Price the outcome and the scope, not the hours. A fixed fee suits a project with a clear deliverable, a retainer suits ongoing access to your judgement, and value-based pricing suits only work where the client can measure the result and you can show your share of it. Hourly billing is the model AI breaks first.
The deciding factor is how predictable your hours are, not how clever your AI tools are. AI shortens drafting, desk research and analysis. It does not shorten listening, deciding or presenting. A diagnostic that took 40 hours might now take 31 while the client values it exactly as before, so your pricing model decides who keeps the nine hours.
Where the hours went on a 40-hour diagnostic
Start with a real project type rather than a feeling. Take a supplier-cost diagnostic that a solo consultant used to sell at $150 an hour: eight stakeholder interviews, a desk review, an analysis model, a written report, a short deck and two client sessions. The table splits the old 40 hours and the hours the same deliverable might take once AI produces the first drafts. The figures are illustrative; the shape is the point, because AI compresses the writing-heavy rows and leaves the people-heavy rows alone.
| Task | Hours before AI | Hours with AI | What changed |
|---|---|---|---|
| Stakeholder interviews and calls | 8 | 8 | AI drafts the question list; the conversations take as long as before |
| Desk research and benchmarks | 6 | 3 | Summaries of public material in minutes, then source checks |
| Analysis and modelling | 8 | 6 | Faster cleaning and first-pass formulas; the judgement is unchanged |
| Report drafting | 10 | 5 | First drafts from your notes; your editing stays |
| Slides and visuals | 4 | 2 | Structure and first slides generated, then reworked |
| Client sessions and presenting | 4 | 4 | No change |
| Checking AI output and sources | 0 | 3 | New work: facts, numbers, tone |
| Total | 40 | 31 | Nine hours saved (22.5%) |
Two details matter. The first is the new checking row: skip it and sooner or later you send a client a confident error, so the saving is not real until the checking is counted. The second is the size of the saving, which is 22.5%, not 70%. Claims about half-day deliverables usually describe the drafting step only. To get your own split, time three projects by task for a month; how much time AI can save a solo consultant each week shows a simple way to log it.
What each model does with the nine saved hours
Run the same project through each model and the differences show up at once. The effective hourly rate is simply the fee divided by the 31 hours you now spend.
| Model | Fee | Your hours | Effective hourly rate | Who keeps the nine hours |
|---|---|---|---|---|
| Hourly at $150 | $4,650 | 31 | $150.00 | The client, whose bill fell by $1,350 |
| Fixed fee priced on the old 40 hours | $6,000 | 31 | $193.55 | You |
| Same fee after a 10% client discount | $5,400 | 31 | $174.19 | Shared |
| Same fee less $1,000 (about 17%) | $5,000 | 31 | $161.29 | Mostly the client |
Under hourly billing your revenue on this project falls by $1,350 for an identical deliverable, and the better you get with AI the more you lose. A fixed fee keeps the saving, but only while the client accepts the price. Once buyers know that AI exists, expect the question "why hasn't this got cheaper?" A fixed fee holds when you can answer in terms of what the deliverable does for them. It cracks when the only answer is how hard you worked.
Retainers have the same problem in a quieter form, and the fix is different. That comes after the quoting method.
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 nowSet a floor from your hours and a price from the buyer's alternative
Two numbers frame every fixed-fee quote. The floor is the lowest fee you would accept, built from your honest AI-assisted hours. The ceiling is what the buyer would spend if they did not hire you. Your price sits between them, nearer the ceiling when the work is hard to replace and nearer the floor when you are bidding against others.
floor = AI-assisted hours x (1 + contingency) x target hourly rate
= 31 x 1.2 x $150
= $5,580
contingency 20%: one extra feedback round, a late data file,
a stakeholder who joins the process halfway
On the diagnostic the floor is $5,580. Suppose the client's realistic alternative is a larger firm's quote of about $9,500. That number is an assumption for this illustration, and in practice you would replace it with whatever you hear in discovery. A fee of $7,000 then sits 25% above your floor and 26% below the alternative. The effective rate over your 31 hours works out at $225.81, which only looks high if you divide by hours, and the client is not dividing by hours.
Do not open with the floor. In discovery ask what the client does today when this needs doing (internal time, an agency, nothing at all) and what a wrong answer would cost them. Those answers set the ceiling, and they are the best raw material for the proposal itself. For turning discovery notes into a first draft quickly, see how consultants use AI to write proposals in under an hour.
Scope boundaries that hold when delivery gets faster
AI makes extras look free, and clients notice. "Could you also add a competitor view?" sounds like a five-minute request once they have heard that a draft takes minutes. A fixed fee survives only if the scope says what is in, how many rounds of feedback are included and what each extra costs. Price extras by the value of the item, not by the minutes it takes you.
