by segment and use case instead: one rep on AI-native companies with heavy inference workloads, another on enterprises modernising their data stack. Build small named-account lists from real signal – who’s hiring platform engineers, who’s publicly wrestling with the problem you solve – rather than handing over a region and a spreadsheet.
Fifty well-chosen accounts beat five hundred cold ones. Early selling in infrastructure is high-touch or it is nothing.
This also solves the fairness problem before it starts. When you eventually hire a second rep, splitting by segment gives each of them a coherent market to learn deeply, rather than an arbitrary line on a map that one of them will resent. Territory disputes at ten people are a symptom of design debt taken on at three.
Comp Design That Doesn’t Guarantee Attrition
The comp plan is where quota design errors become resignation letters. A 50/50 split against an untestable number means your best hire earns base salary for two quarters, does the maths, and leaves.
At this stage the structure should absorb uncertainty rather than transfer it all to the rep:
- Weight base more heavily than the textbook says – closer to 60/40 while the motion is unproven.
- Guarantee or partially guarantee the first two quarters of variable. You’re paying for market discovery, not just closed revenue.
- Put accelerators on the behaviour you need proven – new logos, POC conversions – not just total bookings.
- Let equity carry the real upside. That’s the honest trade at Seed and Series A, and strong candidates understand it.
We’ve written in more detail about the numbers in What Enterprise AE Compensation Actually Looks Like in AI Infrastructure.
Pay for the motion you need proven now. Not the motion you hope to have in two years.
Redesign Is Not Failure
Here’s the part Founders find hardest: you will get this wrong, and that’s fine. The first plan is a draft. Expect to rewrite territory and quota after two quarters of real data, and say exactly that during the offer process.
It costs you nothing. And it earns you a lot – because the candidates worth hiring into a category-creating company have seen bad plans before. What they’re screening for isn’t a perfect number. It’s a founder who’s honest about the uncertainty and structured about reducing it.
That honesty is a hiring advantage. Use it.
If you’re building your first comp plan and want a sense-check against what the market is actually doing, I’d love to talk. You can find how we work with early-stage GTM teams at withvector.io/solutions.
Vector is a specialist recruiting agency helping VC-backed AI and infrastructure startups build their GTM, product, and engineering teams.