For accounting practices
Launch an advisory arm in ninety days, without hiring a CFO.
Your clients already trust you with last year's numbers. They will pay considerably more for next year's. XLCFO gives your people the platform, the method and the training to deliver forward-looking advisory work — and a Chartered Accountant to escalate to when an engagement outgrows them.
The problem
Advisory is every practice's strategy and almost no practice's capability.
Compliance fees are under structural pressure from automation and price competition. Advisory is where the margin went, and the profession has known it for a decade. Most practices have a strategy document that says so.
Very few have made it work — and not because the people are not capable. It is because advisory is a different discipline. A manager who has spent fifteen years preparing accurate financial statements has never been taught to build a three-way forecast, structure a raise, sit opposite a founder and tell them the plan does not fund, or write a board paper that a director will act on. Those are learnable. Nobody hands you the method.
That is the gap. Not software, not enthusiasm — a method your people can follow, with somewhere to escalate when a client outgrows it.
Where we stand
Four commitments, in the agreement, not just on this page.
You are being asked to put a supplier in front of your client base. That is a real risk and it deserves more than reassurance. These are contractual.
We do not touch compliance
We do not prepare statutory accounts, we do not file tax returns, and we are not a tax agent. Your practice remains the accountant of record and keeps the compliance relationship entirely. We work in the forward-looking layer above it.
We never approach your clients
A client you introduce is yours permanently. We will not market to them, will not accept them as a direct engagement, and will not contact them outside the work you have asked us to do. No expiry on that.
Escalation returns the client
When a client needs a sale process, a raise or a covenant reset and we step in, the engagement is scoped, delivered and handed back. You keep the relationship and a share of the fee. Every time.
Our direct work is a different segment
We retain a small number of direct engagements: large multi-entity groups mid-transaction — the work a practice would not staff. The boundary is defined by objective criteria in the agreement, not by our judgement in the moment.
What a practice licence includes
Not software. An advisory arm, assembled.
Plenty of vendors will sell your practice a forecasting tool and leave you to work out how to use it, who delivers it and what to charge. Most of those licences go unused. The tool is the easy part.
The platform
Nine supervised agents running the forward-looking work: month-end close, rolling 13-week and 12-month cash, a driver-based three-statement model, board and investor packs, covenant testing and capital readiness. Running in your workspace, on your clients' ledgers, read-only by default.
The method
The part that cannot be bought elsewhere. Engagement files, materiality thresholds, the approval matrix, run playbooks and the review gate — codified procedures that make a competent accountant produce a senior pack. See the architecture.
Certification
Training that takes a compliance manager to a competent business partner: how to run the diagnostic, what to ask a founder, how to read the exception queue, when to escalate. Assessed, credentialled and refreshed annually.
Your client portal
White-labelled to your practice. Your brand, your logo, your client relationship. Position, runway, close status, the exception queue and the deliverables library — refreshed every cycle. XLCFO's name appears nowhere your client can see it.
The commercial kit
Pricing guidance, proposal templates and the diagnostic that opens the conversation. Practices are historically poor at pricing advisory and it is the most common reason an advisory push stalls. We give you what to charge and how to justify it.
A Chartered Accountant to escalate to
When an engagement exceeds your team — a covenant reset, a sale process, a cross-border structure, an investor's diligence list — a named CA takes it, scoped and priced, and hands the client back. This is the part that lets your people say yes to work they would otherwise decline.
How an engagement runs
The structure is the product.
Four stages, the same every time, on every client. Defaults you can override, and nothing proceeds on an assumption nobody agreed to.
Diagnostic
Run over the client's last closed period, fixing nothing. What is unreconciled, what is unsubstantiated, what is coded inconsistently, which obligations are overdue. The output is an opening findings report — and it is usually the document that sells the engagement, because it shows the client something their current process did not.
Half a day · Produces the proposalEngagement file
Entities, balance dates, functional currencies, materiality thresholds, variance tolerance, the approval matrix, the board cash floor, covenant definitions transcribed verbatim, the reporting calendar and who signs. Agreed with the client in writing before anything runs. Nothing is inferred — a threshold nobody agreed to is worse than no threshold, because the client will assume it was agreed.
Blocks every run until complete · By designThe monthly rhythm
Ledger and reconciliation, then close, compliance calendar and treasury in parallel, then the forecast re-run and scored against its own last version, then the pack. Governance updates open board actions and statutory dates. Status is visible while it runs rather than reported after it.
Close, cash, pack, exceptions · Every cycleReview and sign-off
The supervising agent cross-checks every figure that appears in more than one place — cash per treasury against opening cash per forecast against reconciled bank, revenue in the pack against the closed ledger. Where two disagree, the pack does not go out and the disagreement is the finding. Then a named person signs.
Ten cross-checks · Nothing leaves unsignedDecision gates
Every judgement surfaced, never buried.
The failure everyone fears with automated finance work is a confident number nobody can explain. We designed against it from the start.
