> ## Content Index
> Fetch the complete content index at: https://www.elevatewithall.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# CS 2030 Part 5: Cost to Serve Is a Design Choice
- URL: https://www.elevatewithall.com/cs-2030-part-5-cost-to-serve-is-a-design-choice/
- Published: 2026-11-13T09:03:55.000Z
- Updated: 2026-11-19T20:32:32.000Z
- Description: The support-and-success benchmark covers only part of what it costs to keep and grow customers. Add account management, run the cost on two meters, and bend it with three moves: merge the roles, automate on a roadmap, and raise coverage only as automation goes live.
- Author: Sarabjeet Arora
- Tags: Customer Success, CS 2030

Part 5 of 10 · [CS 2030 series](https://www.elevatewithall.com/cs-2030/)

The CS leader's budget slide shows support and success at about 10% of ARR, close to the industry median. A good meeting. Then the CFO adds one line from the sales budget: the account managers who renew and expand the same customers. The number becomes 14.6%. Nobody had added the two before, because they sit in different budgets. (A composite.)

## **In 30 seconds**

**The problem:** benchmarks count support and success, while the account managers who grow the same customers sit in another budget, so nobody owns the full cost of keeping and growing a customer.  
**The evidence:** the median private SaaS company spends 9% of ARR on support and success, up from 8% a year earlier, even as AI tools spread (SaaS Capital, 2026).  
**The fix:** measure the full cost on two meters, merge the roles, and raise coverage only as automation goes live, **in waves that each beat business as usual within a year.** In an illustrative company, the full cost falls from 14.6% to about 12% of ARR by 2030, **while business as usual holds near 14.6%.**

Part 4 priced outcome pay. This part prices everything it takes to keep a customer, and to grow one.

## Start with the benchmark, then add who grows the account

SaaS Capital's annual survey puts support and success at 10% of ARR in 2023 (more than 1,500 companies) and 9% in 2026 (more than 1,000), up from 8% the year before. It's the right starting point. But it stops at the CS budget, and in many companies renewals and expansion are run by account managers in the sales budget, working the same customers.

**The illustrative company today ($384M existing-customer ARR)**

| Cost                                                                   | $M         | % of ARR  |
| ---------------------------------------------------------------------- | ---------- | --------- |
| Support and success: CSMs, support, onboarding, managers, ops, tooling | $39.4M     | 10.3%     |
| Account managers, their managers and equity                            | $16.7M     | 4.4%      |
| **Full cost to keep and grow customers**                               | **$56.1M** | **14.6%** |

The rule: benchmark the support-and-success line, but manage the full cost.

## Two meters, not one ratio

One ratio hides two jobs. The **keep meter** is everything it costs to hold the base: base pay, support, onboarding, programs, tooling and agents, as a share of ARR. The **grow meter** is outcome pay per dollar of net new ARR, from Parts 3 and 4: about 12 cents in the illustrative company's first year, falling toward 9 cents as each owner covers more.

Mixing them punishes growth. A strong year raises total cost because outcome pay rises, even though revenue rises faster and the closing ARR becomes next year's larger base. The rule: report the keep meter as a share of ARR, the grow meter per net new dollar, and judge the total on closing ARR.

## Three moves bend the curve

Business as usual holds the line: hiring paced to revenue, cheaper locations and cuts when growth slows keep the ratio near 14.6%. It works, but it rations coverage instead of redesigning it. Three moves do better, rolled out in waves.

**Merge the roles, and count the real saving.** One owner replaces the account manager and the CSM (Part 2). But merging 50 AMs and 50 CSMs doesn't leave 50\. In the model, duplicate work and tasks agents take over turn 100 roles into about 87 in the merger year and about 65 by 2030\. The evidence spans a wide range. Bain found CSMs spend about 65% of their time on low-value work AI could automate, which is a ceiling. Salesforce's 2026 State of Sales found sellers spend 40% of their time selling and expect agents to cut research time by 34%. Measured savings are much smaller: an NBER study of chatbot users found 3% of hours saved, most of it spent on other work. The rule: plan on realistic task savings, about a quarter of combined effort at full adoption, not the ceiling.

