The 11-hour proposal-and-scheduling tax on a UK consultancy

At £125 an hour - the mid-tier rate for a UK independent consultant or boutique practice - 11 hours of weekly admin costs £63,250 a year per fee earner. The maths: 11 hours multiplied by £125 multiplied by 46 working weeks. At partner rates of £175 to £200, the same 11 hours costs £88,550 to £101,200.

Consultancy admin has a particular quality that makes it expensive in a specific way: it is concentrated in the highest-stakes tasks. Two to four hours on a proposal that may not close. An hour of pre-call research the client never sees. A 45-minute status email that could have been automated. These require real cognitive effort - which is what makes the standard "just use AI" advice frustrating for most consultancy readers who have already tried it.

The five admin buckets

Bucket 1: Proposal writing (2-4 hours per proposal, and half do not close)

The highest-effort bucket and the one where AI fails most visibly when used incorrectly. Most consultants who have tried AI for proposal drafting report the same experience: the output is "almost right" - which is a worse problem than "obviously wrong", because almost-right requires more editing than starting from scratch. Generic context in, generic output out.

The technique that actually works is different. Instead of feeding the AI the client brief and asking for a proposal, feed it three things: the brief, the structure and framing of a previous proposal you won on a similar engagement, and a short system prompt that defines your voice and the principles you always follow in a proposal. Then ask for a draft of the gap sections only - the parts that change between proposals, not the parts that stay constant.

That context loading - brief plus won precedent plus voice constraints - produces a draft 80% of the way there rather than 50%. Editing time drops from 90 minutes to 20. Claude Projects and Custom GPTs are both well suited to this approach because they save the system prompt and precedent as persistent context so you do not re-enter the setup for every draft. The Claude Projects vs Custom GPTs comparison covers the practical differences if you have not chosen one yet.

Bucket 2: Meeting scheduling (1-2 hours a week)

The calendar negotiation for client calls, internal reviews, and prospect meetings is consistently one to two hours of low-value email exchange. A scheduling link - Cal.com, Calendly, or the scheduling feature built into most calendar tools - eliminates most of this in a single setup under an hour. For the prospect meeting specifically, a scheduling link signals organisational competence before the meeting has happened.

Bucket 3: Project admin and status updates (2-3 hours a week)

Weekly client emails, meeting summaries, status updates. These follow a consistent structure from week to week, which makes them strong candidates for AI-assisted drafting. A Claude Project set up with the client context - objectives, milestones, communication preferences, ongoing issues - produces a weekly status draft in under two minutes. The consultant reviews, adjusts the specifics, and sends. The setup investment is front-loaded: 20 minutes to build the client context. The return is two to three hours a week for the duration of the engagement.

Bucket 4: Pre-call research and preparation (1-2 hours a week)

A brief prompt to ChatGPT or Claude ("summarise what has happened at [company] in the last three months, focusing on [relevant area]") returns a useful briefing in 90 seconds. For recurring client calls, where the context is already known, a saved prompt template that takes three bullet points of status information and returns a structured meeting prep note cuts preparation from 40 minutes to 10.

Bucket 5: Inbox management (1-2 hours a week)

Use AI drafting for emails that follow patterns you have written before: a client asking for a scope change, a prospect asking for credentials, a reference request. These are not unique messages. AI drafting speeds them up without reducing quality. Genuinely novel messages - the difficult conversation, the unusual brief, a complaint - remain human work. That is the right division of labour.

The full UK consultancy picture

The consultancy admin cost breakdown covers all five buckets with specific tool recommendations and three quick wins calibrated for a technically literate but time-poor reader. To see your own annual unbillable cost figure, the admin cost calculator returns it in a few minutes.

The "almost right" problem, stated plainly

The failure mode with AI proposals is almost always the same: generic input produces generic output. Proposals that win are specific. The AI context needs to be specific too - the client's language from the brief, your positioning for this type of work, previous won proposals as precedent. The specificity is the work. The AI handles the structure and the first draft. The final version is still yours.

Apply the same logic to every other writing task. Status updates that reflect the actual project. Pre-call research specific to the relationship. The tool generates the structure. You provide the specificity.

The maths from the other direction

At £125 per hour, a HoursBack Assessment that surfaces 5 hours of recoverable admin per week pays for itself in one half-day of recovered fee-earning time across a fortnight. The question the assessment answers is not whether the admin is costing you money - it is - but which of the five buckets is costing the most in your specific practice, and what the precise intervention is for each one.

For the strategic frame on how to think about AI investment across a consultancy practice, the AI pilot vs AI strategy post covers the positioning question that sits above the tactical fixes in this one.

Book your HoursBack Assessment - £799