A TOEFL teacher clearing $22 per hour on Preply and a TOEFL teacher clearing $56 per hour on Wyzant are selling the same product. The credentials are the same — an MA in TESOL, ten years of test-prep experience, certifications, glowing reviews. The hours of preparation are the same. What is different is the buyer. Different students reach those tutors, from different countries, with different expectations about what TOEFL tutoring should cost.

That gap is what this issue is about.

We scraped 1,135 TOEFL-tagged tutors across Preply, italki, and Wyzant on May 14, 2026 — the three biggest English-tutor marketplaces by user traffic. Cleared rate medians by platform:

Source: Chalk Index scrape, May 14, 2026, Italy-IP authenticated browser session. Platform supply counts pulled from each platform's filter UI same date.

Three platforms, three TOEFL markets, two clearing bands — $20–$22 on the two globally-distributed marketplaces, $50–$56 on the US-anchored one. The mechanism is geographic, and the rest of this issue is what that looks like in practice for someone trying to teach TOEFL for a living.

There is a separate question worth asking first, though: are these three platforms actually where the TOEFL prep market lives? The marketplaces talk about themselves as though they are. The data is more complicated.

ETS doesn't publish current annual TOEFL test-taker volumes. The most recent Test and Score Data Summary (2024) reports score distributions and country breakdowns but no aggregate count. The most recent independent estimate puts the figure at over a million annual iBT test-takers — but that number is dated, and the test has since faced material competition from IELTS, the Duolingo English Test, and other alternatives.

What we can verify from primary sources: TOEFL scores are accepted by more than 12,500 colleges, universities, and licensing agencies across 160+ countries (ETS, 2024 Test and Score Data Summary). The test is administered at 4,500+ test centers in 190+ countries.

Total TOEFL-tagged tutor supply across the three marketplaces we scraped is roughly 6,100 listings — 4,456 on Preply, 1,206 on Wyzant, 484 on italki. Those are listings, not unique tutors; many tutors list on multiple platforms. The actual unique-tutor count is likely 3,500–4,500.

Whatever the current annual test-taker total — 800,000 or 1.5 million or higher — those 4,000-or-so marketplace tutors aren't carrying most of the prep load. Even at one tutoring session per test-taker per year, each tutor would need to deliver hundreds of sessions annually. The math is implausible. Most TOEFL prep happens through other channels.

What we did not measure:

We measured one channel — the three biggest English-tutor online marketplaces by user traffic — and within that channel, tutors who self-tag TOEFL in their skill list. Self-tagging is not the same as TOEFL specialism: on Preply alone, 485 tutors list IELTS, TOEFL, TOEIC, and PTE simultaneously, which strains any honest definition of "specialist" (see methodology section below). The data is publishable as a comparison within that slice. It is not a complete picture of TOEFL prep, and the medians inside it are noisier than they look.

What is outside the slice, briefly: the channels above, plus section-specific tools — products optimized for one of TOEFL's four sections (Reading, Listening, Speaking, Writing) rather than all four. We come back to that below.

The rest of this issue is what those three marketplaces show. The numbers are real and they describe real markets. They describe partial ones.

The mechanism behind the rate gap is geographic. Each of the three platforms aggregates a different national or regional buyer pool, and each buyer pool has a different willingness to pay for TOEFL prep.

The cleanest evidence is traffic geography. Similarweb's April 2026 measurement:

Source: Similarweb, April 2026, "All Traffic" mode. Includes both buyer and tutor sessions.

Wyzant's TOEFL tutor pool is 1,206 listings — 21% of Wyzant's total English-tutor supply (5,767). That is the highest TOEFL specialty concentration of any of the three platforms. Wyzant is overwhelmingly American by traffic: 84% US, the other 16% scattered across Indonesia, Kenya, India, and Canada at single-percent shares.

The buyer pool Wyzant serves is the US-bound international student — a high-school junior or senior preparing for US university admission, or a graduate-school applicant. These buyers come from families paying US private-tutor rates ($40–$100/hour for any subject), often through US-school referrals.

The signals all point in the same direction:

The cleared rate of $56 for TOEFL on Wyzant is the same band as Wyzant English generally ($50 platform-wide median from Wyzant's own exit-popup math). TOEFL is not a premium specialty on Wyzant — it is the default Wyzant rate, applied to TOEFL prep by a buyer pool that pays it for any subject.

italki's TOEFL pool is 484 listings — 11% of italki's 4,305 English tutors. Smaller absolute count than Wyzant; larger share than Preply.

italki structures supply differently from Preply or Wyzant. Every tutor on the platform applies as one of two types: Professional Teacher (training as an educator, prepared materials, structured lesson plans, $10–$120/hour pricing floor) or Community Tutor (conversation practice, no formal credential requirement, $5–$120/hour pricing floor). A tutor chooses one category at application and cannot list as both. In our 313-tutor TOEFL sample, 77% are Professional Teachers and 23% are Community Tutors.

