Every Founder asks the same first question: What should we pay?

It is the wrong first question. The number is the easy part and it is broadly public.

The part that decides whether a strong seller accepts, ramps and stays is how the package is constructed against a motion that is still being invented.

The Numbers, Stated Plainly

For an Enterprise AE selling data or cloud infrastructure into technical buyers, London ranges we see consistently at Seed and Series A:

• Base: £90,000–£120,000. Seed sits at the lower end, Series A at the upper.
• On-target earnings: £170,000–£230,000.
• Equity: 0.15%–0.4% at Seed for a first or second seller; 0.05%–0.15% at Series A.

US-based equivalents run roughly $150,000–$185,000 base against $300,000–$360,000 OTE, with equity bands similar in percentage terms and materially larger in expected value.

These are ranges, not rules. A candidate who has sold a comparable product to a comparable buyer will command the top of the band, and should. The mistake is rarely paying the wrong number — it is paying the right number in the wrong shape.

Why the 50/50 Split Is Usually Wrong Here

The standard enterprise SaaS convention splits OTE evenly between base and variable. That convention assumes the seller controls their outcome: a defined territory, a known buyer, a product that does what the deck says.

None of that is true at Seed in AI infrastructure. The seller is discovering the ICP, testing three messages, and closing deals whose technical evaluation depends on an engineering team they do not manage. Asking them to put half their income behind variables they cannot control does not create urgency. It filters out the people with options.

A 60/40 or 65/35 split is the sensible default for the first two sellers, moving toward 50/50 once the motion is proven and quota is credible. Founders sometimes read this as paying for effort rather than results. It is not. It is pricing risk honestly: at this stage the company owns most of the execution risk, so the company should carry more of it.

Ramp Is Part of the Package, Not a Courtesy

Enterprise cycles in AI infrastructure run six to nine months from first conversation to signature, and longer where a security review or a data-residency question is involved. A seller who starts in January will not close their own sourced deal until the autumn.

Write the ramp into the offer explicitly. Two quarters at full variable paid against a reduced quota is standard and defensible; three is reasonable where cycles genuinely exceed six months. What is not defensible is a full quota from month one with a vague promise that “we will be sensible about it”. Candidates have heard that before and they discount it to zero.

Set Quota Where a Good Seller Hits Eighty Per Cent

Without comparables, quota-setting becomes an exercise in inference. Two anchors help.

The first is capacity: average deal size multiplied by the number of deals one person can realistically run given cycle length and evaluation load. In infrastructure with a heavy proof-of-concept phase, six to eight concurrent live deals is the practical ceiling for one seller without solutions engineering support.

The second is the cost ratio: at Seed and Series A, a quota of three to four times on-target earnings is the usual band. Below three, the model does not work. Above five, you are testing whether the seller can read a spreadsheet.

Then apply the discipline most Founders skip. Set the number so a genuinely good seller lands around eighty per cent, not one hundred and twenty. A quota everybody misses tells you nothing about performance and guarantees attrition at month nine — exactly when their pipeline was about to convert.

Accelerators Are the Cheapest Ambition You Can Buy

Accelerators above quota are underused at this stage, usually because founders are protecting a cash model. The logic is backwards. Every pound paid through an accelerator is a pound against revenue that already landed. Uncapped commission with a step-up above one hundred per cent — commonly 1.5x — costs nothing when it is not earned and signals confidence when it is.

Capping commission at Seed sends the opposite signal, and strong candidates read it precisely: this company expects to miss.

Equity: Be Specific or Be Discounted

Senior GTM candidates in this market are fluent in equity mechanics, and vague answers cost you. Expect to be asked for the number of options, the total outstanding shares, the strike price, the last preferred price, the liquidation preference structure, and the post-termination exercise window.

Give a percentage without those inputs and the candidate cannot value the offer, so they value it at zero and negotiate on cash. That is a self-inflicted wound: you pay more base for equity you already granted.

A ten-year post-termination exercise window is one of the cheapest differentiators available to an early-stage company. It costs the cap table nothing today and removes a real, well-understood risk for the candidate.

What Candidates Are Actually Comparing

Your offer is rarely being weighed against another Series A offer. It is being weighed against a senior role at a well-funded lab or a large cloud provider paying a base close to your entire OTE, in cash, with liquid stock.

You will not win that comparison on money and you should not try. You win on scope, on ownership of a function rather than a territory, and on the honesty of the numbers you put in front of them. Which is why the compensation conversation belongs early — in the first or second conversation, not at offer stage. Raising it early costs one awkward exchange. Raising it late costs the hire.

The Design Question Behind the Number

Compensation at this stage is a statement about what the company believes is true. A 50/50 split against an uncalibrated quota says the motion is proven when it is not. A capped commission says growth is expected to be modest. A vague equity answer says the founder has not thought about the candidate’s position.

Get the structure right and the number becomes a conversation rather than a negotiation. We build these frameworks with Founders before a search opens — how we work.