Where the work requires a judgement — a provision, a capitalisation decision, a revenue recognition question, an arguable tax position, an unmatched intercompany balance — it does not get decided quietly. It stops, it is presented as a gate with the options and the consequence of each, and your person chooses. The choice, the reasoning and the timestamp are written to a permanent record.
Two consequences worth having. Your manager builds judgement because they are making the calls rather than receiving conclusions. And when an auditor, a lender or a client asks in eighteen months why a number is what it is, the answer exists, with a name and a date against it.
Commercials
You are charged on what you use, not what you hoped to use.
A licence that costs the same whether you run two clients or forty is a licence that gets cancelled at renewal. Ours scales with your adoption, which means we only do well when you do.
Practice licence
A flat monthly fee covering the platform, the full method library, the white-labelled portal and your first certification seats. Annual term.
Per active client
A monthly fee for each client you have live on the platform. Nothing for a client you have not onboarded. This is the bulk of what you will pay and it grows only as your advisory book does.
Escalation, when used
Project-priced when you hand something up to a Chartered Accountant. You keep the client and a referral share. Nothing standing, nothing minimum.
Proof before commitment
Run us in parallel on one live client, for one cycle.
You should not take our word for any of this, and we would think less of a practice that did.
How the parallel run works
Pick one real client mid-complexity — not the easiest, not the worst. Your team does the month as they always would. We run the same month alongside, on the same data, without seeing your output. At the end both are put on the table and compared: what each found, what each missed, how long each took, and what the client would have paid for either.
What you are actually testing
Not whether the software works — it will produce something either way. Whether the exception queue is correctly scoped, whether the judgement calls put to you are the right ones, and whether the pack is something you would be comfortable signing your practice's name to. Those are the questions that matter and they are only answerable on a real file.
Straight answers
The questions practices actually ask.
Are you going to become our competitor?
It is the right question and the honest answer is that the structure has to prevent it, because goodwill will not. We do not do compliance at all, we never accept a client you introduce, and escalated work is returned. Our own direct engagements are defined by objective criteria in the agreement — size, entity count, transaction status — so the boundary cannot move quietly. If we ever breach that, you should terminate and tell the profession.
Why would we pay you? We could give our managers Claude and write our own prompts.
You could, and some practices will. A few will do it well. The honest position is that the prompts are the visible ten per cent — what takes the time is the engagement file structure, knowing which four materiality thresholds to set and where, the escalation matrix, the ten cross-checks that catch one agent contradicting another, and the run playbooks. That is years of finding out what goes wrong, and the hard part is not writing it down, it is knowing which questions you failed to ask.
Two things you cannot build yourself at any price. A single assistant session cannot cross-check nine specialists against each other, because there is only one of it — that is an architecture, not a prompt. And when a client asks in two years why a figure is what it is, a chat history is not an audit trail and "the assistant did it" is not a defensible answer.
The third is the one that matters most: a general assistant will not carry your professional indemnity cover, and it cannot take the covenant reset or the sale process off your desk when a client outgrows your team. That is what the escalation path is. If you would rather spend eighteen months discovering all of this yourselves, you should — and we would rather say so now than sell you something you do not need.
Our people are compliance-trained. Can they actually deliver this?
Yes, with the method and the training, and no, without them. That is precisely why the licence includes certification rather than just access. The realistic picture: a capable manager runs their first engagement with support, their second largely alone, and is independent by the third or fourth. What they will not do in that time is handle a sale process or a covenant negotiation — that is what the escalation path is for.
How much of this is AI, and what happens when it is wrong?
The agents do the gathering, the reconciling, the modelling and the drafting. They do not decide anything material and they do not sign. Every judgement is put to a person as a gate, every figure traces to a source document you can open, and the supervising layer cross-checks the pack against itself before anyone sees it. It will still be wrong sometimes — every process is. The design question is whether errors surface before a client sees them, and that is what the review gate exists to do.
What access do you need to our clients' systems?
Read-only to the ledger, and that is the default we prefer to keep. Any write access is named explicitly in the engagement file, limited to stated operations, and revocable by you at any time. We never initiate a payment, file a return or send anything to a third party — those sit outside the agent layer by design, not by configuration.
Who owns the client data, and what happens if we leave?
Your client owns their data and you own the relationship. On termination you receive the data, the deliverables, the substantiation packs and the full audit log in a usable format within a stated window. We would rather you could leave easily and chose not to.
We tried an advisory push before and it did not stick. Why is this different?
Usually it did not stick for one of three reasons: nobody owned it, nobody knew what to charge, or the first engagement was too hard and put everyone off. We address all three — a named person and a named first client inside thirty days, pricing guidance in the kit, and a diagnostic that makes the first conversation easy. If we cannot get your first client live in a month, something is wrong and we would rather confront it then than at renewal.
Which clients is this for?
Owner-managed businesses with enough complexity to need a forward view and enough scale to pay for one — typically those with debt, growth plans, multiple entities, or a transaction on the horizon. It is not for a one-person contractor company, and we will tell you when a client on your list is not a fit rather than take the fee.
Start with one client and one cycle.
Thirty minutes to see whether the method fits your practice, then a parallel run on a live file. No licence commitment until you have seen both packs side by side.