Chess Coach: The Real Price Table and When You Need One

Search "chess coach" and you are, whether you realize it yet or not, about to spend somewhere between ten and two hundred dollars an hour on another person's time. Every page ranking for that phrase either sells coaching directly or lists people who do: a directory, an academy, a retailer's blog, a marketplace. Not one of them asks how good you already are before recommending anything. This page does that first, then publishes the real price table those pages keep behind a click or a message, reports honestly that no neutral source agrees on when coaching starts paying off, and shows with our own numbers what a coach actually catches that a green evaluation bar does not.

What the directories actually show you

Chess.com runs the highest authority coach directory on the internet, searchable by title, country, and name. Checked directly today, its cards show a photo, a title badge, and a short bio, and across every visible profile fetched, exactly one coach stated a rate in plain text: "ONLINE CHESS LESSONS, $60." Every other card requires opening the individual profile or messaging the coach to learn what they charge. There is no price field, and no way to sort by price. That is not a broken page. It is the directory's actual design, and it means the highest authority result for "chess coach" cannot answer the first question a buyer has.

Lichess runs the honest counterexample. Its own coach directory requires a coach's national or FIDE title to be confirmed by Lichess staff before they are listed at all, and it shows real, self set per coach rates directly on the page: $25 an hour, $75 an hour, "From 30 EUR," "40EUR/h." Coaches set their own prices independently, and nothing on the page suggests Lichess takes a cut. If a free platform can publish real numbers next to a verified title, opacity elsewhere on this query is a choice, not a limitation of the format.

Two secondary marketplace pages made that opacity harder to check than expected. Wyzant's chess lessons page and Superprof's online chess lessons page both advertise a low headline rate, $15 an hour and $10 an hour, and both returned a blocked response to a same day fetch attempt. Those two figures were genuinely found in search indexing, so they are reported here, but they could not be independently reconfirmed today, and a headline rate on a lead generation marketplace page is exactly the kind of number that functions as a hook rather than a representative price. Treat both as unverified as of September 2026.

The real price table, by title

Cross referencing five independent sources, two Chess.com forum threads, CircleChess's own published breakdown, EloChaser's own published breakdown, and Lichess's live coach listings, converges on a rough scale that no single source states on its own but that all five roughly agree with:

The two Chess.com forum threads used to help build that table both quote the identical anecdotal scale from posters, attributed to a member writing under the handle tygxc and others: Super GM around $200 an hour, GM around $100 an hour, IM around $50 an hour, FM around $20 to $25 an hour, Candidate Master around $10 an hour. A separate UK based responder in the same threads quoted GBP 20 to 60 an hour for club level coaches and GBP 70 to 150 an hour for top coaches. None of this is a survey. It is forum posters stating their own personal figures, and it is reported as exactly that, but five loosely independent accounts landing on the same rough shape is a meaningfully stronger signal than any one of them alone.

CircleChess, which is itself in the business of selling group chess coaching, publishes a similar table: $20 to $40 an hour for untitled and club level lessons, $40 to $70 for an FM, $60 to $100 for an IM, and $100 or more for a GM. It also lists its own group format at $2 to $3 an hour, small group live classes with a titled coach at $89 to $99 a month, and premium subscription platforms combining video and live instruction from a GM at $79 to $129 a month. EloChaser, which runs its own coach marketplace, publishes a close variant: $15 to $30 an hour for club level coaches rated roughly 1600 to 2000, $30 to $60 for titled or near titled experienced coaches, and $60 to $100 or more for masters and well known international titled coaches. Two single business examples fall outside that pattern on the low end: MyChessTutor, which pairs students with a roster of instructors over Zoom, states one flat starting rate of $45 a session with no published per instructor breakdown, and House of Staunton, a chess equipment retailer rather than a coaching business, tells its blog readers lessons run anywhere from $10 to $150 an hour without narrowing that further for any particular reader.

Nobody agrees on when it starts paying off, and that is worth saying plainly

Every source above sells something, coaching, a marketplace, or equipment, and every one of them also has an opinion on when a coach is actually worth the money. Those opinions do not agree, and naming the incentive behind each one is more useful than pretending a consensus exists.

