TL;DR. Most freelancers and consultants do not avoid raising rates because they lack a method. They avoid it because it feels like a risk with no floor. The evidence says the floor is much higher than it feels. First, standing still is itself a price change: if your rate was last set in August 2021, US consumer prices have since risen 22.4%, so an unchanged rate is an 18.3% real pay cut, and salaried peers in professional, scientific and technical services have received 20.8% more in wages over almost the same period (our calculation from BLS data). Second, clients are not opposed to price increases as such. In the fairness surveys run by Daniel Kahneman, Jack Knetsch and Richard Thaler, 79% of respondents found it acceptable to pass on a cost increase and 75% accepted a cost-driven rent increase at lease renewal, while 91% judged it unfair to raise the price on someone because they had become unlikely to leave. Third, the arithmetic is forgiving: after a 10% raise you can lose 9.1% of your billings and earn exactly the same revenue in 9.1% fewer hours. The playbook below turns those three facts into a sequence: measure your gap, choose the size, time it to a renewal, give notice, say it in the language people accept, and decide in advance what you will do about each client.
Standing still is a price cut
A rate is a promise about the future made in the money of the past. Each year it is left alone, it buys less. The table below measures how much less, using the official US series: the Consumer Price Index for All Urban Consumers (CPI-U) for what your rate buys, and the Employment Cost Index for what salaried people doing comparable work have been paid.
Table 1. What an unchanged rate has lost by August 2026, by the year it was last set (CEOtudent calculation from BLS data)
| Rate last set in | Raise needed to restore purchasing power (CPI-U, August to August 2026) | Real value lost by the unchanged rate | Wage growth, professional, scientific and technical services (ECI, Q2 to Q2 2026) | Wage growth, all private industry (ECI, Q2 to Q2 2026) |
|---|---|---|---|---|
| 2019 | +30.6% | 23.4% | +26.8% | +31.0% |
| 2020 | +28.9% | 22.4% | +23.5% | +27.3% |
| 2021 | +22.4% | 18.3% | +20.8% | +22.9% |
| 2022 | +13.1% | 11.6% | +14.9% | +16.3% |
| 2023 | +9.1% | 8.3% | +10.3% | +11.2% |
| 2024 | +6.4% | 6.0% | +6.0% | +6.8% |
| 2025 | +3.4% | 3.3% | +2.6% | +3.1% |
Method: CPI-U, US city average, all items, not seasonally adjusted, August of each year against August 2026 (index 334.980). ECI wages and salaries, current dollar index, second quarter of each year against the second quarter of 2026, the latest release. “Raise needed” is the ratio of the 2026 index to the base-year index minus one; “real value lost” is one minus the inverse. The two differ because a loss and the raise that repairs it are measured from different bases: losing 18.3% of value takes a 22.4% raise to undo.
Three things stand out.
The raise you need is larger than the loss you feel. A rate set in 2021 has lost 18.3% of its value, but getting back to even takes 22.4%. People anchor on the smaller number and under-correct.
Your salaried peers were repriced for you. Employers in professional, scientific and technical services raised wages 20.8% between mid-2021 and mid-2026. That is not a market signal you have to infer. It is the rate at which the people your clients could hire instead have become more expensive.
The last twelve months alone justify something. From August 2025 to August 2026, CPI-U rose 3.4% and the services index 3.1%. Even a rate reset last year is already behind.
These are US figures. The logic transfers everywhere; the numbers do not. If you bill in another currency, run the same calculation with your national statistics office’s consumer price index and wage series.
What people actually accept: the fairness evidence
The fear behind most rate anxiety is that clients will see a raise as greedy. There is unusually direct evidence on when that happens. In telephone surveys of randomly selected residents of Toronto and Vancouver, collected between May 1984 and July 1985 and published in the American Economic Review in 1986, Kahneman, Knetsch and Thaler asked people to judge specific pricing and wage decisions. Contrasting versions of a question were never put to the same respondent.