![Bar chart: 100 account manager and CSM roles fall to 87 in 2027, 71 in 2028, 66 in 2029 and 65 in 2030 as agents in production rise from 6 to 17. A dashed line at 50 marks the myth that merging halves headcount.](https://storage.ghost.io/c/7a/f3/7af37721-b7fb-465b-980a-791cb1a6c4b3/content/images/2026/11/p5_roles_agents.png)

Roles fall as agents go live, not on the day of the merger. Fictional, illustrative company.

**Automate on a roadmap, not a wish list.** Decide which agents and programs must be stable in production, and by which quarter. Foundations come first: one customer record and a knowledge library. Then CRM and meeting-prep agents, ticket triage and routine resolution, guided setup and an academy. Renewal, value and agent-to-agent work comes later. In the illustrative plan, 6 agents are live by the end of 2027 and 17 by 2030.

**Raise coverage only as automation goes live.** If books grow faster than live automation can support, retention is at risk. ChurnZero’s 2026 study found 81% of CS and revenue leaders say AI cuts manual work, yet net retention has been flat since 2024 and 63% of CS leaders have no consistent way to measure AI’s impact. The rule: time saved becomes cost saved only when coverage moves, and coverage should move only when the automation behind it is live.

**Roll it out in waves, not a big bang.** Start with assist agents, then merge roles, move pay to outcomes and raise coverage, in a pilot slice of about 15% of ARR that you can read within two quarters: a segment, a pod, a region or a renewal cohort. Scale to the next slice only once the pilot beats business as usual. In the illustrative model, on the keep meter **(13.5% of ARR today, once the 1.1% of sales commission for growth is set aside)**, the pilot year costs 0.1 points of ARR (about $0.5M, or $0.1M per $100M of ARR), beats business as usual within a year and recovers its investment in about 18 months. Doing everything at once costs 0.6 points in year one (about $2.5M, or $0.6M per $100M) and ends only 0.2 points lower. **The rule:** gate each wave on results, lowest-risk accounts first.

## Myth vs. reality

| Myth                                       | Reality                                                                                 |
| ------------------------------------------ | --------------------------------------------------------------------------------------- |
| Merging AMs and CSMs halves the headcount. | About 87 of every 100 roles in year one, about 65 by 2030, if automation lands on time. |
| AI tools lower cost to serve.              | They save hours. Cost falls only when each person covers more ARR.                      |

## From the CS 2030 Toolkit: four paths to 2030

From the Cost-to-Serve & Customer Margin toolkit. A fictional, illustrative company with $384M of ARR, simulated quarter by quarter. Today's full cost: 14.6%.

| Path                         | Net retention | Closing ARR 2030 | Full cost 2030 | vs today |
| ---------------------------- | ------------- | ---------------- | -------------- | -------- |
| No cost action (stress test) | 103%          | $432M            | 16.5%          | +1.9 pts |
| Shrinking                    | 95%           | $310M            | 12.8%          | −1.8 pts |
| At target                    | 120%          | $790M            | 12.2%          | −2.5 pts |
| Above target                 | 130%          | $1,088M          | 11.9%          | −2.7 pts |

Outcomes only: the coverage, effort and roadmap logic stays in the toolkit. **The stress test shows the cost of no action at all; most companies hold near 14.6% instead.** The shrinking path improves through cuts, mostly absorbed by attrition, and still misses a 2-point goal.

## What it's worth: the CFO's view

At target in 2030, the full cost is 12.2% against about 14.6% for business as usual: 2.4 points of ARR, roughly $19M a year at $790M **(today’s $384M growing at 120% net retention)**. Rolled out in waves, it lands within a few tenths of a point of that. Assumptions: 3.5% raises, 15% attrition absorbing most reductions, a $1.5M one-time build, and paid success plans and premium support worth 1% of ARR by 2030 (net cost 11.3%). Swap in your own.

![Line chart of the cost to keep existing customers by quarter. Business as usual holds at 13.5% of ARR. A big bang rollout peaks at 14.4% in 2027, then falls to 11.2% by 2030. Rolled out in waves starting with a pilot slice, cost stays within 0.4 points of business as usual, then falls to 11.4% by 2030.](https://storage.ghost.io/c/7a/f3/7af37721-b7fb-465b-980a-791cb1a6c4b3/content/images/2026/11/p5_waves.png)

Pilot, prove, then scale: each wave beats business as usual within a year. Fictional, illustrative company.