CircleChess's own coaching ROI article gives the clearest rating banded recommendation of any source found: beginners, rated 0 to 1000, should use free resources and group lessons rather than a paid coach; intermediates, 1000 to 1600, should mix free platforms with occasional paid sessions; advanced players, 1600 to 2000, are where "regular coaching becomes valuable for breaking through common plateaus"; and experts, 2000 and above, are where "premium coaching with titled players becomes essential." It also names a stagnation trigger rather than a fixed number: improvement stalling for six months or more is the signal worth acting on. That is a company that sells coaching telling its own beginner readers to wait. It is worth taking seriously precisely because it argues against the seller's own immediate upsell.

A Chess.com contributor blog states close to the opposite emphasis. "A chess coach is definitely worth it, especially if you are a beginner chess player," it argues, recommending coaching most for beginners and for serious, goal driven players, while suggesting casual hobbyists can skip it. It recommends two to three weekly sessions below a 1600 rating and one to two sessions above that, and names YouTube, Chess.com's own paid content, and books as the real alternatives to a coach, without ever mentioning Chessable or Aimchess.

EloChaser, a coach marketplace, ties the decision to hitting a plateau rather than to any specific number, and then says something that cuts directly against its own business: "For most players under 1600, a $20 coach and an $80 coach will improve you at a similar rate," arguing that consistent work between sessions matters more than what you pay for the session itself. A marketplace has every reason to steer a shopper toward the pricier listing. Saying the opposite out loud, in its own published article, is what makes that specific claim more credible rather than less.

A fourth number comes from neither a business nor a study. On a Chess.com forum thread titled "Are chess coaches actually worth it," one responder notes that many coaches simply "don't teach players below 1500 rating," which sets a practical floor around 1500 based on which coaches will accept a student, not on any pedagogical argument about when coaching helps. A player in the same thread, self reporting a 2400 rapid and 2538 blitz rating with no paid coaching at all, argues "all the information you need to reach a Master level is already at your fingertips for $0." One highly rated self taught player is not a representative sample, and the thread is included here for exactly what it is: one more genuinely held, sourced opinion, not a settled answer.

Four sources, four different lines: 1000, 1600, a plateau with no number attached, and roughly 1500 set by coach supply rather than pedagogy. No neutral study measured this. The honest answer is that the threshold does not exist as a fact you can look up, only as a spread of stated, attributed opinions, and this page is not going to manufacture a number that none of its own sources actually agree on.

When a human coach genuinely is the right call

Say this plainly, because none of the pages that currently rank for this query do: a human coach is sometimes exactly the right thing to pay for, and pretending otherwise would be its own kind of dishonesty. Three situations, drawn straight from the sources above, make the case on their own merits. You have plateaued for six months or more despite genuine effort, which is CircleChess's own stated trigger and a reasonable one even coming from a seller. You are past the point where free tools cover the gap and want a repertoire and a study plan built specifically around your own recurring mistakes, which is a job no directory, marketplace, or automated score currently does. And you want a second, human check on what an automated summary is telling you, because even a well built score can mislead: GM Noel Studer has argued that Chess.com's own post game Accuracy score is a poor measure of how well someone actually played, since it does not weight a single game losing blunder any differently from a string of minor inaccuracies, and he recommends "simply ignoring this score altogether" on its own. A coach watching the actual game, rather than a single blended number, catches what that number cannot.

What the free tools already do, said out loud

Before paying anyone by the hour, it is worth being straightforward about what is already free. Lichess gives away, with no account upgrade and no paid tier of any kind, unlimited Stockfish analysis and free Maia human move bots, funded by donations rather than subscriptions. That is a real engine and a real model of human play, both for nothing. If your actual problem is remembering an opening you already study, Chessable's own subscription runs $11.99 a month as of September 2026, which is less than a single hour with almost any coach in the table above; we cover that product and its real alternatives in our own breakdown of Chessable pricing and alternatives. And if you are trying to compare AI coaching products specifically rather than hire a person, that is a different question from the one this page answers: an AI branded product called Chessvia actually ranks on this exact search term today, which is a real sign the two categories overlap, and the honest move is to send you to the honest AI chess coach buyer's guide rather than pretend this page already covers it.

One number is genuinely unsettled and worth naming rather than guessing at: Chess.com's own current membership pricing. Its live Help Center article still describes a Gold, Platinum, and Diamond tier structure, a separate Chess.com contributor blog post claims Platinum was discontinued and replaced, and a third party pricing page shows yet another, internally inconsistent set of numbers. Nothing found today resolves that disagreement. If Chess.com's membership price matters to your decision, check chess.com/membership yourself before paying anything.

Chessdrive, the app we build, costs $5.99 a month or $39.99 a year, and it is worth saying plainly what that buys next to a free alternative that already exists: Lichess gives away Stockfish analysis and Maia's human move bots for nothing, and Chessdrive runs the same kind of on device human move model to do one specific job well, which the next section shows directly, rather than replacing what a coach does over months of actual sessions.

The move you could have found, measured

Here is the thing a directory implies only a paid human can tell you: whether a mistake you made was a mistake a player at your own rating would typically make too, or whether it is the kind of thing that should already be fixable. We measured it directly rather than asserting it.

The position comes from the finish of the Opera Game, Paul Morphy against the Duke of Brunswick and Count Isouard, Paris 1858, at the point reached after White's sixteenth move: FEN 4kb1r/p2n1ppp/4q3/4p1B1/4P3/1Q6/PPP2PPP/2KR4 w k - 0 16. Stockfish 17, run to depth 20, says White's queen move Qb8+ is mate in two. The other natural looking queen move, Qxe6+, throws the mate away entirely: the evaluation swings from a forced mate all the way to -1.25, meaning White, who had a mate on the board, is now simply worse. One move ends the game. The other looks similar, checks the king, and is simply wrong.

Maia3, the same on device human move model Chessdrive runs, gives the probability a real human at a given rating actually plays each move in this exact position:

At 1000 and at 1200, the losing capture is more likely to be played than the mate. Somewhere around 1400, that flips, and from there up the mate stays the more probable choice. Look closely at what actually separates a 1200 from a 1500 here: it is not finding the brilliant move, since both ratings find Qb8+ around 15% of the time either way. What changes is that the higher rated player has stopped preferring the plausible wrong one. That is not raw calculation ability going up. It is a specific, fixable, nameable habit going away, and it is exactly the kind of thing a good coach charges an hourly rate to sit across from you and point out.

A second number from the same dataset argues the point even more directly, and it would be dishonest to leave it out because it complicates the story rather than simplifying it. Averaged across the seven non forced, non trivial positions in this same finish, findability barely moves with rating at all: 0.1683 at 1000, 0.1648 at 1200, 0.1638 at 1400, 0.1637 at 1500, 0.1644 at 1800, 0.1682 at 2000, and 0.1765 at 2200. In a sharp, forcing attacking sequence like this one, rating is close to flat on the plain question of whether you find the engine's actual best move. Being 800 rating points stronger buys you almost nothing on raw findability here. What it buys, per the table above, is not repeating the specific wrong idea once you have seen it go wrong.

Both numbers come with the same caveat, stated once and plainly: this is one position, from one historical game, not a study of a rating band across thousands of positions. It illustrates a real mechanism. It is not a population statistic, and it should not be read as one. The pipeline behind it was checked before being trusted: the same Maia3 setup was re run at rating 1400 against 264 already frozen move probabilities from an earlier analysis pass, and the two runs matched to a mean absolute error of 0.0024, close enough to treat the numbers above as a faithful read of what the model actually outputs rather than a fluke of one run.

How to actually decide

Start with a plain rating check, not a marketing claim: if you have not yet worked through what Lichess already gives away for free, an analysis board, unlimited puzzles, and Maia bots that play like a real human at your level, that is the cheapest and fastest thing to exhaust first, and our own guide to reviewing your own games walks through how to actually use that engine output instead of just staring at a number. If you keep making the same category of mistake across games rather than a new one each time, especially around a plateau like the one covered in our piece on why players plateau around 1200, that repetition is the signal worth paying attention to, coach or no coach. And if you have genuinely plateaued for six months, want a repertoire built around your own patterns, or want a second human check on an automated score you no longer trust, the price table above is the real range to budget against, not the click through page that hides it.

None of this requires abandoning free tools to justify a coach, and it does not require pretending a coach is unnecessary just because Lichess is free. Both things are true at once. The honest version of this page's job is smaller than either extreme: tell you what people actually charge, tell you plainly that nobody agrees on exactly when to start, and show you, with a real number, one specific thing worth paying someone to fix.