Table 2. How people judged price and wage changes (Kahneman, Knetsch and Thaler, 1986)
| Scenario (paraphrased) | Respondents | Acceptable | Unfair | What it means for a rate change |
|---|---|---|---|---|
| Grocer passes a 30-cent wholesale cost increase on to customers | 101 | 79% | 21% | Cost-driven increases are widely accepted |
| Landlord raises rent at lease renewal to cover increased costs, even for a tenant on a fixed income | 151 | 75% | 25% | Renewal plus a cost reason is the accepted moment |
| With no inflation, an employer cuts wages 7% | 125 | 38% | 62% | A visible cut is resented |
| With 12% inflation, an employer raises salaries only 5% | 129 | 78% | 22% | Almost the same real cut is accepted when it looks like an increase: people judge in nominal terms |
| Dealer ends a $200 discount and sells at list price | 123 | 58% | 42% | Removing a discount meets less resistance than raising a price |
| Dealer charges $200 above list during a shortage | 130 | 29% | 71% | Exploiting a demand spike is resented |
| Store raises snow shovels from $15 to $20 the morning after a snowstorm | 107 | 18% | 82% | Price rises timed to the customer’s need are punished |
| Grocery chain charges 5% more where it has no competitor | 101 | 24% | 76% | Charging more because the client has no alternative is resented |
| Landlord adds $40 a month on learning the tenant is unlikely to move | 157 | 9% | 91% | Pricing on lock-in is the most punished move of all |
The authors summarise the pattern as a principle of dual entitlement: the customer is entitled to the reference price they have been paying, and the firm is entitled to its reference profit. A price rise that protects the seller’s profit against rising costs is seen as fair. A price rise that exploits the customer’s need or lack of alternatives is not.
For a solo business this is almost a script. The raise that survives is the one that is framed as keeping your business whole (costs, time, the market rate for the work) and delivered at a natural reset point. The raise that damages relationships is the one that arrives at the moment the client depends on you most: mid-project, right after a deadline slipped, or just after they have become reliant on your knowledge of their systems. That last case matters, because long-term clients usually are locked in, and 91% of respondents called it unfair to price on that.
Two caveats. These are judgments by the public about small businesses, not by procurement teams about suppliers, and they are four decades old. What they measure is durable, though: the difference between a cost reason and an exploitation reason, and between a change at renewal and a change at the moment of need.
The arithmetic: how many clients can you afford to lose?
The second fear is volume. If a raise drives clients away, have you gained anything? The answer depends on one number, and it is surprisingly generous.
If you raise your rate by a fraction p and some clients leave, your revenue stays the same as long as the share of billings you lose is no more than p divided by (1 + p). And because the remaining work is billed at the higher rate, you earn that same revenue in fewer hours.
Table 3. Break-even billings loss after a rate increase (CEOtudent editorial framework: arithmetic, not a forecast)
| Rate increase | Share of billings you can lose and keep the same revenue | Hours freed at that break-even point | Revenue if you lose nobody |
|---|---|---|---|
| 5% | 4.8% | 4.8% | +5% |
| 8% | 7.4% | 7.4% | +8% |
| 10% | 9.1% | 9.1% | +10% |
| 15% | 13.0% | 13.0% | +15% |
| 20% | 16.7% | 16.7% | +20% |
| 25% | 20.0% | 20.0% | +25% |
| 30% | 23.1% | 23.1% | +30% |
Formula: break-even loss share = p / (1 + p). At that point revenue is unchanged and the billable hours required fall by the same share. The table ignores costs that scale with clients (onboarding, admin, tools), which make the true break-even slightly more generous, and it ignores the time it takes to replace a lost client, which makes it less so.
Read it against Table 1. A freelancer whose rate dates from 2021 who raises 20% could lose one billing dollar in six and still earn exactly what they earn today, for fewer hours. Which clients leave is not random, either. The ones who leave over a moderate raise are by definition the most price-sensitive, and if you segment first (step 3 below) you decide in advance that the lowest-value work is the work you are willing to lose. There is no published dataset on freelancer churn after repricing, so treat this as a planning rule, not a finding.
