The hard part is not writing the letters. It is knowing which clients need one, what changed since last year, and recording the decision to carry on.
Key takeaways
Treat it as a register, then a mail-merge, then a review — in that order. Most firms attempt the middle step first, which is why renewal season feels like drafting and ends up being archaeology. The tracking layer is what makes the drafting layer trivial: one row per client showing the current letter, its date and period, the scope, the fee basis, the responsible partner, and whether anything material has changed since it was signed.
Once that register exists, a model generating two hundred letters from a fixed template and a clean data source is genuinely low-risk work, because nothing is being invented. Without the register, the same tool is generating two hundred confident documents from assumptions, which is a much worse outcome than not sending them.
The letter is the visible artifact. The thing being decided is whether to carry on. CPA Alberta’s practice review alert on compilation engagements reported that practice reviews of compilation engagements across Canada found file documentation that did not adequately evidence engagement acceptance or continuance procedures under CSRS 4200, alongside missing documentation of the intended use of the compiled information and of the understanding of the entity, its systems and the basis of accounting.
The same alert tells firms, in its list of remedial actions, to update firm tools and methodology — naming the engagement letter, the compilation programs and the reporting templates. It also records that CSRS 4200 applies to compiled information for periods ending on or after 14 December 2021, and that firms performing only compilation engagements had to implement CSQM 1 by 15 December 2023, a year later than assurance firms. If your engagement letter template predates any of that, renewal season is when it gets fixed.
So a well-built renewal workflow produces two outputs per client, not one: the letter, and a short internal continuance record saying who decided to continue, when, and on what basis. The second is the one a reviewer will look for and the one nobody remembers to create.
Plenty of firms try, and there is a real argument for a standing agreement with annual scope schedules. But automatic-renewal wording is not a free pass; it carries its own drafting and notice questions, and our sister firm’s guide to auto-renewal clauses in Ontario business contracts sets out what makes those clauses work or fail in Ontario business contracts. The general anatomy of the underlying agreement is covered in its note on the key clauses in a service agreement.
There is also a professional reason to touch the file annually even where the contract does not require it. An engagement letter that has not been looked at for four years usually describes a scope the firm no longer performs, a fee basis it no longer charges, and a contact who has left. For a law firm the equivalent pressure comes from rule 3.6-1, which requires a fee or disbursement to be fair and reasonable and disclosed in a timely fashion — a rate change communicated only through an invoice is not that.
Five steps, and only one of them is generative. Build the register from your practice-management data and reconcile it against signed documents on file, which is where most firms discover clients with no current letter at all. Flag the exceptions: expired, missing, scope inconsistent with the work actually performed last year, fee basis inconsistent with what was billed. Generate the letters from the template, with variables drawn from the record. Route for partner review by exception rather than in bulk — unchanged clients get a fast approval, changed ones get a proper read. Then track signature and chase.
Vendors package parts of this. Karbon describes its Engagements feature as bringing the client engagement process together from proposals to payments, including creating engagement letters, and its eSignatures & Approvals capability as preparing documents for approval and collecting electronic signatures. Those are capability descriptions, not results; the value depends entirely on whether your register is accurate.
Illustrative. A practice runs renewals every autumn. Historically a partner drafted letters as she got to them, which meant the last thirty clients were still unsigned when the first files arrived.
The rebuilt process starts eight weeks earlier with the register, and the first discovery is the useful one: a number of clients have been served for years on a letter that no longer describes the work. Those become the exception queue and are handled first, by conversation. The remainder are grouped — unchanged, fee change only, scope change — and the letters are generated per group. The partner reviews all scope changes, spot-checks the unchanged group, and approves. Signature requests go out in waves so the chase is manageable. Each approval writes a continuance note to the file automatically, because it is a workflow step rather than something to remember.
The letters were never the bottleneck. The register was, and the review-by-exception design is what converts six weeks of drafting into two weeks of deciding.
A fee increase communicated for the first time inside a generated renewal letter is a poor use of automation. So is a scope reduction. Both are conversations, and the letter should record what was agreed rather than announce it. Similarly, do not let a model decide which clients to renew: it can rank the book by profitability, payment behaviour and scope disputes, and that ranking is an input to a partner’s decision, not a substitute for it.
And where the decision is to stop, the renewal workflow should route to a different document entirely — see how an engagement is ended for cause.
Does every client need a new letter every year?
That depends on the standard applying to the engagement and on what your existing letter says. The point of the register is that you can answer the question client by client instead of applying one habit to the whole book.
Can the model decide what changed?
It can compare last year’s letter with this year’s data and produce a difference list, which is genuinely useful. Whether a difference is material is a professional judgment.
Are electronic signatures acceptable?
Widely used in Canadian practice, and the platforms support them. Confirm what your own professional and insurance requirements say about execution and retention, and keep the audit trail.
For the underlying template and how firms automate the first letter rather than the renewal, see automated engagement letters for accountants.
A 30-minute call is enough to tell you whether AI pays for itself here.
