Price what the client receives, not the minutes it takes. Keep hourly billing for unpredictable work, sell fixed-scope packages for repeatable drafting such as inbox replies, posts and minutes, and use a retainer when a client wants reserved monthly capacity. AI shortens drafting, so pure hourly billing cuts your income unless you raise the rate or change the model.
The sum is simple. If AI cuts your time on a client's work by 40%, you need 1.67 times your old hourly rate to earn the same on the same work, and a client who sees a 67% rise will question it. Packages and retainers avoid that argument by tying the price to deliverables and capacity, which stay valuable however fast you draft them.
Why hourly billing punishes the faster assistant
Hourly pricing rewards slowness and taxes skill. An assistant who takes 8 minutes to answer a routine enquiry and one who takes 3 and a half, after AI drafts the reply and they review it, deliver the same email. On hourly billing the faster one earns 56% less for it. Clients are not wrong to expect some benefit from the speed, and you are not wrong to want your income back, but a stopwatch is a poor place to settle the argument.
A quick sum shows the size of the problem. Say you charge $28 an hour and a client's work took 26 hours a month, so $728. If your AI-assisted process does the same work in 14 hours, hourly billing gives you $392: a drop of $336 a month on one client, with the same results delivered. To stay level you would need $52 an hour for 14 hours, or a rise to $45 and a lower total, and either number invites a conversation about "why does it cost more when it's quicker?". The next sections show how to measure the saving properly, and then four ways to price the same month.
Measure your own saving before you reprice
Don't borrow a percentage from an article. A published figure for how much AI saves is the seller's estimate for the seller's workload, and drafting speed varies hugely by task. Keep a plain time log for ten working days. For every task write the minutes it used to take you, the minutes it takes now including your review, and what the AI did. Include the review time every time; it is the part that doesn't shrink, and leaving it out is how people end up underpricing. Here is an illustrative log for a VA supporting a small business owner.
| Task | Minutes before AI | Minutes now, with review | Time saved |
|---|---|---|---|
| Reply to a routine customer enquiry | 8 | 3.5 | 56% |
| Weekly newsletter draft | 90 | 50 | 44% |
| Minutes and actions from a 45-minute call | 40 | 18 | 55% |
| Five social posts with captions | 75 | 40 | 47% |
| Set of five invoice chaser emails | 25 | 12 | 52% |
| Travel itinerary and confirmations | 60 | 45 | 25% |
| Supplier research summary | 120 | 80 | 33% |
The spread, from 25% to 56%, is the point. Drafting-heavy work such as replies, minutes and chasers shrinks by half or more. Coordination work, where the hard part is checking bookings or gathering facts, shrinks by a quarter or a third. A blanket "AI discount" would overcharge your drafting clients and undercharge your coordination clients, which is why the packages below are built task by task. As a rule of thumb, if a task's saving is under 20% after review, price it as before.
Two habits keep the log honest. Time the whole task from the client's request to the sent draft, not only the typing. And log a bad day as well as a good one: a draft that needed three rounds of corrections belongs in the average.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowOne client's month, priced four ways
Work through one client: the owner of a landscaping company with a six-person crew. Before AI, you spent 26 hours a month on their inbox, quote follow-ups, social posts, invoice chasing and diary, billed at $28 an hour, so $728. Your log now shows 14 hours for the same scope. Here is what each pricing model does with that month.
| Model | Client pays | Effective rate on 14 real hours | Change for you | Main risk |
|---|---|---|---|---|
| Hourly at the old $28 | $392 | $28.00 | -$336 a month | Your income falls with your speed |
| Hourly raised to $45 | $630 | $45.00 | -$98 a month | A 61% rate rise is a hard conversation |
| Package, $650 for a defined scope | $650 | $46.43 | -$78 a month | Scope creep if volumes aren't capped |
| Retainer, 16 hours at $40 | $640 | $45.71 if 14 hours are used, $40 if 16 | -$88 a month | Disputes over unused hours |
The package has the best mix. The client pays $78 less than before, about 11%, which is a visible saving to point to. You earn $46 for each hour you actually work, which is higher than any hourly rate you could comfortably announce. The retainer gives steadier income and gives the client reserved capacity, and it suits a client whose needs change month to month. Hourly billing works for the unpredictable remainder: one-off research, a trip to organise, a project that doesn't repeat. Most VAs end up with a mix: a package for the repeatable core, an hourly rate for extras.