This is also why the break-even framing beats “will anyone leave?”. Somebody usually will. The useful question is whether the loss exceeds the line in the table, and for single-digit and low double-digit raises that rarely happens with a client base that values the work.
The repricing playbook, step by step
Table 4. The CEOtudent repricing playbook (editorial framework)
| Step | What you do | Evidence it rests on | Typical timing |
|---|---|---|---|
| 1. Measure the gap | Find the year your rate was last set and read off the gap in Table 1, or compute it from your own country’s CPI | CPI-U and ECI series | Once a year, same month |
| 2. Choose the size | Take the inflation gap as the floor; add a market or value component only if you can name it | Break-even table; ECI for peer wages | Before any client conversation |
| 3. Segment the book | Sort clients into grow, keep, and reprice-or-release | Break-even logic: the lowest-margin clients are where loss is cheapest | One working session |
| 4. Pick the moment | Tie the change to a renewal, a new statement of work, a new year, or a scope change, never to a crisis | 75% accepted a cost-based raise at renewal; 82% rejected a raise timed to need | 30 to 60 days before the reset point |
| 5. Give notice | Send written notice with a clear effective date | Dual entitlement: the old price is the client’s reference; notice lets them adjust | 30 days minimum, 60 for retainers |
| 6. State the reason | Name costs, the market rate for the work, and what the client gets; never scarcity or their dependence | 79% accepted cost pass-through; 91% rejected pricing on lock-in | In the notice itself |
| 7. Offer one bridge, not a negotiation | A limited grandfather period, a discount you phase out, or a smaller scope at the old price | Ending a discount met 58% acceptance against 29% for a rise above list | Built into the notice |
| 8. Price every new client at the new rate from day one | Stop creating new reference prices at the old level | New transactions take prevailing prices as their reference | Immediately |
| 9. Review in 90 days | Compare billings lost with the break-even line; record who left and why | Break-even table | One quarter after the effective date |
Two steps deserve more detail.
Step 3, segmenting the book. Not every client needs the same treatment. “Grow” clients value the work and are underpriced; they get the standard increase with warmth. “Keep” clients are fine at current value; they get the standard increase. “Reprice or release” clients are those whose effective hourly rate, once you count the calls, revisions and delays, is well below your headline rate. They get the full increase and a clear, polite path out. If a small number of them leave, Table 3 says you are likely to come out ahead.
Step 7, the bridge. The fairness data show that ending a discount is resisted much less than raising a price, 58% acceptable against 29%. That suggests a practical structure for long-standing clients: announce the new standard rate, and give them a named loyalty discount that tapers to zero over one or two billing cycles. They experience the loss of a courtesy rather than an increase, and you arrive at the new rate on a fixed date. Label the discount as temporary in writing: the same paper notes that the most recent price becomes the customer’s reference unless the terms of the previous transaction were explicitly temporary. Do not open an open-ended negotiation; offer one bridge and let the client choose it.
What to say: three notice templates
These are starting points. Keep them short, specific and written. A notice that explains too much sounds like an apology, and an apology invites negotiation.
For a long-standing client.
“From 1 January, my standard rate for this work will be [new rate], up from [old rate]. I have held the current rate since [year], and in that time the cost of running the practice and the market rate for comparable work have both risen significantly. Because we have worked together for a long time, I will apply a loyalty discount of [x]% to invoices in January and February, so the new rate takes full effect from 1 March. Nothing changes in the scope or the way we work. I am glad to talk it through if useful.”
For a retainer at renewal.
“As we come up to the renewal of our agreement on [date], the monthly retainer for the next term will be [new amount]. This reflects increased costs and brings the retainer in line with my current rate for new clients. The scope stays as it is. If a smaller scope would suit you better at a lower figure, I am happy to propose one.”
For a client you are prepared to release.
“From [date], my rate for this type of work will be [new rate]. I understand if that no longer fits your budget, and if so I am happy to help with a handover and recommend someone for the work.”
