This is The Agentic Commercial Model, a newsletter about the operating model crisis hiding inside the AI gold rush.
Let me start with the sentence that should keep more SaaS leaders awake than it does.
You can build a world-class agentic product, price it intelligently, and still lose, because the operating model underneath it was designed to sell something that no longer exists.
Issue 02 ended on the move from access to consumption to outcomes. This issue is about the thing that move actually demands, and the thing almost every company treats as an afterthought. Because changing your pricing model is the easy part. You change a pricing page in an afternoon. What you can't change in an afternoon is how your entire commercial organisation behaves, who's incentivised to do what, when, and why. And that, not the pricing model, is where the agentic transition is going to separate the companies that survive from the ones that quietly decay.
Here's the uncomfortable truth: comp plans are not a detail you fix after the operating model. Comp plans are the operating model, made visible. Show me how you pay people and I'll show you what your company actually believes about how value gets created, regardless of what the strategy deck says.
The shift nobody is pricing in: from a moment to a motion
For thirty years, the SaaS commercial model has been built around a moment. The close.
A rep finds an opportunity, runs a process, negotiates, and signs a contract. At the moment of signature, value is booked, the rep is paid, and the customer is handed off, to onboarding, to customer success, to a renewal motion that picks up again twelve months later. The entire machine is engineered around that single transactional event. Quota is bookings. Commission is paid on signature. The org chart has a clean line between the people who sell (before the moment) and the people who serve (after it).
This worked because the thing being sold was access. You sold someone a seat, a licence, a right to log in. Whether they extracted value from that access was, commercially, somebody else's problem, usually a customer success team that existed downstream, under-resourced, and structurally disconnected from the people who'd made the promises.
Agentic AI breaks this completely, because agentic AI isn't access. It's action. And action only generates value when it's used, repeatedly, well, on the right problems, in ways that compound over time.
The unit of value is no longer the contract. It's the ongoing, evolving, monitored, continually-renegotiated exchange between what the customer spends and the value they actually realise. That's not a moment. It's a forever motion. And you cannot run a forever motion on a comp plan built for a moment.
Snowflake did the hard thing first - and told everyone why
The most useful case study here isn't a failure. It's the company that understood the problem early and did the structurally painful thing, in public, before its IPO.
When Snowflake moved to consumption pricing, the obvious change was to the pricing model. But Snowflake's leadership understood that the pricing change was the trivial part. The hard part was that paying reps on bookings, on the moment of signature, actively worked against the business. When you pay reps on bookings in a consumption model, you create incentives to over-sell a customer ahead of their actual usage, or to slow deals down to inflate the commitment. The rep is optimising for the signature. The business lives or dies on the consumption that comes after it.
So Snowflake rebuilt the comp model. They moved to a structure split between committed bookings and recognised consumption, paid out as it's earned explicitly to align the rep's incentives with the customer's actual realisation of value rather than the size of the initial signature. As their CFO put it, in a consumption model you need the rep to be involved in getting the customer to actually use the product, so a large portion of pay had to move onto consumption itself.
But here's the part most people miss, and it's the whole point of this issue. Snowflake didn't stop at the comp plan. They changed the org. One of the first things they did was eliminate the separate customer success function entirely, the view being that everyone owns customer success, and salespeople are involved precisely because their comp is now tied to ongoing revenue rather than the close.
Read that again. They didn't add a customer success team to chase value realisation after the sale. They dissolved the artificial wall between selling and serving, because in a consumption world that wall is the problem. The logic was direct: the more time a rep spends with a customer, the more opportunities they recognise for that customer to consume, which leads to more value realised and more revenue so the incentive structure should make customer engagement the rep's job, not someone else's.
That is what taking the operating model seriously looks like. Not a new SKU. A different shape of company.
What Snowflake learned the hard way
It wasn't clean, and this is the part worth dwelling on because it's where most companies will get the implementation wrong.
At one point Snowflake had reps on 100% consumption-based comp, and it didn't work, reps struggled, not just with delayed payouts, but with the entire mental model. If your pay is based on how much a customer uses, and a good finance function can forecast that usage accurately from historical data, how does a rep actually outperform? In a usage-based world it's genuinely harder for a rep to blow past their number, because the finance org has become so good at predicting consumption.
So the answer wasn't a single model. Snowflake landed on a hybrid: territory profiles that emphasised bookings in greenfield areas and consumption in mature ones, matching the incentive to the actual nature of the work in that part of the business. And critically the part nobody wants to fund, they invested in the tooling and infrastructure to monitor customer consumption, so reps had the data needed to drive value-focused conversations rather than flying blind.