SCOPE: SUPPLIER-COST DIAGNOSTIC (fixed fee $7,000)
Included
- Up to 8 stakeholder interviews, up to 45 minutes each
- One written report, up to 20 pages, with a ranked list of cost drivers
- One 90-minute readout to the leadership team
- Two rounds of consolidated feedback, each returned within
10 working days of the draft
Assumptions
- Client supplies spend data within 5 working days of kick-off
- One named decision-maker approves the report
Extras (fixed prices, agreed in writing before work starts)
- Each interview beyond 8: $450
- Each extra report section (up to 4 pages): $600
- Additional readout session: $500
- New data source after week 2: re-quoted
Not included
- Implementation support, training, ongoing monitoring
(these can be offered as a separate retainer)
The extras are deliberately not hourly. An interview beyond the eighth costs you roughly three hours with preparation, notes and follow-up, and AI trims the write-up, but the client is buying another view of their business. Review the list after every project and add the extras clients really asked for. If the same extra comes up three times, move it into the core scope and the core price.
The assumptions block earns its place too. Most fixed-fee overruns come from late data and unlimited revisions, not from the AI step, and writing them into the scope turns a delay into a conversation about the timeline rather than a silent loss. The HR consultant proposal template with scope, fees and exclusions shows the same structure for people work.
Retainers: sell access and outputs, not an hours bank
An hours-bank retainer says "20 hours a month for $3,000". When AI cuts the work behind your usual monthly outputs to 15.5 hours, one of three things happens. You quietly under-use the bank and the client senses it, or you pad work to look busy, or you tell the client and discount. All three are avoidable by changing what the retainer sells.
| Retainer type | What the client buys | What AI does to it | What to write in the agreement |
|---|---|---|---|
| Hours bank | A number of hours each month | Exposes unused hours and invites a discount | Avoid for new clients; if you keep it, add a minimum term and a cap on rolled-over hours |
| Access plus outputs | Calls, a response-time promise and named monthly outputs | Lifts your margin; the client sees outputs, not timesheets | The outputs, the number of calls, the response time (for example one working day) and a fair-use limit |
| Outcome-linked | A base fee plus a bonus tied to a metric | Neutral on hours; depends entirely on measurement | The metric, the baseline, the measurement date and a cap |
Here is what a rewrite looks like for a retainer that already exists. The price stays the same; the promise changes.
BEFORE
Advisory retainer: 20 hours a month, $3,000, unused hours lapse.
AFTER
Monthly advisory pack, $3,000, three-month minimum:
- one written KPI review (2 pages)
- two 60-minute calls
- written answers to up to 10 questions, each within one working day
- one deep-dive on a topic of the client's choice every quarter
If delivering the pack takes you 15.5 hours, your effective rate is $193.55 and the client has no timesheet to hold against it. One condition makes this honest: the contract must no longer say hours. If an existing agreement is hours-based, keep reporting hours truthfully and change the structure at renewal, explaining that you are moving to a defined monthly pack.
Spend part of the saved time on something the client can see. A quarterly benchmark note that takes four hours and gives the client a decision to make is a better use of 4.5 spare hours than a quiet discount. Over a year it also gives you something concrete to point to when the renewal conversation starts.
Value-based pricing: five conditions, and where it fails
Value pricing makes the fee follow the economic result: a share of savings, extra revenue or risk avoided. It is the most attractive model for the consultant and the hardest to run well. Score each proposal against five conditions, one point each.
- The client can state the outcome in money, starting from a baseline they trust.
- The result shows up within a window you can wait for, ideally twelve months or less.
- Your contribution can be separated from the other changes the client is making.
- The person signing controls the budget and cares about the money, not the hours.
- You can live with the downside if the number comes out smaller than hoped.
Four or five points: value pricing is realistic. Two or three: use a base fee plus a capped success fee. Zero or one: use a fixed fee and stop there.
A proposal that passes: a distribution business loses about $120,000 a year to pick-and-pack errors, a figure taken from the client's own audit. You expect a 30% reduction, worth $36,000 a year. Pricing at 15% to 20% of year-one value gives $5,400 to $7,200, and your floor is $5,580, so the range works. All five conditions are met, because the client has a baseline, a time window, a separable change, a finance-minded sponsor and a cushion if the saving disappoints.
A proposal that fails: a positioning project for a services firm. There is no baseline for "brand strength", sales will move because of three other changes, and the owner wants the work finished in six weeks. At best the proposal scores two. Value pricing here turns into an argument about whose efforts produced next quarter's revenue, and you lose that argument because you cannot prove your share.
