Get Started
Start an AI agency

Mission 8 of 8

Scale your agency without burning out, and see your capacity month coming.

Around client eight, delivery, retainers and outreach add up to more than a month. You'll build the P&L model that shows when, then plan the first contractor, the hiring tests and the next 90 days.

80 minIntermediateFreeStep 1 free

What you'll do

Build your twelve-month P&L

Run a model on your own numbers to find the month your hours run out.

Choose and test your help

Decide contractor or hire, understand IR35 for a small UK agency, and score candidates with five practical tests.

Delegate and plan 90 days

Hand over delivery using your SOPs, weigh up turning the package into software, and fill in a 90-day plan.

  • The honest economics: where margin comes from, and what does and doesn't scale
  • Four weekly numbers and three things you'll stop doing
Before you start
  • Your real numbers: setup fee, retainer, hours per setup so far, hours per client per month, pipeline close rate
  • Python 3 for pnl.py (or build the same model in a spreadsheet; the formulas are in the script)
  • The ten SOPs from mission 6, or at least the first three. Delegation without them isn't delegation, it's hoping
Your progress0/6
01/06Build the P&L model and find the month you run out of hours0 done
STEP 01 of 06~20 min

Build the P&L model and find the month you run out of hours

A twelve-month model on your numbers that shows revenue, margin and, most importantly, your hours per month.

Around client eight, something changes. Delivery hours, retainer hours and your protected outreach blocks add up to more than a month. The way through is arithmetic first, then delegation, then a decision, so we start with the arithmetic.

A service business has one input that doesn't scale: your hours. Every other line in the P&L can grow.

So the model's job is to show you, before it happens, the month where your delivery hours plus your sales hours pass what you can actually work. Everything in this mission follows from that month.

Build the model

  1. Save the model below as pnl.py. Edit the top block with your numbers from the first seven missions; the rest is arithmetic.

    It prints a table and writes pnl.csv so you can chart it.

    pnl.py
    # pnl.py  -  a 12-month agency P&L you can argue with. Edit the top block.
    # Run: python pnl.py            Writes pnl.csv for the spreadsheet.
    import csv
    
    new_clients_per_month = [1, 1, 1, 2, 1, 2, 2, 1, 2, 2, 2, 2]   # from your pipeline
    churn_pct_per_month   = 0.03          # share of clients lost each month
    setup_fee             = 2900
    retainer              = 550
    tool_cost_per_client  = 45            # passed through at cost + 20%
    
    your_hours_per_setup  = 14            # falls with each client; start realistic
    your_hours_per_client_month = 2.5     # monitoring, tuning, report, support
    contractor_rate       = 35            # £/hour, if you delegate delivery
    delegate_from_month   = 7             # 0 = never; month you start handing setups over
    contractor_share      = 0.7           # share of setup hours the contractor does
    fixed_costs           = 260           # n8n VPS, software, insurance/12, accountant
    your_sales_hours_per_month = 32       # the protected outreach blocks
    your_capacity_hours   = 140           # hours you can work a month, honestly
    
    rows, clients = [], 0.0
    for m, new in enumerate(new_clients_per_month, start=1):
        clients = clients * (1 - churn_pct_per_month) + new
        setup_rev = new * setup_fee
        retainer_rev = clients * retainer
        tools_rev = clients * tool_cost_per_client * 1.2
        tools_cost = clients * tool_cost_per_client
        setup_hours = new * your_hours_per_setup
        delegated = setup_hours * contractor_share if (delegate_from_month and m >= delegate_from_month) else 0
        contractor_cost = delegated * contractor_rate
        your_delivery_hours = (setup_hours - delegated) + clients * your_hours_per_client_month
        your_hours = your_delivery_hours + your_sales_hours_per_month
        revenue = setup_rev + retainer_rev + tools_rev
        costs = tools_cost + contractor_cost + fixed_costs
        profit = revenue - costs
        rows.append(dict(month=m, clients=round(clients, 1), new=new,
            revenue=round(revenue), retainer_rev=round(retainer_rev),
            costs=round(costs), profit=round(profit),
            margin_pct=round(profit / revenue * 100) if revenue else 0,
            your_hours=round(your_hours), over_capacity=your_hours > your_capacity_hours,
            profit_per_your_hour=round(profit / your_hours) if your_hours else 0))
    
    with open("pnl.csv", "w", newline="", encoding="utf-8") as f:
        w = csv.DictWriter(f, fieldnames=rows[0].keys()); w.writeheader(); w.writerows(rows)
    