The lesson isn't "copy Snowflake's comp plan." The lesson is that the comp plan was the output of a deeper redesign of roles, of org structure, of the data environment, of what "success" even means for a rep. As one analysis put it, the cultural shift was that sales is no longer about the close; it's about the adoption curve. You cannot import that shift by editing a commission table. You have to rebuild the environment the table sits inside.
The failure mode: hope dressed up as a renewal motion
Now contrast that with what most companies are actually doing.
The default agentic transition looks like this. Ship the AI product. Change the pricing page to add consumption or credits. Leave the sales comp plan almost entirely intact, still weighted to bookings, still paying on the close. Hand the customer to a customer success team after signature. Hope the customer realises enough value that, twelve months later, they renew.
That's not a commercial operating model. It's a hope-based one. And it has a specific, predictable failure mode in a consumption or outcome world: the revenue that was "booked" never materialises, because nobody downstream was incentivised, or equipped to ensure the customer actually used what they bought and got value from it.
In a seat-based world, this was survivable. The customer paid for the seat whether they logged in or not. In a consumption world, unused capacity is unrealised revenue, and unrealised revenue is a renewal that's already dead it just doesn't know it yet. The customer success function, in most companies, is structurally set up to discover this too late: under-resourced, disconnected from the commercial incentive, measured on activity rather than value realised, and brought in after the promises were made by people who've already been paid and moved on.
The industry is slowly waking up to this. The most forward-thinking view is that in consumption and outcome models, expansion becomes continuous the more a customer uses, the more they expand and the renewal stops being a calendar event at all. Some are now restructuring into integrated "pods" where account executives and CSMs share a single variable comp pool based on total book-of-business growth, specifically to force collaboration and stop reps closing bad-fit deals that hurt the pod's retention number. That's the right direction. But notice what it requires: a fundamental rewiring of who's paid for what, not a tweak.
Everything you knew about selling is changing
Let me be precise about the scale of what's shifting, because "comp plans need updating" radically undersells it.
The old model sold access and optimised for the moment of sale. The agentic model sells action and has to optimise for a perpetual, evolving value exchange. That single shift cascades through every part of the commercial organisation:
The job of the seller changes. It's no longer to close and move on. It's to land, then stay embedded monitoring how the customer uses the product, why they use it, what value they realise, how that evolves, where new use cases emerge, and whether the customer is positioned to advocate to others. The sale is the beginning of the commercial relationship, not the end of it.
The operating cadence changes. The old rhythm was annual: sell, onboard, go quiet, reappear at renewal. The new rhythm is continuous. Value has to be conveyed and demonstrated constantly, not surfaced in a QBR once a quarter. Usage has to be monitored always, not audited at renewal. Learnings about how customers actually extract value have to flow back to customer success and product teams continuously, not in an annual review.
The org design changes. The wall between sell and serve has to come down, because in a forever-value-exchange model the distinction is incoherent. Someone has to own the perpetual relationship, and they have to be incentivised on the thing that actually matters value realised over time — not on a signature that happened once.
And the thing that makes all of this real, or exposes it as theatre, is incentives. You can put "customer-centric" on every slide. You can stand up a customer success function and call it value realisation. But if your reps are still paid on the close, your CSMs are still measured on activity, and your renewal team still operates on hope, then the operating model hasn't changed at all. You've just added vocabulary. The incentives will tell the truth your strategy deck won't.
Why this is an existential threat, not an optimisation problem
Here's where I'll be blunt, because hedged takes are useless to anyone actually doing this.
A world-class, genuinely relevant agentic product is no longer enough to win. That's the part that's hard for great product organisations to accept. You can have the best agent in your category and still lose, because your commercial operating model is leaking value at every seam, selling access in a world that pays for action, optimising for moments in a world that runs on motions, walling off the people who sell from the people who ensure value is realised, and paying everyone for behaviours that made sense in 2018.
This is not an optimisation problem you can defer to next year's planning cycle. It's structural. The companies that treat operating-model and incentive redesign as the strategy, the way Snowflake did — will build a different kind of business: a forever business, with customers genuinely at the centre, where the commercial model gets stronger every time a customer extracts more value. The companies that treat it as an afterthought a comp tweak bolted onto last year's org chart will spend the next two to three years watching their net revenue decay while their product team can't understand why a great product isn't translating into durable revenue.
The product was necessary. It was never sufficient. The operating model is where this is won or lost. And your comp plan is the clearest, most honest signal of whether you've actually changed it or just told yourself you have.
Next issue: NRR Is Lying to You — why your retention metric is masking structural decay. If you're using net revenue retention to reassure the board that the agentic transition is working, there's a strong chance the number is hiding exactly the problem this issue describes.
First published in the Agentic Commercial Model newsletter on LinkedIn, June 25, 2026. Read the original on LinkedIn.