The Audit
A measured account of your firm, in two weeks
A firm calls because the owner is drowning. They describe the symptom — we need to automate intake, our CRM is a mess, we should be using AI. Every one of those is a proposed solution wearing the costume of a problem. The audit replaces your theory of your business with a measured account of it.
What it is
- A measured baseline — where the hours go, how many times information is re-entered, where work waits and for whom.
- A ranked list of findings, each with an hours-per-month figure and a confidence rating attached.
- A recommended build — the top two or three items, scoped, priced fixed, acceptance criteria written before any work starts.
- A list of what we don't recommend, and why.
What it isn't
- Not an implementation. Nothing gets built during the audit — not even the easy thing.
- Not a technology assessment. We're not grading your software, we're following your work.
- Not a headcount study. We don't recommend cutting people, and we say so out loud at kickoff.
- Not an AI strategy. AI is the last of four layers, and in a meaningful number of engagements the right answer contains very little of it.
Six instruments
Not an interview, and not a questionnaire
Six specific things done in a specific order, five of which produce a number.
The Trace
Follow one real, completed instance of work end to end. Who touched it, in what system, what they had to look up elsewhere, how long the step took — and how long it then waited. Two workflows minimum, four maximum.
The Count
Four integers pulled from each trace: touches, re-entries, systems crossed, waits. These become the spine of the executive summary and the before-figures every later claim is measured against.
The Ledger
Handling time per instance × instances per month, by role, at your loaded rates. Every figure carries a confidence rating from 1 to 5 — published next to the number, so when we say a figure is a 5 you know why.
The Map
One page: every system you pay for, what it holds, who administers it, what it costs, and what connects it to the next one — where the honest answer is often a person's name.
The Shadow
One question to everyone: what do you keep track of somewhere other than where you're supposed to? Shadow systems aren't misbehaviour. They're documentation of where the official system failed, written by the person who had no other option.
The Interview
Three levels, three incompatible accounts of the same firm. The gaps between them are findings. We interview the people doing the work without the owner in the room — and we agree that in the proposal, before it's awkward.
Scoring
The Reclaim Score
The point isn't precision. It's that the ranking becomes explicable to a client who disagrees with it.
Reclaim = ( H × C ) ÷ ( E × R )
Confidence multiplies, it doesn't filter.
A large but poorly evidenced number gets pushed down the list rather than thrown out. That's the honest treatment of a number nobody has measured yet.
Risk divides.
A high-value change that touches billing has to be genuinely high-value to outrank a safe one. On a first engagement the cost of a visible failure isn't the rework — it's the relationship.
The deliverable
Presented, never emailed
Twelve to twenty pages, eight sections, written for the owner rather than for us. The readout happens in person, or on video with cameras on, and the document is handed over at the end.
A report emailed in advance gets skimmed for the price, and every finding we spent two weeks measuring reads as an opinion. Presented, the numbers land in the room — and we get to watch which one makes you sit forward.
We lead with the item you can act on the next morning, for free, without us. Not because it's the most impressive thing we found, but because it's the one that proves the rest.