    print(f"{'m':>2} {'clients':>7} {'revenue':>8} {'retainer':>8} {'profit':>7} {'mgn':>4} {'hrs':>4} {'£/hr':>5} over?")
    for r in rows:
        print(f"{r['month']:>2} {r['clients']:>7} {r['revenue']:>8} {r['retainer_rev']:>8} "
              f"{r['profit']:>7} {r['margin_pct']:>3}% {r['your_hours']:>4} {r['profit_per_your_hour']:>5} "
              f"{'YES' if r['over_capacity'] else ''}")
    print(f"\nYear: revenue £{sum(r['revenue'] for r in rows):,}  profit £{sum(r['profit'] for r in rows):,}  "
          f"recurring at month 12: £{rows[-1]['retainer_rev']:,}/mo")
  2. Run it from the terminal.

    Terminal
    python pnl.py

    With the defaults (a modest pipeline, 3 percent churn, your setup hours already down to 14, delegation starting month 7) it prints something like this:

    Output
     m clients  revenue retainer  profit  mgn  hrs  £/hr over?
     1     1.0     3504      550    3199  91%   48    66
     2     2.0     4090     1084    3741  91%   51    73
     3     2.9     4658     1601    4267  92%   53    80
     4     4.8     8713     2653    8236  95%   72   114
     5     5.7     6330     3123    5814  92%   60    97
     6     7.5    10335     4130    9737  94%   79   124
     7     9.3    11407     5106   10043  88%   64   158
     8    10.0     8943     5503    7890  88%   61   129
     9    11.7    12870     6438   11397  89%   70   164
    10    13.4    13865     7344   12319  89%   74   167
    11    15.0    14832     8224   13213  89%   78   170
    12    16.5    15769     9077   14080  89%   82   172
    
    Year: revenue £115,316  profit £103,936  recurring at month 12: £9,077/mo

How to read the table

Read it in this order. The hours column. With these defaults you stay under 140 all year because delegation starts in month 7; set delegate_from_month to 0 and run it again, and watch the hours climb past capacity around month 9 to 10.

That's your capacity month. It's the date by which you need a contractor delivering, which means you need one hired a month before, which means you need the SOPs written a month before that.

Profit per your hourThis is the number that tells you whether the business is working. It should rise every quarter as setup hours fall and recurring revenue grows; if it's flat, you're growing revenue by working more, not by building anything.

Recurring at month 12The retainer line is the business. Setup fees are cash flow.

The honest part

Margins in the 90s above look wonderful because the model doesn't pay you. Your own time is the cost. Divide the year's profit by your total hours and that's your hourly rate, before tax.

For the defaults it's around £130 an hour, which is a healthy small agency, and that's with one person doing 50 to 80 hours a month plus a contractor.

Now change churn to 8 percent, close rate down so new clients are 1 a month flat, and setup hours back to 22, and watch what happens to the year. Run the pessimistic case. It's the one to plan from.

What scales and what doesn't

Here's the split:

  • What scales: retainers (they accumulate), tool pass-through (it tracks clients), the SOPs (they get used more), referrals (they compound with clients).
  • What doesn't: your hours, custom work, anything not in the package. Every hour on something outside the template is an hour the model didn't plan for.
  • Where margin comes from: productised delivery. The fourteen-hour setup that was 25 hours for client one. The two-and-a-half-hour client-month that was six. That's it. There's no trick.
Warning
The numbers in the table above are the script's output on illustrative defaults, not a forecast and not what anyone earned. Your pipeline, churn and hours will be different. The point of the model is to run it on yours.
Checkpoint
pnl.py runs on your numbers, you've written down your capacity month with delegation off, and your profit per hour for the pessimistic case. That month is the deadline the rest of this mission works back from.
If it breaksMy profit per hour is under £40 even in the optimistic case + 1 more
My profit per hour is under £40 even in the optimistic case
One of three things: setup hours are still too high (you're not reusing the template; go back to mission 6's SOPs), the retainer is too low for the delivery hours (mission 2's floor rule), or the pipeline is too slow (mission 4's blocks are being skipped). The model will tell you which if you change one input at a time.
I don't want to hire anyone
That's a legitimate choice and the model supports it: set delegate_from_month to 0, find the capacity month, and that's your client ceiling. Price above it (raise Standard, drop Starter) so the ceiling is a good income. A deliberate ten-client agency at a strong retainer is a fine business. Just make it a decision, not a drift.
Still stuck? Ask the community