Ask ten CS leaders how they comp their CSMs and you'll get ten different answers, several of which will be slightly apologetic. "It's complicated." "We're working on it." "The model made sense when we built it but the business has changed." CSM compensation is one of those questions that companies get right by iteration rather than by design, which means most are somewhere in the middle of an iteration, running a model they're not fully confident in.
That's not a critique. It's an acknowledgement that the question is genuinely hard. Unlike Sales compensation, where the model is well-worn (base plus commission on revenue closed, various spiffs for pipeline activity), CS compensation has no settled consensus and a set of objectives that are harder to tie to individual outcomes.
Why CSM compensation is harder to get right than sales comp
Sales compensation works because the salesperson's contribution to a revenue event is legible. They owned the deal. They closed the deal. The revenue is attributable.
CS is different. A customer renewing is partly the CSM's relationship work, partly the product's quality, partly the sales promise made at the start, partly the wider support experience, partly market conditions. The CSM contributed, but isolating the contribution is genuinely difficult. This is why "pay CSMs on retention" sounds straightforward and is, in practice, complicated.
A few specific problems:
Attribution is murky. If a CSM has a portfolio of 30 accounts and 28 renew, is that good? Depends on the baseline. Depends on the segments. Depends on whether the two who churned were already flagged as at-risk before that CSM took them over. Raw retention metrics can be gamed (by avoiding at-risk accounts, by managing down expectations, by delaying the inevitable) or they can mislead through no fault of the CSM at all.
The time horizon misaligns. Sales comp is paid at deal close. Retention outcomes manifest over 12 months or more. A CSM who does excellent work in Q1 may not see the reward until Q4 or the following year. That's a real motivational problem.
Expansion creates a conflict. If CSMs are partially comped on expansion and partially on retention, they're structurally incentivised to prioritise accounts where they can upsell, not accounts where the relationship most needs attention. That can make retention worse, not better.
The components CSM comp usually includes (and the problems with each)
Base salary. The floor. Most CS leaders would agree that CSMs should have a higher base-to-variable ratio than salespeople: typically 70/30 or 80/20, versus 50/50 or 60/40 in Sales. The rationale is that the work is relationship-intensive and requires consistency regardless of whether renewals are due. A heavily variable comp model for CSMs creates the wrong urgency.
Gross revenue retention (GRR) component. Tying variable comp to GRR (the proportion of contracted revenue retained, excluding expansion) is the most common model. The problem is attribution (see above) and the time lag. Some teams pay this quarterly based on trailing 12-month rolling retention; others pay annually. Neither is perfect.
Net revenue retention (NRR) component. Adding expansion into the metric. This is where the conflict between retention and expansion work lives. A CSM chasing NRR may deprioritise an account with no expansion opportunity even if that account is at meaningful churn risk.
Activity-based metrics. QBRs completed, health score checks, playbook adherence. These are the "did the CSM do the job we asked them to do?" metrics. They're useful as leading indicators and as a floor for variable comp, but they have the obvious weakness that activity doesn't guarantee outcomes. You can run a perfect QBR process on an account that churns.
Customer satisfaction (CSAT / NPS). Some teams include customer satisfaction scores in comp. The problem is manipulation risk: CSMs who know their CSAT scores are tied to their earnings will avoid the hard conversations that might generate a low score. CSAT is best used as a coaching signal, not a compensation signal.
What the best models have in common
The CSM compensation models that seem to generate the least dysfunction have a few things in common.
They're simple enough for a CSM to explain in a sentence. If a CSM can't tell you what they're being comped on without a spreadsheet, the model has too many components. Complexity reduces motivation because people can't see a clear line between their behaviour and their reward.
They separate what the CSM controls from what they don't. A CSM who is set a GRR target for a portfolio that includes accounts with legacy contracts, poor onboarding history, and competitors actively pricing them out is being held accountable for things outside their control. The best models adjust for context, not to let CSMs off the hook, but to make the incentive meaningful.
They include a team component alongside an individual component. Pure individual metrics in CS create the wrong dynamics: CSMs hoard accounts, avoid helping colleagues with at-risk situations, and optimise their portfolio rather than the team's outcomes. A partial team component (even 20-25% of variable) changes the incentive structure and usually makes the team more collaborative.
They're reviewed annually, not set and forgotten. The business changes. The product evolves. The customer base segments differently. A compensation model built for a company with 50 customers and 5 CSMs is usually wrong for a company with 300 customers and 25 CSMs. Treating comp as a living model rather than a fixed policy prevents the drift.
The expansion question
Whether to comp CSMs on expansion is the most contested design question. There's a reasonable case on both sides.
For: CSMs have the best relationships. They see expansion opportunities earliest. If they're not rewarded for surfacing and supporting expansion conversations, the incentive to do so is weak.
Against: Expansion complicates the CSM's relationship with the customer. When the customer knows their CSM has a financial interest in selling them more, the dynamic shifts. The CSM's value proposition (trusted advisor) can be compromised. And expansion selling requires skills and a commercial confidence that not all CSMs have or want to develop.
The resolution I've seen work best: comp CSMs on identifying and qualifying expansion opportunities (a defined signal event), not on closing them. The commercial close goes to Sales, but the CSM is rewarded for the origination. This preserves the relationship dynamic while aligning the CSM's incentive with expansion activity.
CS Ops and comp visibility
CSM compensation has a CS Ops dimension that often gets overlooked. The data required to calculate and report on comp metrics (renewal rates by CSM, expansion signals attributed to CSM activity, CSAT by account) is operational data. Getting that reporting right, so that comp outcomes are transparent and calculable, is a CS Ops function as much as a finance function.
CSMs who don't trust their comp reporting disengage from the model. Transparency and accuracy in the data layer are a prerequisite for the incentive working.
The short version: CSM compensation is harder to design than sales comp because attribution is murky, time horizons are long, and the objectives (retain, expand, satisfy) can conflict. The models that work best are simple, separate what the CSM controls from what they don't, include a team component, and are reviewed annually. On expansion: comping CSMs for opportunity origination rather than commercial close preserves the relationship dynamic while aligning incentives. CS Ops owns the data layer that makes comp reporting transparent: without that, the model is aspirational rather than functional.
Working on this in your business?
We help UK SMEs and scale-ups turn this kind of thinking into action.
Read next
Building customer success in a company that doesn't think it needs it yet
Why tying CSM pay to renewals can backfire
When Sales and CS share the same customer: the commercial ownership problem
Building CS Before You Have CS Ops: What to Do First
The first CS hire: what you actually need versus what you think you need