Notice what none of them contain: scarcity (“I am in high demand”), urgency (“you need to decide this week”), or anything that points at how much the client depends on you. Those are the reasons the survey respondents punished.
When not to raise, and what to do instead
A raise is the wrong tool in three situations.
Mid-project or mid-crisis. Wait for the reset point. A raise timed to a client’s moment of need is the snow-shovel case.
When the work itself has become cheaper to produce. If AI tools now do much of what you used to do by hand, a pure rate increase on the old basis invites the question of what the client is paying for. That calls for a change of pricing model, not just a higher number; see our comparison of freelance pricing models and the guidance on what to charge for AI-assisted work.
When you cannot name the reason. “Everyone is raising prices” is not a reason a client can repeat to their own finance team. Costs, the market rate for the work, a larger scope or a better outcome are.
The CEO and the student in repricing
The CEO in you treats the rate as a capital allocation decision reviewed on a schedule, not a feeling revisited in a panic. Once a year, same month, you run Table 1 for your own numbers and decide. The student in you treats every repricing round as an experiment: you record who stayed, who left, what they said, and how the break-even line compared with reality, so the next round is based on your own evidence rather than on dread.
For the structure of the offer itself, pair this playbook with the productized service pricing framework and the broader guide to pricing your expertise in the AI era. And because a raise changes cash flow before it changes anything else, fold the new rate into your variable income budget the month it takes effect.
FAQ
How much should I raise my rates?
Use the inflation gap since your rate was last set as the floor: for a rate from 2023, US consumer prices have risen 9.1% to August 2026; for 2021, 22.4%. Add a market or value component on top only if you can name what justifies it. For most practitioners who have not repriced in several years, the floor alone is a double-digit number.
How often should I raise my rates?
Once a year, at a fixed and predictable point such as the new year or contract renewal, is easiest for clients to plan around and matches the principle that changes at a reset point are the most accepted. Even in a low-inflation year a small adjustment keeps your rate from sliding; over the twelve months to August 2026, prices rose 3.4%.
How much notice should I give?
At least 30 days in writing, and 60 days for retainers or clients with budget cycles. The purpose of notice is to let the client adjust their reference point before the new price applies.
Should I grandfather existing clients?
Grandfather through a bridge, not indefinitely. A time-limited loyalty discount that tapers to zero gets you to the new rate on a fixed date, and the fairness evidence suggests ending a discount meets less resistance than raising a price. Indefinite grandfathering turns your best clients into your worst-paid ones.
What if a client pushes back?
Restate the reason once, offer the single bridge you prepared, and if needed offer a smaller scope at a lower total rather than the same scope at the old rate. If they still decline, check Table 3: losing a client below the break-even share leaves your revenue intact and frees hours for better-paid work.
How many clients will I lose?
There is no reliable published figure for freelancers, and anyone quoting one is guessing. What you can know in advance is your break-even: after a 10% raise you can lose 9.1% of billings without losing revenue; after 20%, 16.7%.
Sources
- US Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers (CPI-U), US city average, all items and services, not seasonally adjusted, series CUUR0000SA0 and CUUR0000SAS, monthly data 2019 to August 2026.
- US Bureau of Labor Statistics, Employment Cost Index, wages and salaries for all private industry workers and for private industry workers in the professional, scientific and technical services industry, current dollar indexes, series CIU2020000000000I and CIU2025400000000I, quarterly data 2019 to the second quarter of 2026.
- Kahneman, D., Knetsch, J. L. and Thaler, R. H. (1986). Fairness as a Constraint on Profit Seeking: Entitlements in the Market. American Economic Review, 76(4), 728-741. Survey design, samples and the acceptable and unfair shares for each scenario in Table 2; the principle of dual entitlement.
Tables 1 and 3 are CEOtudent calculations from the sources above. Table 4 and the notice templates are CEOtudent editorial frameworks, not findings of the cited studies.
This content was compiled with the support of AI following in-depth research, then written and prepared for publication by the CEOtudent editorial team.
This post is also available in:
