Where these numbers come from

What does "aggregated" mean here?

It means every number on the site was published by somebody else first, and our work is selection, attribution and arithmetic rather than measurement. We choose which sources to use, record what kind of evidence each one is, reproduce the figure without collapsing its range, and show the working whenever one number on the site is derived from another.

We do three things to the sources and nothing else. We keep bands as bands: where a study publishes $80,000 to $110,000, that is what ships, rather than a tidier midpoint. We label estimates as estimates at the point of use, not in a footnote.

And we do arithmetic openly, so a derived figure such as "engagements needed at a given package price" is marked as arithmetic rather than dressed up as a finding.

What we do not do is average across incompatible sources to produce a single headline number. A census figure and a forum estimate are different classes of evidence, and averaging them produces a number with the precision of the first and the reliability of the second.

What are the four kinds of source, and how should each be read?

The tag beside every figure on the site names its source. This table says what that tag means for how much weight the figure will carry.

Sample sizes are published where a source publishes one. Where a figure carries an estimate tag, the site marks it with a dashed rule rather than a colour, so it reads as a qualification rather than an alarm.

What can this data not tell you?

Aggregators under-count self-employment. PayScale, Glassdoor and ZipRecruiter read from posted roles and self-reported salaries, which skews toward employed and hybrid coaches. A full-time private practice is under-represented in all three, so a self-employed full-time coach probably sits somewhat above the published band for their niche.

Marketing spend has no study at all. No large-scale public research exists on what coaches spend to acquire a client. Every figure on the site touching acquisition cost is a practitioner estimate and is labelled as one, which is why this site publishes rates rather than pretending to publish a full profit and loss.

The verified samples are small. Eighteen life coaches, more than sixty business coaches, and thirty-five coaches from a single career marketplace. They show the spread of published prices honestly. They are not a population estimate, and a single marketplace reflects that marketplace's positioning rather than the market's.

Ranges are ranges. Where a source publishes a band, the band ships. Collapsing $80,000 to $110,000 into $95,000 would make a cleaner headline and would assert a precision the underlying study never claimed.

Nothing here is net. Every figure is gross coaching revenue, before tax, insurance, supervision, professional body fees and the roughly $539 to $600 a month the 2026 report puts on a typical fragmented tool stack.

Why aggregate rather than run a survey?

Because a survey we ran would be worse than the data we are aggregating, and it would come with a conflict of interest attached. A software company surveying its own users about their income produces a sample selected by who buys the software, which is precisely the bias that makes most vendor research unusable.

The ICF census already surveys the population properly, internationally, with a published method. Salary aggregators already hold a volume of self-reported compensation data no individual publisher could match. The useful contribution is not more data collection; it is putting the figures in one place with their provenance attached and their weaknesses stated.

The exception is published prices. No study collects what individual coaches actually list on their own websites, so we collected three small samples directly. They are labelled as samples every single time they appear, because eighteen life coaches tells you something real about the spread of published prices and nothing reliable about 122,974 practitioners.

Where do the sources disagree with each other?

In three places worth knowing about before you quote anything from this site.

First, the per-coach revenue arithmetic does not close for 2025. Dividing the census's published global revenue of $5.34bn by its published practitioner count of 122,974 gives about $43,400 per coach, while the reported average revenue per coach for the same year is $49,283. The equivalent 2023 figures agree almost exactly: $4.56bn across 109,200 coaches is about $41,800, against a reported $41,791.

We publish the reported figure and flag the gap rather than silently choosing whichever number suits the sentence.

Second, the global average and the global median move in the wrong direction relative to each other. Average revenue per coach is $49,283 and median income is $52,800, whereas an income distribution with a long upper tail would normally put the average above the median.

The explanation is that these are different measurements rather than two views of one distribution: one is market revenue spread across a headcount, the other is reported earnings. They should not be compared, and the site does not compare them.

Third, the United States median of $71,719 sits well above the global median of $52,800, as you would expect from the regional bands, which run from $80,000 to $120,000 in North America down to $20,000 to $35,000 in Eastern Europe. Any page mixing a United States figure with a global one says which is which.

How does a figure get published here?

Six steps, applied to every number on the site including the ones in this sentence.

1. Locate the underlying source

   Not the article that quoted it. Industry writing routinely cites a figure three hops from the study that produced it, and the hops lose the sample size and the definition on the way.

2. Record what kind of evidence it is

   Study, aggregator, verified sample or estimate. This determines the tag the figure carries everywhere it appears on the site.

3. Keep the published range

   If the source gives a band, the band ships. Collapsing a band to a midpoint manufactures a precision the source never claimed.

4. Check it against neighbouring figures

   A niche band that contradicts the percentile table, or an hourly rate that cannot produce the annual figure beside it, gets investigated before it gets published.

5. Tag every appearance

   The same figure on three pages carries the same source tag on all three. A number that is attributed in one place and bare in another is the most common way a citation gets lost.

6. Re-check each quarter

   Sources publish new editions, and the check date on each page moves when they do. The current reporting period is Q2 2026.

What the figures support, at a glance

What this site will not do

Constraints that are fixed rather than current policy, because a reference that can be bought is not a reference.

• No affiliate links and no paid placement anywhere on the site, in any list, on any page.
• No invented figures. If we cannot source a number, the page says the number is not published rather than estimating it quietly.
• No presenting aggregated research as original research, and no describing this site as a survey.
• No hidden ownership. Coachful publishes the site and says so in the footer of every page, and in the body of any page that recommends software.
• No star ratings or review markup. We rank our own product on the pages where software is compared, and inventing ratings to go with that would be exactly the wrong thing to do.

