Work we can do without the client

One line that does not move, however much of their data we hold. Setting up a mailbox for a client, holding their authentication code and knowing their National Insurance number lets us do the administration on their behalf. It does not let us be them. Passing an HMRC identity check in a director's name, or signing a declaration as them, is impersonation whoever set up the email account — and on a VAT registration it is a declaration with their name on it. Everything in the red lane below stays in the red lane. Everything else is genuinely ours to get on with.

Lane 1 — do it now, nobody to wait for

Lane 2 — prepare it all, then ask once

Do every part that does not depend on the missing piece, then send one request rather than three. Most of the delay in this portfolio has come from asking for things one at a time.

Lane 3 — they have to be there

Why the Gateway sits in lane 2 and not lane 1

It is the most tempting one to move, and it should not move. We hold the identity pack for the Strong-band clients — UTR, VAT, National Insurance, date of birth, postcode — and once we control the mailbox the recovery email lands somewhere we can read. Mechanically it would work.

But the questions HMRC asks at that screen are an identity check on the director, and answering them as them is not the same as doing their admin. The workable version is the same call with the director on the phone for ninety seconds while we do everything else — which is lane 2, not lane 3, and is a very different ask from "please log in and figure this out". Book it that way and the Strong band clears quickly.

Order to work it

Statutory filings we hold the code for

Confirmation statements, strike-off objections, auth code requests. Ten statements at £50 each removes five live strike-offs. Nothing here needs a reply from anyone — but director approval in writing is still required before filing, and that is one email, not a conversation.

HMRC chases

UTR orders, UTR chases, VAT chases. All done with the company number and our own submission references. Batch the calls — four in one sitting beats four hold queues.

Our own records

Director corrections, nominee reconciliation, unattributed documents. Not billable, and the reason the rest of the data cannot be trusted until it is done.

Portal, then social, per client

The portal build creates the Google account. The social channels run off that account, so doing them in that order means one login instead of five. See the timings below.

Then, and only then, the single client request

By this point the only things left are the ones genuinely needing them. One email per client covering all of it, not a drip.

Portal and social — how long per client

Stage	Our time	Needs the client?	What happens
Prep	10 min	No	Read that client's `INSTALL.html`, check the config, check the brand pack. Most of the build is already done in advance.
The install itself	35 min	No	Eight steps, timed in the runbook: Apps Script project, paste the bundle, run setup, form, calendar, suppliers, Business Profile, first posts.
The three unknowns	—	Yes	Which mailbox it runs from, bank details for the invoice footer, and where they actually trade. Only they know these.
Handover session	45 min	Yes	One call. The agenda is the three unknowns — treat them as the meeting, not as prerequisites for it.
Social channels	75 min	No	YouTube, Google Business Profile, Facebook Page, Instagram, X — all from the Google account created above. Plus the AI drafting queue.
Wrap-up	10 min	No	Record handles and recovery routes, log completion, send the handover note.

So: about 2 hours 15 minutes of our time per client, of which 45 minutes is the client on a call. At ten tasks a day, portal plus social across all 44 is admin-days — but the real constraint is not our time, it is getting 44 directors onto a 45-minute call. Book the sessions in waves and do the social work while waiting; it needs nobody.

The install can run before the session. Everything except the three unknowns is buildable in advance, so the client sees a working system rather than a blank project. That is worth more than the setup fee.

Five clients do not get a public brand yet. Blue Tech, CRP, Grays, Pinegrove and Willow Croft are on a live strike-off or have no director. Putting a company on five social platforms while it is being struck off creates a public shopfront for something that may not exist in two months — and the handles have to be recovered afterwards. Their portals and channels stay built and waiting until the register is fixed.

What it is worth

Two-thirds of the queue and nearly all of the billable value can be started today without a single client reply. That is the argument for doing the register and record work first: it is the part nobody can block.