The coaching profession nearly doubled its revenue in two years. The 2025 ICF Global Coaching Study puts global coaching revenue at $5.34 billion, up from $2.849 billion in 2023, earned by a record 122,974 practitioners (15% more than two years earlier). In the same stretch, the biggest name in online homework help went the other way: Chegg cut 22% of its workforce in May 2025, then another 388 roles, about 45% of staff, that October, blaming students’ rapid adoption of generative AI and falling Google referrals.
Both businesses sell expertise. One grew and one shrank because they sell different things. That difference is the most useful lens for anyone choosing between coaching, tutoring, and consulting in 2026, and it is a better guide than the usual advice to “pick the one that matches your personality.”
The question that sorts all three models
Ask what the buyer is actually paying for when they hand you money. Not what you know; what they get.
- A tutoring client pays for understanding. They want a concept explained, a skill taught, an exam passed. The deliverable is knowledge transferred into their head.
- A coaching client pays for change. They already know roughly what to do. They pay for structure, accountability, and someone who notices when they drift.
- A consulting client pays for a fixed problem. They want a diagnosis and a recommendation, and they want a person who stands behind it. The deliverable lands in their business, not their head.
This framing matters now because AI is very good at the first deliverable and much weaker at the other two. A chatbot can explain the quadratic formula at 2 a.m. for free. It cannot sit across from a manager who keeps avoiding a hard conversation and hold them to it next Tuesday, and it cannot sign off on a vendor migration and take the call when it goes wrong. Chegg’s collapse was the commodity end of the “understanding” market getting repriced to zero, which is exactly where you would expect it to happen first.
Side by side: what each model sells, and to whom
| Tutoring | Coaching | Consulting | |
|---|---|---|---|
| Buyer purchases | Understanding of a subject or skill | Behavior change, with accountability | A diagnosed, solved business problem |
| Who usually pays | Parents, students, adult learners | Individuals, or employers for their staff | Businesses, from budget lines |
| Natural pricing unit | Per hour or lesson | Per program or engagement | Per project or retainer |
| Exposure to free AI | High for homework help and test prep; lower for live, high-stakes, or hands-on skills | Low for accountability; moderate for generic advice | Low for judgment and sign-off; high for research and slide production |
| Typical marketplace take | Wyzant keeps 25%; Preply takes 100% of trial lessons, then 18% to 33% | Usually direct; newer profile tools like LinkedIn Advice Sessions take 0% for now | Almost always direct, via referral or network |
| Sales cycle | Days | Weeks | Weeks to months |
The fee row is the one people skip. Wyzant’s help center says tutors keep 75% of their posted rate, but they keep the full rate for students they refer themselves. Preply’s commission model keeps the entire first trial lesson with each new student, then takes 33% from new tutors, with the cut falling toward 18% as your total teaching hours accumulate. Both structures teach the same lesson: the marketplace is a client acquisition channel you pay for on every hour, and the cheapest tutoring hour is one with a student you found yourself.
Tutoring: still a real income, but not the homework-help version
Tutoring has the fastest path to a first dollar of the three. The marketplaces bring demand to you, the sales conversation is short, and nobody asks for a credential beyond evidence that you know the material. That is still true. What changed is which tutoring is worth doing.
The work that survives AI looks less like “explain this worksheet” and more like one of these:
- High-stakes, deadline-driven prep where a missed result costs real money: professional licensing exams, graduate admissions tests, bar and CPA review. Students pay for a human who has seen where people fail.
- Skills that need a live human on the other end: conversation practice in a language, music, interview rehearsal, public speaking. A model can drill vocabulary; it is a poor substitute for someone who reacts to you in real time.
- Parents buying oversight, not answers. A weekly session that makes sure a teenager actually does the work is closer to coaching than to content delivery, and it is priced that way.
If your plan is generic homework help for high school math on a marketplace, the Chegg numbers above are your warning. If your plan is preparing nurses for a licensing exam you passed yourself, you are selling something AI makes easier to deliver, not something AI replaces. Our deeper guides on tutoring platforms and rates and building a premium tutoring business cover platform choice and pricing in detail.
Coaching: growing fast, and splitting by credential and niche
The ICF figures are the strongest evidence that people will pay humans for accountability even when the information is free. Dividing the study’s $5.34 billion in revenue across 122,974 practitioners works out to roughly $43,000 per coach. That is an average across full-time executive coaches and people coaching a few clients on weekends, so treat it as a sanity check on the income claims you see online, not as a forecast.
Two other findings from the same study are more useful than the headline:
- 73% of coaches say clients and organizations expect a coaching credential. You can legally coach without one in most places, but the buyers with the biggest budgets (employers paying for leadership development) increasingly screen for it. Whether the cost is worth it depends on who you sell to, which we break down in the life coaching certification guide.
- Most coaches already sell something else. The ICF reports 60% also offer training and 57% also offer consulting. The “pure coach” is the exception. In practice, coaching is often the relationship layer on top of a consulting or training business.
The coaching businesses that work tend to be narrow (a specific role, a specific transition, a specific industry) and priced by program rather than by session, because the change the client is buying takes months, not an hour. If you are weighing a program built around group sessions or a course, the hybrid coaching and course model is the next read.
Consulting: the highest ceiling, because the buyer has a budget line
Consulting is the slowest of the three to start and the most lucrative once it works, for one structural reason: the buyer is a business spending money it has already budgeted for a problem, not a person spending discretionary income. That changes the price conversation entirely. A parent compares your hourly rate to other tutors. A company compares your fee to what the unsolved problem is costing it.
The independent professional market has grown into that. MBO Partners’ 2025 State of Independence study, released September 9, 2025, counted a record 5.6 million U.S. independents earning more than $100,000 a year, up 19% from 2024 and 86% from 2020. The same study found 74% of independents use generative AI to improve their output, which says something about where the remaining value sits: AI is doing the research and drafting, and clients are paying for the judgment around it.
This is where my own experience lives. After 20+ years running IT operations at a large telecom, the fractional COO work I do through Ops Harmony has taught me that clients almost never hire for what you know. They can find the information. They hire someone to own the outcome: to look at their operation, say “this is the bottleneck, here is the order we fix it in,” and still be on the hook three months later when it is half done. Every engagement that went well had a clear owner and a clear scope. Every one that drifted had me supplying knowledge when the client needed a decision made.
If that description fits your career, the practical path in is your existing network and a narrowly defined problem you have solved before. Our guides on starting a consulting business without a Big Four pedigree and building a pipeline from LinkedIn go step by step.
Choosing, using the buyer test
Start from who would pay you and why, then let the model follow:
- If the most valuable thing you have is knowledge someone needs to learn, tutor, but move up the stakes ladder (exams, professional skills, live practice) and away from anything a free chatbot answers well. Use marketplaces to get your first students, then build a referral base you do not pay 25% to 33% on.
- If the most valuable thing you have is having made a change others are trying to make, coach, but pick a narrow audience and sell a program, not hours. Decide early whether your buyers will expect a credential.
- If the most valuable thing you have is years of solving a specific kind of business problem, consult, and expect a slower start with a much higher ceiling. Your first client is far more likely to come from someone who already trusts your work than from any platform.
Plenty of people end up running two of these at once, which the ICF data shows is the norm rather than a compromise. A consultant who coaches the leaders implementing their recommendations, or a tutor who adds an accountability program for parents, is simply selling understanding, change, and solved problems to the same buyer. The test still applies to each offer separately: be clear about which one the client is paying for, and price it that way.
For more ways to turn experience into income beyond these three models, see seven ways to monetize what you already know, and for the bigger shift behind all of this, read our analysis of why AI labs now pay credentialed experts $200 an hour while generalist work gets cheaper.