This is The Agentic Commercial Model, Season 2: the build. Issue 02 set out the loop. This is how you pay for it.
There is one law in commercial management and it has never been repealed. People do what they are paid to do.
Not what the strategy deck says. Not what was agreed at the offsite. Not what the town hall applauded. Whatever the plan pays for, in the month it pays, is what happens on the floor, and every leader who has ever tried to change behaviour without changing the plan has learned this at their own expense.
Which makes the current state of sales compensation the most reliable evidence available about what this industry actually believes.
Alexander Group's 2026 survey found 95% of XaaS companies are changing their compensation plans this year, and only 18% believe their current plans are effective. An entire industry redesigning something it already knows does not work, with almost no confidence the next version will be better. Meanwhile median enterprise quota to on-target-earnings multiples pushed from 4.5 times in 2024 to 4.8 times, and most reps still missed quota while pay rose.
Higher quotas. Higher pay. Lower attainment. Lower confidence.
That is not a plan needing a tweak. That is a plan measuring the wrong thing with increasing precision.
Why the quota broke
The bookings quota was a good instrument for a specific world. Revenue was captured at signature, delivery was somebody else's problem, and the gap between what a rep sold and what a customer received was commercially irrelevant.
That relationship has dissolved. The same survey found most technology organisations now run two or three pricing models simultaneously, each with a different relationship between seller activity and revenue realisation. A rep can close identical contract values under three different structures and produce three completely different outcomes for the business, and the quota records all three as the same number.
An instrument that cannot distinguish between a good result and a bad one is not a measurement. It is a ritual.
However the deeper problem is not measurement, it is timing. The quota pays for a moment. The loop earns across a lifetime. Pay for the moment and you have bought the moment, which is precisely what most companies have done and precisely why their agentic transformation stalls somewhere between the announcement and the renewal.
Pay the loop, not the close
If the operating model is a loop, the plan has to carry weight at all four stages. Not one stage with three sentiments attached. Four stages, four payable measures.
Land pays for activation, not signature. The measure is time to first verified value, and the rep holds it through activation rather than handing off at contract. A smaller deal live in three weeks is worth more to the loop than a larger one stuck in implementation for two quarters, and the plan should say so in cash.
Realise pays for verified value delivered. This is where the instrument from Issue 03 stops being a reporting exercise and becomes payroll. Outcomes defined before the contract, detected by system events, weighted by contribution margin. Without that infrastructure this measure is unpayable, which is the honest reason most companies will not attempt it this cycle.
Expand pays for discovered growth. Growth found in the usage map and converted, rather than argued at renewal. Weighted more heavily in mature territories, less in greenfield, because the underlying work is genuinely different.
Advocate pays for published proof. Named case studies, verified metrics, citable evidence. Season 1 established that this material now decides whether the next buyer ever learns you exist. If it decides pipeline, it belongs in a plan. If it is not in a plan, it will not happen, and you will keep asking why nobody harvests the references.
Four weights. Which stage carries most depends on your motion and your maturity, and anyone quoting you a universal split has never run one of these.
The hard part nobody has solved
Here is where I stop prescribing and start being honest, because this is the genuinely unsolved problem and most writing on the subject skirts it.
Value realises months after the rep did the work. Sometimes many months. A plan paying on realised value is a plan paying people in the second half of the year for work done in the first, and reps discount future money brutally. Defer too much and your best people leave for a competitor still paying on bookings, which is a real and immediate risk rather than a theoretical one.
Four mechanisms exist and none of them are clean.
Tranche the payment. A portion at land, the balance as value verifies. Simple, and it halves the immediate reward for the hardest part of the job.
Bank and draw. Credit the rep as value realises, pay against the balance on a regular cycle. Smoother, and it turns your comp plan into an accounting system most organisations cannot administer.
Raise base to fund the deferral. Honest, expensive, and it reduces the leverage that makes strong reps strong.
Pay a premium for patience. Deferred variable at a higher rate than immediate variable, explicitly pricing the wait. My preferred answer, and the one finance will resist hardest, because it costs more in cash to buy behaviour you could have had for free under the old model. That resistance is the whole reason the industry stays stuck.
Whichever you choose, name the trade-off out loud. Every option here transfers risk between the company and the rep, and plans that hide which direction the risk moved get discovered in month four by the people carrying it.
The only design test that matters
Before any plan goes live, run one question against every weight in it.
How would a smart, decent rep exploit this without technically breaking a rule?
Not a bad rep. A good one, behaving rationally inside the incentives you built. If you cannot answer, you have not finished designing.
Pay on time to first value and reps will avoid complex enterprise accounts that activate slowly, which is a strategic catastrophe dressed as efficiency. Pay on verified outcomes and reps will lean on customers to confirm value early, and will target accounts likely to succeed regardless of anything they do. Pay on discovered expansion and reps will wait for telemetry rather than create demand. Pay on published proof and the same three friendly customers will appear in every case study you produce.
None of that is dishonesty. It is people doing what they are paid to do, which is the law we started with. The plan is a machine for producing behaviour, and if you did not model the behaviour, you did not design the machine.
To the CRO, plainly
Your compensation plan is your commercial strategy. Not the deck, not the operating model diagram, not this newsletter. The plan.
If it still pays primarily on bookings, then whatever you announced this year, your actual strategy is to close contracts and hope. Your people already know this. They read the plan, they worked out what pays, and they adjusted within a fortnight. The only person still describing the strategy in terms of value realisation is you, in meetings, to other executives.
To finance, comp designed for cost control produces exactly the behaviour cost control deserves. If the deferral premium looks expensive, price the alternative, which is a transformation that never happens while you fund it at 15 to 25% of new revenue anyway.
And to the reps, you will game whatever we build, and I do not say that as an accusation. It is the system working. Which is why the design test above matters more than any split, and why the plan you receive in January tells you more about what your company believes than anything it says in January.
Reps behave exactly as you pay them.
So decide what you actually want, and then pay for that.
Next issue, the season finale: The Boardroom Issue. Diligencing a business whose NRR can lie, and the 100 day commercial model plan.
First published in the Agentic Commercial Model newsletter on LinkedIn, September 10, 2026. Read the original on LinkedIn.