A coordination-heavy client, where AI saves less
A second client shows why one discount rate would be wrong. This one is the owner of a removals firm. Before AI you spent 20 hours a month on booking confirmations and diary changes (8 hours), messages to crews and suppliers (6), customer enquiry replies (4) and review requests (2), billed at $30 an hour, so $600. Your log now reads 6.5, 5, 1.8 and 0.7 hours: 14 in all. Only the replies and review requests shrank much, because crew messages and diary changes depend on other people's answers, not on drafting speed.
The time saved is 30%, against 46% for the landscaping client, so the price should fall by less. A $560 package for a defined scope implies $40 an hour on the 14 real hours and costs the client $40 (about 7%) less than before. Billing hourly at $30 would give you $420, a loss of $180 a month. A client who pays 7% less for faster confirmations has little to complain about, and the smaller reduction is accurate, because you measured it.
Building a package ladder from measured hours
Work backwards from your log. For each package, list the deliverables, multiply by the measured minutes, add the fixed jobs, and see what hourly rate the price implies. If the implied rate is below your floor, change the scope or the price, not the arithmetic. A floor is easy to set: say you want $3,000 a month, carry $300 a month of overheads such as software, equipment and insurance, and bill 65% of a 160-hour month, which is 104 hours. That gives $3,300 over 104 hours, a floor of about $32 an hour, or about $32.30 once $60 a month of AI subscriptions is included. A package priced below that is working for the client at your own expense.
| Package | Price a month | What is included | Hours behind it | Implied rate |
|---|---|---|---|---|
| Inbox Starter | $350 | Up to 100 drafted replies, calendar upkeep, a five-line daily summary | 8.8 (100 replies at 3.5 min, 2 hours of diary, 1 hour of summaries) | about $39.60 |
| Owner Desk | $650 | Starter plus 12 quote follow-ups, 8 social posts, a weekly invoice chase | 13.7 (adds 2, 1.6 and 1.3 hours) | about $47.40 |
| Owner Desk Plus | $1,100 | Owner Desk plus a monthly newsletter, lead qualification emails for up to 20 enquiries, a weekly report, expenses prep | 24.4 | about $45.00 |
Every row clears the floor, and the step from Starter to Owner Desk is where a client with a growing business will naturally land. Note what the table doesn't do: it doesn't promise a number of hours. A client buys a defined result, such as replies drafted and sent within a set time, and the hours are your concern. That separation is what lets you keep the benefit of working faster. The companion tutorial on how a VA business can package AI-assisted services covers naming and positioning these tiers.
Boundaries that keep a package profitable
A package without edges is an open-ended hourly contract at a worse rate. Write four limits into every one.
- A volume cap. "Up to 100 drafted replies a month" with a stated price for each extra block of 25. Without it, the client who sends 400 emails pays the same as the one who sends 60.
- A turnaround. "Drafts the same working day if received by noon." It gives the client a reason to value the package and gives you a measurable promise.
- A review limit. One round of changes per draft. A second round is a conversation, and a third is a signal that the brief was wrong.
- Named exclusions. Phone calls, bookkeeping, anything that touches the client's money or legal documents, unless you list it. It protects you and makes upgrades easy to propose.
Here is how skipping the first limit shows up. A VA sells a $650 package without a cap. In month three the client's business gets busier, and 340 emails arrive instead of 100. At 3.5 minutes each that is nearly 20 hours of replies alone, so the effective rate falls from $47 to under $25. The fix is a mid-month message offering the next block, written before it is needed: "Your inbox is running at 3.4 times the package volume. Two options: add a 100-reply block for $200, or we move you to Owner Desk Plus at $1,100." Clients accept a clear offer far more readily than a surprise invoice.
Retainer terms that stop arguments
A retainer sells reserved capacity, so the terms need to say what is reserved and what happens to unused time. The usual dispute is about hours that go unused one month and are wanted the next. Decide the rule in advance and put it in the agreement. Here is a short set of terms for the 16-hour retainer at $40, which is $640 a month.
