HandoffForge — Client Onboarding Command Center

The onboarding wait is the one interval a practice never measures and every client remembers.

Not because it is unmeasurable — because the question everybody asks about it is the wrong one. “How long does onboarding take?” you already know, roughly, and roughly is why nothing changes. HandoffForge asks a narrower question that can be answered exactly: of the days between the money being committed and the first billable hour, how many were yours — and the answer is almost never the one the story in your head gives.

The number nobody computes

Five courts over the same interval, and they are a partition of the calendar rather than a model of one. Every single day is charged to exactly one of them — yours (you had not yet asked, or you were holding what had already arrived), theirs (you had asked in writing and were waiting), booked (a date both sides agreed to and neither could move), external (a bank, a portal, an IT department) and unattributed (an ask with no timestamp, printed and never folded into theirs). The five sum to the whole with no residual, which is why there is no weight anywhere in the application and nothing to tune. In the worked consultancy that ships with it, 33% of a 631-day book was the practice’s own court, not the client’s.

The rule that makes it honest

An item enters the client’s court at the timestamp of the first written ask, and not one hour earlier. Every hour before that is yours by definition, however reasonable the reason — there is no field anywhere in the application in which to type “I was busy”. Chasing is never your court either: once you have asked, the baton is theirs however irritating the wait. And ties are charged to the client on purpose, so a large figure in your own column cannot be an artefact of the rule.

The wait that began before anybody was asked

Every item you request has a received date, because receiving is an event with an attachment on it. Almost none has a first-asked date. In the sample practice the mean gap from commitment to the first written ask is 17.3 days; the gap from that ask to the thing arriving is 5.1 days. 77.2% of the wait on client-held items happened before the client knew there was anything to send. Both numbers are “days before the next thing happened”, measured the same way on the same items — the one like-for-like comparison in the file, and the application says so on the screen that would otherwise tempt you into a different one.

What actually set the start date

On an onboarding waiting on six things, five of them are irrelevant: the date the work began was set by whichever arrived last. That item is the gate, and the figure worth having beside it is the slack — how long the runner-up had already been sitting there. A gate with one day of slack behind it is not a gate, it is a queue, and fixing it would have bought you a day. In the sample, Data export was last to arrive on 3 onboardings with a mean of 4.0 days behind it.

The intake pack, measured over the union

A pack grows by accretion: every past disaster adds a question and nothing is ever removed. The hit rate here is measured over the union of what you asked for and what the work turned out to need, which is what makes it impossible to game — a rate over the pack alone could be raised to a hundred per cent by asking for one thing, and adding a question nobody opens makes this number worse rather than better. In the sample: 84 items requested, 56 ever referenced by the work, 17 things the work needed that the pack never asked for, and a hit rate of 55.4% over the union of both mistakes.

The screen that argues with the rest of the application

Referrals look faster than outbound. Small clients look faster than enterprise. Almost none of it survives contact with the arithmetic. HandoffForge applies the standard sample-size formula and prints, beside every pair of groups, how many onboardings it would take in each before a difference that size could be told from luck. In the sample practice, 0 of 15 comparisons are settled and the rest would need more onboardings than most practices sign in a decade. That is not a defect in the arithmetic; it is the reason these decisions are made on judgement, and this is the only tool of its kind that says so out loud.

What it refuses to do

This is the part worth reading before you buy. No split for an onboarding under 4 recorded handoffs. No book-wide share under 8 completed onboardings. No median for an item under 5 observed waits. No turn-around distribution under 6 spans on each side — and even above that floor it will not race the two medians, because a span in the client’s court bundles every ask outstanding at the time and a span in yours does not. No rate is ever applied to the days work stood still: that is deferral, not damage, and what you were doing instead is not something the application knows. And there is no benchmark anywhere for what a reasonable onboarding takes, because there is no such dataset that is not made up — every threshold on the scorecard is one you set.

What you get

One HTML file. 21 screens, 9 calculators that load your own figures off the ledger at the press of a button, 6 ready-written documents — the onboarding email that actually carries a date, asking for the thing you forgot, chasing without apologising, a kickoff agenda that produces a start date, telling a client the date has moved including the half that was yours, and a post-mortem to yourself — plus a guided tour and a printable review. Double-click it and it opens in your browser. No account, no installation, no network request of any kind: it works with the wi-fi off, and no client name, fee or date of yours ever leaves the machine.

A 21-sheet companion workbook. Every conclusion in it is a live formula, not a picture of a number. Relabel one span on the handoff ledger, or change when the clock starts on Assumptions, and the court split, the standing balance, the sample-size arithmetic and the scorecard all follow. There is a CHECK row that is zero or the decomposition is broken — a formula, not a claim — and a final sheet listing what the workbook refuses to compute and why.

Two guides and a licence. Twenty minutes to the number a practice has never put a figure on, and a screen-by-screen reference that spends as much time on where the software stops as on what it does.