Guides
Nine sections of a virtual assistant agency plan and the test each must pass
Nine sections of a virtual assistant agency plan, the test each has to pass, and the three pages that carry the numbers: capacity, cost and the worst month.
What to take away
- A plan for this trade is a capacity document, not a vision document. If it does not say how many delivered hours you can produce and what they cost, it says nothing.
- Nine sections do the work. Each one below carries a test. A section that fails its test is decoration.
- Write the operations and capacity sections first. The market section is the easiest to write and the least useful, which is why most plans open with it.
- A plan you will not reread in ninety days is a plan you wrote for somebody else.
This page describes structure and tests. It contains no figures, because a plan built on borrowed numbers fails at the first month-end reconciliation.
The nine sections and their tests
| Section | The test it has to pass |
|---|---|
| What you sell | A stranger can repeat your scope back, including one thing you refuse |
| Who buys it | You can name the trigger event that makes someone start looking |
| Delivery model | It says who does the work when the named person is away |
| Capacity | It converts headcount into delivered hours with stated assumptions |
| Cost of delivery | It produces a cost per delivered hour you could defend line by line |
| Pricing and terms | Price sits above that cost, and unused reserved time has a stated rule |
| Staffing plan | Each hire is triggered by a measurable condition, not by a month |
| Cash and runway | It shows the worst month, not the average month |
| Risks and stop rules | Each risk has a signal you would actually notice |
Nothing else earns a page. Mission statements, competitor grids and five-year projections are where plans go to become unfalsifiable.
Nine sections and their tests
- What you sellstranger repeats scope, one refusal
- Who buys itname the trigger event
- Delivery modelwho covers when named person away
- Capacityheadcount to delivered hours, stated assumptions
- Cost of deliverydefensible cost per delivered hour
- Pricing and termsabove cost, unused time rule
- Staffing planhires triggered by measurable condition
- Cash and runwayshows worst month, not average
The three pages that matter
| Page | The one number on it | Why the plan fails without it |
|---|---|---|
| Capacity | Delivered hours the bench can produce | Everything downstream is priced against a guess |
| Cost of a delivered hour | Loaded cost divided by delivered hours | You have no floor, so no price is defensible |
| The worst month | Cash at the end of it | The plan only works if nothing goes wrong |
Page one: capacity. Headcount, realistic delivered hours per person per month after training, absence and handovers, and the total that produces. Write the assumptions in the margin. When the plan turns out wrong, you want to know which assumption failed.
Three pages that matter
- 1Capacity delivered hours the bench can produce
- 2Cost of a delivered hour loaded cost divided by delivered hours
- 3The worst month cash at the end of it
Page two: cost of a delivered hour. Fully loaded people cost divided by delivered hours. Include supervision, tools issued, and recruiting cost spread across expected tenure. This number is the floor under every quote you will ever issue.
Page three: the worst month. Two clients pause, one assistant resigns, and one invoice pays late. Show the cash position at the end of that month. If it is survivable, the plan is fundable. If it is not, the plan needs a smaller opening, not a braver forecast.
Where the numbers come from
Every figure in the plan should have a source line beside it. There are only three legitimate kinds.
Where each number comes from
Is the figure from your own records?
use it, available from month two
is it a quote you actually received?
- Your own records.Best, and available from month two. Until then you have none, so say so rather than borrowing.
- A quote you actually received.From a tool vendor, an insurer, a bookkeeper. Attach it to the page it supports.
- A published requirement, read literally.The IRS guidance on starting a business sets out the structure choice, the identification numbers and the recordkeeping expectation. Those are obligations rather than estimates, so they can go into the plan as stated.
Anything else is a number somebody else made up, and it will be the number your plan fails on.
The compliance section people skip
A plan for a remote staffing business has one dependency that a local business does not: your obligations follow your people and your clients across jurisdictions.
Two paragraphs cover it. First, worker status, which the Labor Department's page on misclassification myths explains is settled by the economic reality of the relationship rather than by the label on the agreement. Second, the wider registration and insurance layer, mapped in the SBA business guide and by the state or states you and your clients sit in.
Write what you have settled and what is still open. An open item with an owner and a date is a plan. An open item nobody has named is a surprise.
Staffing triggers instead of a hiring schedule
Do not write "hire the second assistant in month five". Write the condition.
Staffing triggers, not dates
- Reserved hours sold exceed stated share of capacity two months
- One person carries more than one unabsorbable client
- Training pipeline, not sales pipeline, becomes the constraint
Conditions survive contact with reality. Dates do not. How you fill the role once a trigger fires belongs in the hiring and training guide.
What to leave out
Leave out the five-year revenue curve. Leave out the competitor grid built from public websites, which measures marketing rather than operations. Leave out any claim about market size that you cannot connect to a client who would sign.
What to leave out
- Five-year revenue curve
- Competitor grid from public websites
- Market size you cannot tie to a signing client
- Pricing you have not costed
- Marketing plan before one delivered client
Leave out pricing you have not costed. The reasoning that produces a defensible price is in the pricing and profit guide, and it belongs there rather than as an assertion in a plan.
Leave out the marketing plan until you have delivered for one client. What you learn in that first month changes who you would sell to and how, and the marketing and growth guide is more useful once you have that experience than before it.
Keeping it alive
Set a review date ninety days out and put it in the calendar with the plan attached. At the review, change the assumptions rather than the conclusions. If delivered hours per person came in lower than you wrote, correct that line and let the rest of the document recalculate.
A plan that is never wrong was never specific. The order in which the opening decisions get made, and which ones are expensive to reverse, is set out in the guide on how to start a virtual assistant agency, and the plan should follow that order rather than fight it.
Keep every superseded version, dated. When a lender, a partner or your future self asks why you expected something, the old version answers honestly. And the broader setup context, formation through growth, is mapped in the startup and market guide.
Common questions
How long should the plan be?
Short enough that you reread it. Nine sections, a page of arithmetic, and the risks. Length is not evidence of thought.
Does a lender need something different?
A lender needs your worst month and your repayment source, presented plainly. The rest of the document is for you.
Should I include a competitor analysis?
Only if you have something checkable, such as what a competitor's published scope excludes. A grid of adjectives compiled from homepages tells you who writes well.
What if I have no numbers yet?
Then write the formulas with the inputs blank and say they are unknown. A plan that admits which cells are empty is more useful than one that fills them with invention.
