Margin recovery workshop with Meridian ops leadership; built scenario model and reviewed Q3 SKU profitability cuts.
Read every entry.
Send four to a partner.
Reads every entry against engagement rules, rate cards and narrative standards — and escalates only what a partner actually needs to see.
Approving this week pushes the engagement past budget
Kim and Marks logged 24h this week — entries are clean and billable. The project sits at 94% spend with two weeks still scoped. Approving lands it at 101%. Worth a re-baseline with Meridian before clearing.
One escalation, end to end. Nothing leaves until you submit.
The problem
Why this goes wrong today.
- 01 Volume defeats review
When four hundred lines arrive at once, an approver checks whether the fields are filled in. That is not what approval was for.
- 02 The mechanical errors hide the real ones
Missing codes and wrong dates crowd out the entry that is genuinely mis-scoped — the only one that changes a number.
- 03 So everything is approved
The queue clears because it has to. The errors travel downstream into the invoice, where a client finds them instead.
How it works
What it actually does.
- 01 Read everything, not a sample Every entry on every engagement, every cycle. Sampling is what lets a bad narrative reach a client invoice.
- 02 Check against the rules that apply Engagement terms, rate cards, task codes, narrative standards and client-specific billing guidelines — whichever govern this entry.
- 03 Correct what is mechanical Wrong task code, wrong rate, a narrative that fails the standard for a reason the agent can fix without judgement. Corrected in place, with the change logged.
- 04 Escalate what needs judgement Everything else goes to a human with the reason attached, so the review is a decision rather than an investigation.
From every entry to the few that matter
One cycle: 22 timesheets and 6 expense reports. Seventeen auto-approved, five nudged, two still to submit — and four that need a partner. $142k cleared, $18.4k held.
I reviewed 22 timesheets and 6 expense reports. 4 need your call.
Inputs and outputs
What it reads, and what it does.
- Submitted and drafted time entries
- Engagement terms, rate cards and task-type definitions
- Client billing guidelines where they are configured
- The approval chain for the engagement
- Rule and rate-card compliance
- Narrative quality against standard
- What approving would cost
- Auto-approves what passes
- Corrects what is mechanical
- Escalates what needs judgement
- Returns what the author has to fix
Anything needing judgement: a week that would push an engagement past budget, a booking it cannot tell is stale or miscoded, a narrative that is accurate but reads poorly, a policy exception. It says what it could not resolve, and why.
Next step
See it run against one of your live engagements.
Bring an engagement that worried you. We will show you where the margin went, and the week Octayne would have told you.