The middle route is a base fee that covers your floor plus a success fee with a cap. These terms are an illustration: base $5,600, plus 10% of verified savings above $20,000 in the first twelve months, capped at $3,000.
| Verified first-year savings | Success fee | Total fee | Effective rate over 31 hours |
|---|---|---|---|
| $10,000 | $0 | $5,600 | $180.65 |
| $20,000 | $0 | $5,600 | $180.65 |
| $36,000 | $1,600 | $7,200 | $232.26 |
| $60,000 | $3,000 (cap) | $8,600 | $277.42 |
Write the measurement rule into the agreement: who measures, from which records and on what date. "Verified" without that definition invites a dispute. The other ways value pricing fails are predictable enough to list at the proposal stage: the client's own analyst redefines the metric, other projects muddy attribution, the result arrives after your window, or a sceptical buyer says "it only took you twenty minutes with ChatGPT". The answer to the last one is condition one: you are being paid for the $36,000, not for the twenty minutes.
Telling clients you use AI without apologising for it
Say it once, in the proposal and again in the engagement letter, and be specific. Name the tasks (drafting, summarising public sources, formatting, data cleaning). State what stays human: the interviews, the conclusions, the recommendations and the final review. Say where client material goes: only into tools covered by business terms, never into free consumer tools, and not at all if the client has said no. The client-data checklist for consultants using AI covers the data side in more detail.
How we work. I use AI tools to speed up drafting, summarising
public sources and tidying data. The interviews, analysis choices,
conclusions and final review are done by me personally, and I am
accountable for every page of the report. Your confidential
material is handled only in tools covered by business terms that
exclude it from model training, and I will follow any written
restriction you place on AI use. The fee is fixed for the
deliverables listed above and does not vary with the hours spent.
Only write what is true of your own setup. If you cannot honestly say "business terms that exclude it from model training", change the setup first and the sentence second.
When a client asks for a discount because of AI, a reply that works is short and keeps the conversation on outcomes:
[first name], fair question. AI does speed up my drafting, and that is part of why this lands in four weeks rather than seven. The fee covers what you receive: a ranked list of cost drivers you can act on, and my accountability for it being right. If the budget is tight I would rather reduce the scope than the standard, for instance one fewer readout, and I can quote that version today.
Offering less scope instead of a lower fee protects the price floor and tells the client you are not discounting your judgement. If they still want the full scope for less, you have found where the floor really is.
Back-test three past projects before you change a price list
Take your last three projects and fill in the actual hours, then estimate the AI-assisted hours honestly by task, including checking time. The sums below are illustrative, using the same $150 rate.
| Project | Hours before AI | Hours with AI | Hourly revenue before | Hourly revenue with AI | Effective rate if the old fee holds |
|---|---|---|---|---|---|
| Diagnostic | 40 | 31 | $6,000 | $4,650 | $193.55 |
| Facilitated sessions (a series of four) | 24 | 19 | $3,600 | $2,850 | $189.47 |
| Advisory retainer (one month) | 12 | 9 | $1,800 | $1,350 | $200.00 |
| Total | 76 | 59 | $11,400 | $8,850 | $193.22 |
On these three projects, hourly billing would cost $2,550 of revenue, or 22.4%, for identical work. A workable rule: if hourly revenue on a project type would fall by more than 15% once AI-assisted hours replace the old ones, move that type to a fixed fee or an outputs retainer this quarter. If it falls by less, leave it and look again in six months, when your estimates have real project data behind them.
The same back-test tells you which model to start with. Match each project type to the situation, not to your preference.
| Situation | Model to start with | Check before you send it |
|---|---|---|
| Describable deliverable, scope fits on one page | Fixed fee | Scope list, two feedback rounds, priced extras |
| Open-ended advice, unpredictable requests | Access plus outputs retainer | Response-time promise, fair-use limit, three-month minimum |
| Result measurable in money within a year | Base fee plus capped success fee | Baseline, measurement rule, cap |
| Competing against firms that quote hourly | Fixed fee with itemised scope | Show content and outcomes, not hours |
| Client insists on hourly | Day rate or hourly with a monthly floor | Floor at roughly 80% of expected hours |
Consultants whose work AI barely touches
Not every practice feels this pressure equally. A facilitation consultant who spends 70% of a project in a room with the client sees a smaller change than the diagnostic above: perhaps preparation falls from ten hours to seven and everything else stays. For that kind of work a day rate with a clear minimum still functions, because the client is buying your presence and judgement in the room, which no tool replaces. Run the back-test anyway. If your saving comes out under 10%, you probably do not need to restructure anything this year, and the money is better spent on the proposals that win the work.