- **Risk and guardrail:** the first quarters of each wave cost more, because the merger, the build and outcome pay land before automation pays back. Waves keep that bump under half a point. If gross retention falls more than a point for two quarters, hold the next wave.

## Red Flags: is your full cost unmanaged?

If three or more are true, it is:

- **Two budgets:** support and success are benchmarked; account-management cost is never added.
- **One ratio:** keep and grow cost are judged together.
- **Hours, not coverage:** agents are live, but ARR per person is flat.
- **Coverage first:** books grow before the automation that supports them is live.
- **Layoffs first:** reductions are planned without counting attrition.

## Objections, answered

| Objection                            | Answer                                                                                                        |
| ------------------------------------ | ------------------------------------------------------------------------------------------------------------- |
| "We're at benchmark already."        | The benchmark covers part of the cost. Add the account managers.                                              |
| "Merging the roles is a cost cut."   | Year one saves little. Savings come as automation lands, so budget the transition.                            |
| "Growth makes the ratio look worse." | Only on starting ARR. Measure on closing ARR and split the meters.                                            |
| "Payback takes two years."           | Not in waves. Each wave beats business as usual within a year and recovers its investment in about 18 months. |

## Team exercise: add it up

Bring this to your next CS and finance review:

- **Add it up:** support and success plus account-management cost, as a share of ARR.
- **Split it:** keep cost versus outcome pay.
- **Date it:** which three automations must be live by next year end, and which coverage step does each one unlock?

## Signals to watch

My view: by 2030, the best CS organisations will report one full cost to keep and grow customers, on two meters, falling while gross retention holds. Watch for:

- **Boards:** asking for the full cost, not the CS line.
- **Coverage:** ARR per owner rising each year after agents go live.
- **Planning:** coverage steps tied to automation going live, not to budget cycles.

What would prove me wrong: retention falling in companies that hit their coverage targets.

**Next in CS 2030, next week:** Verified Value. You can't pay for outcomes you can't prove, so how do the customer, the vendor and the CSM agree on one record of value? The workforce plan, forecast and operating rhythm behind this part go deeper in The Engine Room. All parts are on the CS 2030 hub.

## **Your turn**

Does anyone in your company add account-management cost to cost to serve?

The year-by-year path to 2030 and two free worksheets are free for members.

[Join free to see it ](#/portal/signup/free) 

---

If you'd like to pressure-test this against your own organization, message me on LinkedIn.

---

*Disclaimer: this is a personal perspective, written in my own capacity. It is not connected to, endorsed by or based on any information from my employer. All companies, costs, headcounts and scenarios are illustrative and fictional. Survey data is used for direction, not as benchmarks for any specific company. Views are my own.*

**Sources**

- SaaS Capital, [2023 Spending Benchmarks for Private B2B SaaS Companies](https://www.saas-capital.com/wp-content/uploads/2023/10/Spending-Benchmarks-for-Private-B2B-SaaS-Companies-2023-W1.pdf?ref=elevatewithall.com), 2023 (1,500+ companies)
- SaaS Capital, [2026 Spending Benchmarks for Private B2B SaaS Companies](https://www.saas-capital.com/blog-posts/spending-benchmarks-for-private-b2b-saas-companies/?ref=elevatewithall.com), June 2026 (1,000+ companies; support and success reported together)
- Bain & Company, [Customer Success at a Crossroads: Evolve with AI or Fade Away](https://www.bain.com/insights/customer-success-at-a-crossroads-evolve-with-ai-or-fade-away/?ref=elevatewithall.com), December 2025 (practitioner survey, August 2025, n=235)
- Salesforce, [State of Sales, 7th edition](https://www.salesforce.com/news/stories/state-of-sales-report-announcement-2026/?ref=elevatewithall.com), February 2026 (4,050 sales professionals; vendor research)
- Humlum, A. and Vestergaard, E., [Large Language Models, Small Labor Market Effects](https://www.nber.org/papers/w33777?ref=elevatewithall.com), NBER Working Paper 33777, 2025; revised in 2026 as “Still Waters, Rapid Currents” (AI chatbots, Denmark)
- ChurnZero, [2026 Customer Revenue Leadership Study](https://churnzero.com/press-release/ai-research-customer-growth/?ref=elevatewithall.com), October 2026 (580 CS and revenue leaders; vendor research)