Should commission still be tied to closing when AI-run outreach does most of the pipeline-building?

The shift

The work that filled the top of the funnel — researching accounts, writing and sending personalized sequences, following up, booking meetings — goes from scarce rep-hours to near-free and automated. Commission plans were built on a chain where a rep's effort produced pipeline and pipeline produced closes, so paying on the close implicitly paid for the whole chain. When AI severs the effort-to-pipeline link, the question is what commission is actually buying.

The axioms

• Commission is tied to the close because the close is the outcome the business gets paid for, and it's the least gameable point in the funnel to measure. (scarce: verifiable outcome)
• Paying on the close implicitly compensates the whole chain of rep effort that led to it — prospecting, outreach, nurturing — because that effort was the rep's and was expensive. (scarce: rep effort across the funnel)
• The rep who built the pipeline is the rep who should close it and get paid for it, because pipeline-building and closing were done by the same person and the relationship carried through. (scarce: continuous human ownership of an account)
• Activity that fills the funnel is worth rewarding (via SDR comp, ramp, quotas) because that activity was the scarce input that gated pipeline. (scarce: outreach capacity)
• Which accounts to pursue is a judgment call worth a rep's time because targeting well is hard and the cost of chasing the wrong ones is high. (scarce: judgment on where to spend effort)
• Someone accountable must own the commitment made at close — the price, the terms, the promise — because the buyer is committing budget and their own credibility to a person. (scarce: accountability + trust)
• Comp plans reward measurable volume because volume was a real proxy for effort, and effort was the thing that moved outcomes. (scarce: rep capacity as the bottleneck on results)

Invalid axioms

1. Paying on the close compensates the rep's effort across the whole funnel. The close-based plan was a clean way to pay for prospecting, outreach, and nurturing without measuring each — because a human did all of it and that labor was the cost. AI now does most of the top-of-funnel work at near-zero marginal cost, so a large share of what the commission implicitly paid for is no longer the rep's labor. Habit-trap: orgs keep full close-based commission rates as if the rep still performed the whole chain, quietly overpaying for work the tooling now does.
2. Activity that fills the funnel is worth rewarding as scarce input. SDR comp, dialing quotas, and outreach-volume targets priced the funnel-filling as the expensive, gating step. It's now cheap and automatable. Habit-trap: SDR comp structures and activity-based ramp targets persist as line items even though the activity they reward is no longer the constraint — and can be produced without the person.
3. The rep who built the pipeline is the one who should be paid for building it. Attribution assumed a human did the building. When AI-run outreach generates the meeting, the "pipeline-building" credit baked into the close commission is being paid to a rep for work a system did. Habit-trap: comp keeps crediting sourced/self-generated pipeline to individual reps as a differentiator, when sourcing is increasingly a shared automated capability rather than an individual's effort.

Unchanged axioms

1. Commission is tied to the close because the close is the outcome the business gets paid for and the least gameable point to measure. AI made the funnel cheap to fill but didn't make revenue arrive any earlier or make the close any less real. Paying on the verifiable outcome is more defensible now, not less, precisely because upstream activity has stopped being a signal of anything.
2. Someone accountable must own the commitment made at close. The buyer commits budget and internal credibility to a specific person; AI has no standing to make or answer for that commitment. This is where a commission is genuinely earned — for absorbing the risk of the promise, not for the volume of touches that preceded it.
3. Deciding which accounts are worth pursuing is judgment worth paying for. When outreach to anyone is free, chasing the wrong accounts is nearly costless to start and expensive to finish, so the value shifts to the call about where to point the effort. That's judgment on ambiguous, high-stakes targeting — not something the automation resolves by producing more volume.
4. Closing complex, multi-stakeholder deals is human work that the pipeline engine doesn't touch. Reading the room, navigating internal politics, structuring a face-saving concession, and carrying trust to the point of signature stay scarce. The AI built the meeting; it didn't close the deal, and the part it didn't do is the part worth the largest share of the comp.

New axioms

1. When AI builds most of the pipeline, the org has to decide how to split credit and comp between the tooling and the human who closes — a close that AI sourced, AI-nurtured, and a rep only finished isn't obviously worth the same commission as one a rep worked end to end, and there's no established way to price the difference.
2. When outreach activity is free, comp has to reward something other than volume, and the replacement metric isn't built yet. Activity was the measurable proxy; remove it and the plan needs to pay for judgment (which accounts, which concessions), relationship quality, and close ownership — none of which the org currently instruments as cleanly as it counted calls.
3. When the same automated pipeline could be worked by fewer reps, the org has to decide whether commission rates should fall as tooling absorbs the funnel — if AI does the work commission used to implicitly pay for, keeping rates flat overpays, but cutting them treats the close (still scarce) as if it got cheaper too. Separating "the funnel got cheap" from "the close did not" inside a single number is the unsolved design problem.
4. When pipeline is generated centrally by a system rather than sourced by individuals, territory and account ownership have to be re-drawn — the fairness logic of "you built it, you own it, you're paid for it" breaks when a shared engine builds it, and nothing has replaced the ownership model that comp plans assume.

Where it breaks

Comp plans still pay full close-based commission as though the rep performed the entire funnel (invalid: the rep no longer builds most of the pipeline) while the org can't yet price what the AI contributed versus what the human did (new: credit-splitting between tooling and closer is unsolved) — so the plan either overpays for automated work or, in trying to correct, cuts the close rate and underpays the one part that's still scarce and human. This calls hinge on how fast agentic outreach becomes genuinely autonomous rather than rep-supervised; the faster the pipeline-building runs without a human in the loop, the harder it gets to justify paying the closer for having "built" it, and the more urgent the credit-splitting problem becomes.

ai-axioms.vercel.app/axioms/marketing/sales-commission-structure-ai-axioms