Running a CAS Practice: Staffing, Cadence and Technology

A CAS practice runs on three things: dedicated people, a repeatable monthly cadence, and a technology stack that automates the transactional layer so the humans can spend their hours on review and advisory work. Get those three right and the practice scales; get any one wrong and every month-end becomes a fire drill.

This is the second post in a three-part series on building and running a CAS practice. Part 1, What Is a CAS Firm covers what CAS is and how engagements are scoped. Part 3 covers pricing models and engagement classification under SSARS 27. This post is about operations: who does the work, when, and with what tools.

Staffing a CAS practice

The single clearest finding in the profession's benchmark data is that successful CAS practices do not treat CAS as a side assignment. In the 2024 CPA.com and AICPA PCPS CAS Benchmark Survey, nearly four-fifths of practices, 78%, have dedicated staff working only in CAS. The same survey put median net client fees per professional at $156,250, up 29% from the 2022 survey, which is the economic argument for dedicated staffing: focused teams serve more clients per head.

The Journal of Accountancy's practitioner guidance on starting a CAS practice makes the same point from the firm side: ideally, have someone devoted to CAS full time. Staff who split their weeks between compliance deadlines and CAS clients will reliably sacrifice the recurring work to the deadline work, and CAS clients notice a late close faster than they notice anything else.

A workable structure for a growing practice looks like this:

• Staff accountants own transaction processing, coding, and reconciliations across a pod of clients
• A senior or supervisor runs the month-end close and prepares the reporting package
• A controller reviews every package before it reaches a client and handles the escalations
• A CFO-level advisor covers the engagements that include forecasting, KPI work, or board-level reporting

Two roles worth adding that compliance practices rarely carry, both named in the Journal of Accountancy guidance: a project manager who owns deadlines and handoffs across the client portfolio, and a client success advocate who owns the relationship health that a controller reviewing workpapers will not see.

The operating cadence: the week and the month

CAS is a rhythm business. The deliverable is not a document; it is a close that lands on the same day every month. The cadence below is a template to adapt, not a standard to cite.

Daily and weekly:

• Transactions sync from bank and card feeds and get coded, with exceptions flagged rather than guessed at
• AP runs on a fixed weekly cycle so approvals never bottleneck on month-end
• Open items go on a client-facing list with owners and dates, not in someone's inbox

Month-end:

1. Cut-off: receipts, final bills, and payroll entries in by a stated business day
2. Reconciliations: every balance sheet account, with aged items investigated rather than rolled
3. Review: the senior closes, the controller reviews, and review notes clear before anything leaves the firm
4. Delivery: the reporting package goes out on the committed date, with variances against budget or prior year annotated
5. The advisory conversation: for insights-level engagements, a standing monthly call that works from the package, not from memory

The piece most firms skip is the escalation path. Every close surfaces items that need a client decision: an uncategorized transfer, a loan that showed up mid-month, a payroll adjustment. Name the path in the engagement letter: who at the client answers, within how many business days, and what the firm does when the answer does not come. An open question without a named owner is how a five-day close becomes a fifteen-day close.

The technology stack, by category

Tool choices change fast; categories do not. A CAS practice needs coverage in five places, and the discipline is standardizing on one choice per category across the client base rather than inheriting whatever each client brought with them.

Standardization is also what the benchmark data rewards: in the 2024 survey, practices continually investing in technology served a median of 100 clients against 67 for all respondents. If you are choosing a stack from scratch, the startup-stage view of these same decisions is in How to Start Your Own Accounting Firm.

AI in a CAS practice

The practical question is not whether to use AI but where it already sits in the stack and what to pilot next. The most grounded published framework comes from CohnReznick's CAS leaders, writing in the Journal of Accountancy in January 2026 on AI use cases for CAS. Their sequence is worth following as written.

First, audit the workflow before buying anything: map the core processes across the client life cycle and your top clients, and flag the repetitive tasks, data bottlenecks, and error risks. Second, identify the AI already embedded in tools you own, from the accounting platform to the office suite, before adding new vendors. Third, pick one high-friction workflow, AP being the obvious candidate in most CAS practices, and run a 30-to-60-day AI automation pilot, measuring time saved, added review time, error reduction, and client feedback before expanding.

Their cautions deserve equal weight: do not over-automate past the point of review, do not treat AI output as a substitute for professional judgment, and do not neglect data governance when client financial data is involved.

One workflow that framework does not cover is the tax question that arrives mid-engagement. CAS puts the firm in the room for operating decisions, and operating decisions raise tax questions: an equipment purchase, a new state, an owner comp change. Marble's Intelligence agent is built for that moment: ask the federal or state tax question in plain English, get a citation-backed answer linked to primary authority, and generate a client-ready memo or email for your review, all without pulling a tax partner into the close calendar. Sign up for Marble.

Frequently asked questions about running a CAS practice

How many staff does a CAS practice need?

There is no published headcount-per-client standard, and any ratio you see quoted is someone's internal rule of thumb. The benchmark evidence points to structure rather than size: 78% of practices in the 2024 CPA.com and AICPA PCPS survey have staff dedicated solely to CAS, and dedicated teams reported higher fees per professional. Start with dedicated people, even if that is one person, and layer review roles as the client count grows.

Do you need a CPA to run CAS engagements?

Much of the transactional layer is performed by skilled non-CPA accountants, and no separate CAS license exists. A CPA matters at the controller and CFO layers, for credibility in advisory conversations, and whenever the work approaches financial statement preparation or attest territory, where professional standards and the independence rules come into play.

What software does a CAS practice need?

Coverage in five categories: a cloud general ledger with a managed close process, AP and AR automation, payroll integration, practice management for workflow visibility, and a standardized reporting layer. The specific vendors matter less than standardizing one stack across clients so staff work the same way on every engagement.

How should a CAS practice start with AI?

Follow the published practitioner sequence: map your workflows first, inventory the AI already in tools you own, then run a 30-to-60-day pilot on one high-friction workflow and measure the results before expanding. Keep human review on everything client-facing and treat data governance as a prerequisite, not an afterthought.