Retainer: 16 hours a month at $40, billed on the 1st for that month
Rollover: up to 4 unused hours carry into the next month only, then expire
Ceiling: no more than 24 hours available in any single month
Response: requests acknowledged within 4 working hours, drafts within 1 working day
Extra hours: $45 an hour, agreed by message before work starts
Review: a 20-minute call on the last Friday to check hours and priorities
Notice: 30 days by either sideTest the rollover rule with numbers. If the client uses 10 hours in month one, 4 roll over, so month two has 20 hours available. If they use all 20, month three starts at 16. The ceiling of 24 stops the client from banking time for a busy season you can't staff. For a method to set the underlying hourly figure, the freelance day rate calculator tutorial works through it when AI speeds up delivery.
The other half of the sum: what the freed hours can earn
Everything above lowers the price per client a little, and that can feel like a loss. It stops feeling like one when you count capacity. Say you bill 104 hours a month. At 26 hours per client you could serve four clients and earn 4 x $728 = $2,912. At 14 hours per client the same 104 hours cover seven, and seven clients at $650 earn $4,550, which is 56% more. Allow for the time it takes to find and onboard clients and to switch between them, and plan on five or six rather than seven: six clients at $650 is $3,900, still 34% more than before.
Capacity is why pure hourly billing is the worse choice: it hands the whole gain to the client and leaves you with spare hours and no income from them. Whether you can fill those hours depends on your marketing more than your pricing, and the tutorial on how virtual assistants use AI to manage more clients covers the workload side of taking on five or six at once.
Tool costs: fold them in or pass them through
Sort every tool into one of three buckets and treat each bucket differently.
- Tools that serve all your clients. Your own assistant subscription is the obvious one: ChatGPT Plus is $20 a month and Claude Pro is $20 monthly or $17 on annual billing. These are overhead. Put them in your base price and never itemise them.
- Tools bought for one client. A client who wants you inside their own workspace buys the seat. ChatGPT Business is $25 per user monthly or $20 on annual billing, Claude Team is the same, and both need two seats, so a one-person business is paying for two. Write it into the proposal so it isn't a surprise; business plans of this kind don't train on your content by default, which is often the reason a client asks for one.
- Costs that scale with volume. Transcription minutes, stock images, paid API usage. Include an allowance in the package and bill anything above it at cost plus a small handling charge, with the rule stated in writing.
The first bucket is smaller than people fear. Suppose your stack is one assistant at $20 and a transcription tool at a placeholder $15 a month. Across five clients that is $7 each, about 1% of a $650 package. Itemising it would cost more in awkward conversations than it earns. The tutorial on the AI tools VAs use across five clients shows a stack with realistic costs, and the confidentiality checklist explains why client data belongs in a business plan or the client's own account, not a free personal one.
What published rate guides say, and why to distrust them
Everyone asks what the market charges, and the honest answer is a spread. goLance's rate guide, updated October 2026, lists $10 to $20 an hour for junior assistants, $20 to $35 for mid-level, $35 to $55 for senior and $55 to $90 for experts, and doesn't explain how it got them; the same page shows its own marketplace's typical range at $5 to $10, which tells you how far platform averages sit from the guide's bands. A managed-assistant agency's July 2026 guide gives $10 to $14 for entry level, $14 to $22 mid-level, $22 to $35 senior and $25 to $60 for specialists, and prices a full-time managed assistant at $1,600 to $2,500 a month and a part-time one, about 20 hours a week, at $800 to $1,200. It also shows lower hourly figures for assistants in lower-cost regions and $25 to $75 for the highest-cost markets, which explains why a rate quoted from another market is no benchmark for yours.
Both guides are published by businesses that sell assistants or compare them, neither gives a sample size, and they disagree on what "senior" means. Use them as outer limits only. A better test costs an afternoon: collect three public price lists or quotes for comparable work from providers your clients could realistically hire, work out the effective hourly or per-deliverable price of each, and place yourself against them on speed, specialism and reliability rather than on the lowest number. If you serve clients in the highest-cost markets and your rate sits towards the top of the published bands for specialist work, that is normal; if you are a generalist priced there, expect pushback.
Telling an existing client about the change
Don't announce "AI discount" or "AI surcharge". Both frame the change around your tools, and the client's interest is the outcome. Frame it around what they get, give notice, and show the sum. Here is wording you can adapt.
Hi [first name],
From 1 January I'm moving your support from hourly billing to a monthly package, Owner Desk, at $650. It covers the same scope you use now: inbox drafts and replies, quote follow-ups, social posts, the weekly invoice chase and your diary.
Over the last three months you averaged 26 hours, or $728 a month. The package is $78 less, and you'll get same-day drafts if they reach me by noon, plus a one-page summary at month end. You won't need to approve hours any more.
If your volumes grow, I'll tell you before they pass the cap, and we can add a block or move up a tier. If you'd rather stay on hourly billing, my rate from January is $45.
Thanks, [your name]How you describe your tools to a client is a separate question, covered in whether freelancers should tell clients they use AI. Two parts of this message do the work. The $78 reduction gives the client a visible gain, and the stated alternative at $45 shows that the package is the better deal rather than a trick. Send it at least 30 days ahead, and keep the old terms for anything already agreed.
Pricing like this is a commitment, so review it after 90 days: compare the hours you actually worked against the hours behind each package, and move any package whose implied rate has fallen below your floor.
Questions virtual assistants ask when they reprice
Should I tell clients I use AI to draft?
Yes if asked, and better to say it first in your terms: describe drafts as AI-assisted and human-reviewed, state which client data goes into which tool, and let clients opt out for sensitive material. Pricing by outcome keeps the conversation on quality and turnaround rather than on minutes.
How big a discount should a retainer carry?
A small one at most. A retainer gives the client reserved capacity and you steady income, so a few percent below your package-equivalent hourly rate is a reasonable starting point; that is my rule of thumb, not a market rule. Add a rollover cap so unused hours don't pile up into a dispute.
What if a client asks me to pass the AI savings on?
Show them what they receive: faster turnaround, a quality check on every draft and a price that fell for the same scope. A client buys results and reliability, not your typing speed. If you have to concede something, offer extra scope in the first month rather than a permanently lower rate.
Do I charge extra for my AI tools?
Fold tools that serve every client into your base price, because they are overhead. Charge separately only for tools bought for one client, such as a seat in their own workspace, at cost and with written notice. Itemising a $20 subscription across five clients invites haggling over pennies.
Further reads
- AI Assistant vs Virtual Assistant: Which Should You Pay For? — How clients weigh your fee against an AI assistant.
- Consulting Pricing Models After AI: Fixed Fee, Retainer or Value? — Fixed fee, retainer or value pricing after AI.
- Marketing Agency Pricing After AI: Retainers, Packages and Deliverables — Retainers and deliverables priced when AI speeds production.
- Life Coaching Rates During Automation: How to Reprice Packages — Repricing packages when automation takes over some work.
- ChatGPT Prompts for Virtual Assistants: Inbox, Scheduling, Reports — Prompts for the inbox and calendar work you are pricing.
- Event Planning AI Agent or Virtual Assistant: Which Fits a Small Planner? — Decide task by task whether supplier chasing and RSVP follow-ups suit an AI agent or a virtual assistant, with monthly costs and a rule by event volume.
- Chat Support Virtual Assistant Software: Which Fits a VA Business? — Which shared inbox or live chat tool suits a VA serving several client shops, what each really costs at eight clients, and whether AI drafts pay.
- Grant Writer Rates: What to Charge When AI Drafts Faster — Freelance grant writer rates by hour and by project, plus a worked quote showing how AI drafting should change your pricing, and why percentage fees are out.
- Subtitling Rates per Minute: What to Charge When AI Drafts First — Price AI-drafted subtitles by review time: published vendor bands, a per-minute formula, a three-clip timing test and quotes for QC, proofreading and rush.
- AI Email Management for Virtual Assistants: Rules, Drafts, Handover — Triage labels, draft-only replies, escalation rules, delegated-access safety and a daily digest, with a client approval handover a VA can run from week one.
- Amazon Seller Virtual Assistant or AI: Which Tasks Go to Whom — A task-by-task split for Amazon sellers: what a virtual assistant should do, what AI should draft, what stays with you, and what each costs.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: goLance freelance virtual assistant rate guide (updated October 2026) and Stealth Agents virtual assistant pricing guide (July 2026), both read in November 2026 and used only as published ranges; ChatGPT and Claude plan prices from the vendors' pricing pages.