Margin depends on touch time
Every bill a staff member keys is margin. The practice scales linearly with headcount and the pricing rarely keeps up with wage inflation.
By role
CAS is the growth line in most firms and it scales with staff, which is the problem — the work is largely transactional, the margin depends on how much of it a person touches, and hiring is the constraint. Automating the transactional layer is the only structural answer, and it does not require moving clients onto one platform.
The situation
Every bill a staff member keys is margin. The practice scales linearly with headcount and the pricing rarely keeps up with wage inflation.
You could take more clients if you could staff them. Experienced people are scarce and juniors need supervision that consumes your seniors.
Different systems, different charts, different conventions. Method does not transfer, so training is per-client rather than per-firm.
Partner and manager review is the quality control and it is the most expensive hour in the firm being spent checking arithmetic.
Across forty clients you find out something is late when the client asks, which is the worst possible discovery mechanism.
What was done, by whom, on what basis — varying by staff member, which matters when a client is examined or when the person leaves.
In most CAS practices the work splits roughly into transaction processing, close execution, and advisory. The first is the majority of the hours and the least of the value, and it is the only part that scales linearly with headcount.
Automating it does not remove your staff — it moves them up. The firms that have done this report the same headcount serving materially more clients, with juniors doing exception handling and review rather than keying, and seniors spending their time on the advisory work that clients actually pay a premium for.
The obvious version of this is standardising every client onto one accounting system, and firms that try usually find it stalls — clients resist, migrations consume the practice, and the ones who agree are not the ones causing the problem.
The layer runs on top of QuickBooks, Xero, Sage Intacct, NetSuite, and the rest. Your clients keep what they have; your method standardises anyway. That is the arrangement that actually rolls out across a book.
The Controller Agent reviews every journal entry against pattern rather than sampling, and surfaces what is unusual with the comparison attached. That does not replace partner review — it changes what the partner is reviewing, from arithmetic to judgement.
Firms tell us this is the change that most affects capacity, because senior review time is the scarcest resource in the practice and it was being spent on the part a machine checks more thoroughly.
Where your firm also audits a client, the usual independence considerations apply to anything that looks like performing management functions. The permission model helps — agents draft and your staff commit, with the client approving — but this is a conversation for your risk partner rather than something we would advise on. We flag it rather than leaving you to discover it.
Questions
Tell us how many clients and where the hours go, and we will scope what automating the transactional layer takes.