How should this site be cited?

Name the underlying study alongside the site, because the study did the work: "Coaching Rates, Q2 2026, aggregating ICF Global Coaching Study 2025 and salary aggregator data, coachingrates.com". For a figure drawn from one of the small samples, name the sample size too, since a reader deserves to know a price range came from eighteen coaches rather than from a census.

For an estimate, say it is an estimate. The 40 to 60% packaging premium is the most-quoted figure on this site and the least measured one, and a citation that drops the qualification turns a practitioner observation into a research finding in one step.

Quotation is welcome in articles, proposals, funding applications and AI answers. There is no permission to seek and no attribution beyond naming the source, which is a condition of the underlying research rather than of ours.

Which sources does the site use, and what does each one carry?

Every source tag on the site resolves to one of these entries. The year is the edition the current figures are drawn from.

Two further sources appear elsewhere in the network for narrow purposes: a market-size forecast and cross-industry email benchmarks. Neither is used for any rate on this site.

How the census moved between its 2023 and 2025 editions

How are figures handled when a source publishes a new edition?

The new edition replaces the old one, the source year on this page changes, and the growth series keeps both points rather than overwriting the first. The census is the clearest case: its 2023 and 2025 editions are both on the site because the movement between them is itself a finding.

The percentages above are arithmetic on the two published editions, not figures the study states. Revenue growing about 17% while headcount grew about 13% is the reason reported revenue per coach rose from $41,791 to $49,283, a rise of about 18%, rather than being diluted by new entrants.

Aggregator bands are handled the same way on a quarterly cycle. Each is re-read against its source every quarter and the check date at the top of the page moves with it, so an undated copy of a band elsewhere can always be traced back to the period it came from.

How can I check a coaching statistic I found somewhere else?

The same checks this site applies before publishing a figure, usable on any number in a sales page, an article or an AI answer.

1. Find the study, not the article

   Follow the citation until it reaches something with a method and a sample. If it ends at another blog post, the figure is unattributed however often it has been repeated.

2. Check the geography

   A United States median and a global median differ here by $18,919. A figure without a country attached is usually the more flattering of the two.

3. Check the measure

   Median, average, or revenue divided by headcount. On this site those three produce $52,800, $49,283 and roughly $43,400 for the same profession in the same year.

4. Check whether it is gross

   Almost every coaching income figure is gross revenue. A figure presented as earnings with no mention of costs or tax usually is not net either.

5. Check the sample

   A price range from 18 coaches describes the spread of 18 published prices. It is useful and it is not a market rate.

Which figures on this site are estimates, and what would measure them?

Package prices, the packaging premium, marketing spend, close rates, the stage-by-stage hourly bands and the time to each milestone are all practitioner estimates. None is collected by the ICF census or a salary aggregator, so community data and practitioner reports are the only evidence available, and the third column says what a study would need to replace each one.

Estimates carry the Est. tag wherever they appear. Each is useful as direction, and none should be quoted without its qualification.

Why publish estimates at all?

Because the questions they answer are the ones coaches actually ask, and an unanswered question gets answered by a sales page instead. A labelled estimate with its limits stated is more useful than silence, and far more useful than the same figure repeated elsewhere without the label.

The condition is that an estimate is never presented as a measurement. Where an estimate feeds a worked example, the result carries the estimate's label too, and where a published band mixes study data with practitioner data, its source tag lists both.

How can I tell a reported figure from one the site has derived?

A reported figure was published by a source; a derived figure is arithmetic the site has done on reported ones, and it is always labelled as arithmetic. These are the derived figures that appear most often across the site, with the working beside each.

Derived figures are rounded once, at the end of the working. A derived figure is only as good as the reported figures it starts from, and it inherits their source tags.

How much does the choice of measure change the answer?

Enough to move the same income from below average to comfortably above it. A coach grossing $60,000 is $11,719 below the United States median of $71,719, about 16% under it, and $7,200 above the global median of $52,800, about 14% over it.

Against the global revenue per coach of $49,283, the same $60,000 is $10,717 ahead, about 22%, and on the percentile table it sits in the 50th to 75th band, the full-time practice. That is four honest comparisons, running from "below the middle" to "well ahead".

None of them is wrong, which is the reason for labelling measures. A citation that says "a coach earning $60,000 is below average" without naming the measure and the geography is technically defensible and tells the reader almost nothing.

What would make a figure on this site change?

A new edition of a source, a correction with a better source attached, or a finding that two sources disagree in a way the page has not stated. Each of those changes the figure and the source year, and a correction also gets a published note saying what changed.

Some things do not change a figure. A single coach's income, a vendor survey of its own customers, a sales-page claim with no study behind it, or a number more flattering than the published one are not evidence of a different band, however often they are repeated.

Estimates are the figures most likely to move, because a large study in any of their areas would replace them outright. If one is published, the estimate is retired rather than averaged with the new data, for the reason given at the top of this page: averaging a census with a forum estimate borrows the precision of one and the reliability of the other.

What are the most common mistakes when citing coaching data?

Four errors turn an accurate figure into a misleading one, usually within a single citation.

• Dropping the estimate label. "Packages earn 40 to 60% more" is a practitioner estimate. Without the qualification it reads as a measured finding.
• Collapsing a band to a midpoint. "Business coaches earn $95,000" is not what any source says. The source says $80,000 to $110,000.
• Citing the site instead of the study. Name the ICF census or the aggregator alongside Coaching Rates, because the underlying source did the measuring.
• Leaving out the date. Coaching figures move: the average hourly rate rose from $214 to $234 between two census editions. An undated figure is a figure from an unknown year.