Guides

GP Per Desk: The Metric That Actually Justifies an AI Sales Employee

Why gross profit per desk, not "hours saved" or headcount, is the number that should decide your BD automation spend - and how signal-led BD moves it more than hiring does.

TB Team Boilr

· September 1, 2026 · 13 min read

Abstract dark liquid-metal texture representing agency margin and gross profit

TL;DR

"Save 4 hours a day" does not get an AI tool past a P&L review. Gross profit (GP) per desk - the number agency owners already report to their board every month - does. GP per desk is annual gross profit from placements divided by the number of billing desks (fee earners), and UK agencies typically run somewhere between £80,000 and £250,000+ per desk depending on discipline and seniority mix [1] [2]. There are only two ways to move it: add desks, which dilutes GP per desk before it lifts it, or grow GP on the desks you already have. Signal-led BD - finding the right companies at the right moment and running more qualified conversations per consultant per week - is the second lever. This article shows the formula, the two-lever framework, how to track it, and how boilr moves GP per desk without adding headcount.

Why "Time Saved" and "Headcount Reduction" Are the Wrong Pitch

Most AI sales tools are sold on hours freed up per week. That is a real benefit, but it is the wrong unit for the person who signs the contract. An agency owner or MD does not manage "hours." They manage a P&L, and the number that P&L is built on is gross profit per fee earner - GP per desk.

"Hours saved" is unfalsifiable. A consultant who gets 90 minutes back a day can spend it on LinkedIn, on lunch, or on a genuinely better-targeted call. The owner has no way to verify which happened.
"Headcount reduction" is the wrong framing for growth-minded owners. Most agency leaders are not trying to cut consultants - 84% expect sales growth in 2026, but only 47% plan to hire more staff to deliver it [3]. The pitch they need is more output per desk, not fewer desks.
Time and headcount both skip the P&L line that matters. A tool can save time and still not move billings if the freed-up hours are not converted into qualified conversations with buyers who are actually hiring.
GP per desk is what the owner already tracks. It sits directly above net margin in most management accounts, it is comparable across the floor and across time, and it is the number used to decide who gets promoted, who gets a second desk, and whether the agency can absorb a slow quarter.

Reframe the pitch around GP per desk and the conversation changes from "will this save time" to "will this move the number I am already accountable for." That is a much easier internal sale for an owner to make to their board or to themselves.

What GP Per Desk Actually Is (and How to Calculate It)

GP per desk - sometimes called gross profit per consultant, GP per head, or fee-earner productivity - measures how much gross profit each billing desk generates in a period, usually a rolling 12 months.

The Formula

GP per desk = Total gross profit from placements ÷ Number of billing desks (FTE fee earners)

Gross profit, not revenue. On perm placements this is the fee itself (typically 15-25% of first-year salary); on contract and temp desks it is the margin after pay rate, not the full bill rate [1].
Billing desks, not headcount. Count full-time-equivalent fee earners who carry a target - exclude resourcers, ops, and delivery-only staff who do not own a P&L number.
Use a rolling 12 months, not a single month. Recruitment revenue is lumpy - one big perm placement can distort a monthly number. Annualise it or use a trailing-12 view.
Pro-rate ramping consultants. A consultant three months into their first desk should be weighted by tenure, not counted as a full desk from day one, or the average understates mature performance.

Worked Example

A 12-consultant agency books £1.44m in gross profit across the trailing 12 months. Two consultants are resourcers with no billing target, so the agency has 10 billing desks.

£1,440,000 ÷ 10 = £144,000 GP per desk.

That sits inside the typical UK range, where competent consultants generate £120,000-£200,000 in gross billings (and correspondingly less in GP once margin is applied), with specialist top performers exceeding £300,000 in billings and healthy Revenue-per-Recruiter benchmarks quoted around $250,000+ [1] [2]. A consultant billing under roughly a third of their fully-loaded cost is generally considered unprofitable [1]. Agencies that consistently push past £150,000-£200,000+ GP per consultant are usually running a disciplined mix of new-logo hunting and account expansion rather than pure cold outreach - see our guide to account expansion for recruiters for the second half of that equation.

Why GP Per Desk Beats "Billings Per Consultant" as the North Star

Agencies track a lot of per-consultant metrics. GP per desk is the one worth anchoring an AI BD investment case to, because it already nets out the variables that make other metrics misleading.

Metric	What it hides	Why GP per desk is cleaner
Billings / revenue per consultant	Ignores margin - a perm desk and a low-margin contract desk look identical	GP already reflects the real margin the agency keeps
Placements per consultant	Treats a £4,000 fee and a £40,000 fee as equal	GP weights placements by value, not count
Headcount / total GP	Rewards adding desks even if new desks under-perform	GP per desk exposes dilution the moment it happens
Activity metrics (calls, meetings)	No link to whether activity converts to fee-paying work	GP per desk is the outcome activity is meant to produce

This is also why "hours saved" fails as a pitch: it is an activity metric one level removed from GP per desk, with no guaranteed line back to it.

The Two Levers That Move GP Per Desk

There are exactly two ways to grow total agency gross profit, and only one of them grows GP per desk.

Lever 1: Add More Desks (Headcount)

What it does: Grows total GP, but a new consultant typically needs 3-8 months to reach full billing productivity [4], during which they generate little or no GP while still costing salary, NI, and management time.
The hidden cost: Fully-loaded cost per hire runs to roughly £4,000-£5,500 before onboarding and ramp-time productivity loss are added [5] - and 90-day retention for new recruiters fell to 84.6% in 2025, down from 93.9% the year before, meaning a meaningful share of that investment does not even survive the ramp [6].
The dilution effect: Every new desk that has not yet ramped drags the average GP per desk down before it ever pushes it up. Ten desks at £144,000 GP each becomes eleven desks that need to hit £1.584m combined just to hold the average - and the new hire alone will not get there for months.

Lever 2: Grow GP on the Desks You Already Have (Signal-Led BD)

What it does: Increases the numerator (GP) without touching the denominator (desks) - the only lever that moves GP per desk directly and immediately.
The mechanism: More of a consultant's week spent in qualified conversations with companies that are actually about to hire, instead of research, list-building, and cold outreach to companies with no live need.
Why timing matters: Signals such as funding rounds, executive hires, and job-posting velocity can surface 48-72 hours before a role is posted publicly, sometimes weeks earlier for funding and expansion signals - reaching the hiring manager while the need is still unadvertised and before competitor agencies have engaged.
Why it compounds without new cost: Recruiter call time hit 286 minutes per week in Q1 2026, double what it was in Q1 2024 - activity is already intensifying on flat headcount [7]. Redirecting that activity toward better-timed targets, rather than adding more of it, is where the GP actually comes from.

Manual BD vs Signal-Led BD, Per Desk

Per-desk activity	Manual BD	Signal-led BD (boilr)
Company research & list-building	2-3 hrs/day, reactive to job boards	Continuous, 10,000+ sources monitored 24/7
Signal timing	Sees the job posting when everyone else does	Often 48-72 hrs ahead of the posting, weeks ahead on funding/expansion
ICP fit of outreach targets	Gut feel, inconsistent across the floor	Scored against the agency's defined ICP before it reaches the consultant
Decision-maker contact	Manual LinkedIn search, 20-30 min/lead	Auto-enriched and attached to the signal
Qualified conversations per week	Capped by hours spent researching, not by consultant skill	Capped only by consultant capacity to verify and send
Effect on GP per desk	Flat or declining as fee pressure rises	Same desk, more qualified pipeline, no new headcount cost

How to Track GP Per Desk in Your Agency

A metric only changes behaviour if it is visible weekly, not just at year-end. Track it alongside the leading indicators that feed it.

Metric	Cadence	What it tells you
GP per desk (trailing 12 months)	Monthly	The lagging outcome - direction of travel per consultant
Qualified conversations per consultant/week	Weekly	Leading indicator - whether BD volume is healthy before GP lands
Signal-to-conversation rate	Weekly	Whether targeting quality is improving or drifting
Time-to-first-outreach after signal	Weekly	Speed advantage over competing agencies chasing the same brief
New-desk ramp curve (GP per desk by tenure month)	Monthly	Whether Lever 1 (hiring) is diluting the floor average
GP per desk by consultant tenure cohort	Quarterly	Whether experienced desks are still growing or plateauing

Pull trailing-12 GP from your finance system or ATS/CRM (Bullhorn, RecruiterFlow, Spott).
Define billing desks - exclude non-quota resourcers and delivery-only staff.
Divide, then segment by tenure cohort so ramping consultants do not distort the average.
Set a weekly leading-indicator dashboard - qualified conversations and signal-to-conversation rate, not just activity counts.
Review monthly against the trend line, not a single month, since perm placements are lumpy.

How boilr Moves GP Per Desk Without Adding a Desk

boilr is an AI sales employee for recruitment - one per consultant, working alongside the desk rather than replacing it. It runs the top of the BD motion so the consultant's time goes into conversations that are more likely to convert to fee-paying work.

Signals - detects funding rounds, executive moves, expansions, and job-posting velocity across 10,000+ sources, often 48-72 hours before a role hits a job board.
Companies - finds and enriches target client companies, scored against the agency's ICP so the consultant is not deciding fit from scratch every time.
Candidates - sources candidates against open or anticipated roles, so a signal converts into a pitch with a shortlist attached, not just a cold intro.
Tasks - turns research and sourcing into a drafted, ready-to-send outreach task in the consultant's inbox; the consultant verifies and sends.
Company Brain - a shared memory of the agency's ICP, winning angles, and outreach patterns that survives consultant churn, so a new desk inherits what the floor already knows instead of starting from zero.
Integrations - connects to Bullhorn, RecruiterFlow, Spott, calendars, and email, so GP-per-desk tracking pulls from the same system BD already runs in.

Kept deliberately human: outreach personalisation, relationship-building, discovery calls, negotiation, and closing. boilr delivers a verified, scored, ready-to-act signal - the consultant still decides how to work it.

Mistakes Agencies Make When Chasing GP Per Desk

Mistake #1: Measuring Activity, Not Outcome

Why it fails: Call counts and email volume can rise while GP per desk stays flat if the targets are wrong. Recruiter call time has already doubled since Q1 2024 without a matching jump in agency profitability across the board [7].

Fix: Track qualified conversations and signal-to-conversation rate, not raw activity.

Mistake #2: Hiring to Fix a GP-Per-Desk Problem

Why it fails: A new desk dilutes the average for 3-8 months before it can lift it, and nearly one in six new recruiters do not make it past the first 90 days [4] [6].

Fix: Prove out Lever 2 (signal-led BD on existing desks) before adding Lever 1 (headcount).

Mistake #3: Treating All Desks as Equal

Why it fails: A blended floor average of £150,000 can hide a top desk at £300,000 and a struggling desk at £60,000 that needs a different intervention entirely.

Fix: Segment GP per desk by tenure and by discipline before deciding where to invest.

Mistake #4: No Leading Indicator Before the Lagging One

Why it fails: GP per desk is trailing by definition. Without a weekly leading metric, a problem is invisible until it has already cost a quarter of billings.

Fix: Track qualified conversations per consultant per week as the early warning signal.

Mistake #5: Ignoring Existing Database and Account Depth

Why it fails: New-logo hunting alone is the most expensive way to grow GP per desk. Existing candidate databases delivered an 80% positive ROI in one 2026 survey, ahead of job boards at 72% [3].

Fix: Pair signal-led new-business BD with account expansion into existing clients - the two levers reinforce each other rather than compete for the same hours.

A 90-Day Plan to Move GP Per Desk

Days 1-30: Baseline and Define

Calculate trailing-12 GP per desk today, segmented by tenure cohort. Define billing desks precisely. Set the weekly leading-indicator dashboard (qualified conversations, signal-to-conversation rate).

Days 31-60: Instrument Signal-Led BD

Configure ICP scoring per desk or per niche. Turn on signal monitoring so the top of the funnel is automated rather than manual. Track time-to-first-outreach after each signal fires.

Days 61-90: Measure and Compare

Compare qualified conversations per consultant per week against the Day 1-30 baseline. Compare early GP movement on desks running signal-led BD against desks still working purely manually. Decide whether to scale before considering Lever 1 (headcount).

Before you approve another headcount request, calculate what signal-led BD could do for the desks you already have. Try boilr free or book a 20-minute demo to see the maths for your floor.

Frequently Asked Questions

What is GP per desk in recruitment?

GP per desk (also called gross profit per consultant or GP per head) is a recruitment agency's total gross profit from placements divided by its number of billing desks - full-time fee earners who carry a target. It measures how much each consultant actually contributes to the agency's margin, rather than raw revenue or placement count, and is typically calculated on a trailing-12-month basis to smooth out lumpy perm fees.

What is a good GP per desk for a recruitment agency?

UK benchmarks put competent consultants at £120,000-£200,000 in gross billings annually, with specialist top performers exceeding £300,000, and one widely used Revenue-per-Recruiter benchmark treats $250,000+ as healthy [1] [2]. Because gross profit is a percentage of billings (roughly 15-25% on perm, 10-20% on contract), actual GP-per-desk figures run lower than gross billings figures - the right comparison is always against your own historical trend and your discipline's typical margin.

How is GP per desk different from billings per consultant?

Billings per consultant measures gross revenue generated; GP per desk measures the margin the agency actually keeps after pay rates and placement costs. Two consultants can have identical billings but very different GP if one works high-margin perm roles and the other works thin-margin contract desks. GP per desk is the more accurate profitability metric because it already accounts for that difference.

Does hiring more consultants increase GP per desk?

Not immediately, and often not at all in the short term. A new consultant typically needs 3-8 months to reach full billing productivity [4], during which they add cost and dilute the floor average rather than lifting it. GP per desk only rises from new hires once they have fully ramped - growing GP on existing, already-productive desks moves the number faster and without the ramp-time cost.

How does signal-led BD improve GP per desk?

Signal-led BD replaces reactive, job-board-driven prospecting with early detection of buying signals - funding rounds, executive hires, expansions, and job-posting velocity - often 48-72 hours before a role is posted publicly. That lets a consultant spend their week in more qualified conversations with companies that are actually about to hire, rather than in research or cold outreach to companies with no live need, which increases GP without adding a new desk.

What counts as a "billing desk" when calculating the metric?

A billing desk is a full-time-equivalent consultant who carries an individual billing target - typically 360 consultants, or split BD/delivery pairs where the BD half owns the target. Resourcers, delivery-only staff, and ops roles without a personal target should be excluded, or the average will understate what your actual fee-earners are producing.

How often should agencies track GP per desk?

Review the trailing-12-month GP-per-desk figure monthly to smooth out lumpy perm placements, but pair it with a weekly leading-indicator dashboard - qualified conversations per consultant and signal-to-conversation rate - so problems surface before they show up in the lagging GP number a quarter later.

Can AI tools actually move GP per desk, or just save time?

Time saved on research and list-building is only a business outcome if it converts into more qualified conversations with companies that are genuinely hiring. Tools that just automate admin without improving targeting quality save hours without moving GP. Tools built around verified, scored signals - like boilr - aim the freed-up time at better targets, which is the mechanism that actually shows up in GP per desk.

Sources

Information sourced from public industry reports and benchmarks as of September 2026.

Back to all articles