At the other end sits a research-heavy consultant who writes long reports. The drafting rows carry most of the hours, the saving is larger, and the fixed fee or outputs retainer matters sooner. The question to ask is where your hours sit, not what your job title is.
A review you can finish in one afternoon
- Time three projects by task for a month, or reconstruct them from calendars and drafts.
- Estimate the AI-assisted hours task by task and add a checking row. Be stingy with the savings.
- Work out the floor with 20% contingency, and ask in your next discovery call what the client's alternative costs.
- Write the scope boundaries and the extras list for your most common project.
- Rewrite one existing retainer as access plus outputs, ready for its next renewal.
- Draft the AI-use paragraph for your proposals and check that every sentence is true of your setup.
If you would rather work through the numbers with someone else looking at them, that is what a call about your own project mix covers. Either way, change one pricing document at a time and compare the next three projects with the back-test.
Questions consultants ask when AI shortens the work
Should I charge less now that I use AI?
Not automatically. Price on what the deliverable is worth to the buyer and what their alternatives cost, not on your hours. If a client pushes for a discount, offer a smaller scope before a lower standard, for example one fewer readout. If you are bidding against firms that quote hourly, show an itemised scope so the comparison is about content, not effort.
Is hourly billing finished?
No, but it punishes speed. Hourly still suits open-ended advice where nobody can write the scope down. If you stay hourly, add a monthly minimum or a day rate with a floor, so that the faster you work the less you are penalised. For anything with a describable deliverable, a fixed fee or an outputs retainer will usually pay better.
What if my AI-assisted estimate turns out too low?
Build in about 20% contingency, log actual hours against the estimate on your next five projects, and adjust the contingency from the real overrun. Cap feedback rounds, name the data you need and by when, and price extras separately. A fixed fee goes wrong mostly through unlimited revisions and late client data, not through the AI step.
Do I have to tell clients I use AI?
That depends on your contract, any client rules on AI and the standards of your field, so check them and ask a qualified adviser if unsure. As a working habit, say it once in the proposal and engagement letter, describe the tasks, and confirm that a person reviews everything and stays accountable. Saying it up front is an easier conversation than explaining it after a client finds out another way.
Further reads
- Marketing Agency Pricing After AI: Retainers, Packages and Deliverables — The same retainer logic for agencies selling packages and deliverables.
- How to Price Accounting Services: Quote Fixed Fees With AI Support — Fixed-fee quoting in another professional service, with a worked scoping method.
- Virtual Assistant Pricing When AI Does the Drafting: Rates and Packages — How rates and packages change when drafting gets faster.
- ChatGPT for Freelancers: Turn One-Off Clients Into Monthly Retainers — Turning a finished project into a monthly arrangement.
- Is an AI Consulting Retainer Worth It After Your First Project? — How a buyer judges a retainer, useful when you write your own.
- AI Consulting Report Writing: From Findings to a Client-Ready Draft — Where the drafting hours actually go in a report.
- How to Price Interior Design Services: Compare Fee Models With AI — One three-room project priced hourly, fixed-fee and cost-plus, with margin checks, stress tests and client wording for each model.
- Best AI Tools for Management Consultants: Seven From Research to Decks — Seven AI tools for independent consultants from research to decks, with a confidentiality tier rule, costs and a worked two-week diagnostic.
- Freelance Day Rate Calculator: Set Your Rate When AI Speeds Up Delivery — A step-by-step freelance day rate calculation in USD, with billable days, overheads, AI tool costs and four ways to price when AI speeds up delivery.
- 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.
- Life Coaching Rates During Automation: How to Reprice Packages — How to reprice life coaching packages once AI handles prep and admin: hours-based versus milestone-based structures, with a worked example.
- Virtual CFO Pricing: Build a Monthly Package With AI-Prepared Reports — How a firm can price a monthly virtual CFO package: three tiers, the reports in each, review time, tool costs and add-on rules.
- Consultant SEO for ChatGPT Search: Get Your Practice Recommended — Make a consulting practice easy for AI search to identify and cite: entity pages, case studies, reviews, schema and a fixed prompt test you rerun monthly.
- Consulting Engagement Letter: Scope, Fees and AI-Use Wording — A clause-by-clause engagement letter for independent consultants: scope, fees, change control, confidentiality and plain AI-use wording you can adapt.
- Consulting Utilisation Rate: Track It Weekly and Price Projects With AI — Billable versus available hours, bench time, a weekly AI roll-up from timesheets and a worked project price checked against your rate card.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: no vendor pricing was needed; every figure in this tutorial is illustrative arithmetic you can replace with your own hours and rates.