The rates displayed on italki search-card listings are trial-lesson rates, not transacted session rates. italki tutors set per-lesson-type prices on their profile pages: a Trial Lesson typically discounted to attract first-time bookings, and a separate TOEFL Prep lesson (or Business English, Conversational English, etc.) at the actual transacted rate. The $21 Professional Teacher / $14 Community Tutor trial-rate medians from our scrape are the listing-card displays; representative profile-level captures put the transacted Professional Teacher TOEFL session rate closer to $25–40 per 60-minute lesson. See methodology section for the per-lesson rate architecture.

As of Issue 1's reference date, italki has closed English-tutor applications entirely. Neither the Professional Teacher nor the Community Tutor track is accepting new English applicants (May 11–17, 2026 application window, captured directly from italki's policy page; other languages such as German and Catalan remain open). The English-tutor pool we scraped is essentially the entire addressable supply on italki right now — new entrants are not arriving. This is a supply-side constraint that doesn't exist on Preply or Wyzant, and it matters analytically: a closed-supply pool stabilizes at different rates than an open pool actively absorbing new applicants. The italki English numbers we report are closer to equilibrium than the Preply or Wyzant pools in that specific sense.

italki's traffic geography is dispersed: US 22%, UK 8%, Mexico 4%, Japan 4%, France 3%. The top five countries sum to only 41% of total traffic. The remaining 59% is a long international tail.

The most striking signal: Baidu — China's primary search engine — is italki's #1 referring website at 30.51%, growing +808% year-over-year. italki has substantial Chinese user flow that Wyzant and Preply show essentially none of. Japan in the top 5 (3.6% of traffic) is the world's largest TOEIC market by volume; italki captures some of that learner flow as well.

The italki TOEFL buyer is more likely to be an Asian or European student applying to a non-US university that accepts TOEFL, or cross-shopping between TOEFL and IELTS for global admissions. The trial-rate display anchor is $20 — close to italki's platform median trial display of $18, suggesting TOEFL is a small premium over generalist trial rates. At the transacted-session level, italki TOEFL rates run materially higher (typical Professional Teacher TOEFL: $25–40 per 60-minute lesson per profile-level captures).

Preply has the largest absolute TOEFL tutor pool (4,456 listings, more than 3× Wyzant's) but the lowest TOEFL share of pool (10.5%). Preply's English-tutor pool is the biggest of the three, simply because Preply itself is the biggest platform — 42,522 total English tutors, against italki's 4,305 and Wyzant's 5,767.

Preply's traffic geography is the most globally distributed of the three: 17% US, 6% UK, 6% Spain, 5% Italy, 4% Mexico. Spain and Italy combined are nearly as large as the US share — a Romance-language-learner profile. The largest single referring source is performance-marketing affiliate networks (Admitad, cashbackdeals.es, app.impact.com), all of which are growing 100%+ year-over-year.

The Preply TOEFL buyer is more likely to be a non-US-bound international student in Europe or Latin America applying to Canadian, British, Australian, or international-school programs that accept TOEFL alongside IELTS. The buyer pool clears at $10 platform-wide. TOEFL on Preply clears at $22 — a roughly 2× lift over Preply's generalist median, the largest within-platform specialty premium of the three.

The platform a TOEFL tutor lists on is not a commission-rate decision. It is a buyer-pool decision. Each platform's TOEFL buyer is a different person with different expectations, different geographies, and different willingness to pay.

Your buyer is most likely a US-based parent or US-bound international student paying US private-tutor rates. They expect a US-credentialed tutor, fast response time, and the ability to schedule in US time zones. Wyzant's algorithmic surface curates aggressively — the default-sort English search returns a 50-card surface against a 5,767-tutor pool; specialty searches like TOEFL expose more (we captured 582 unique TOEFL tutors). Competing for visibility means competing for slots in Wyzant's "best match" algorithm.

The rate ceiling is high: $56 median, p90 around $100, premium tier extending into the hundreds. Wyzant is also a multi-subject marketplace — TOEFL tutoring competes with math tutoring, music lessons, and SAT prep for the same buyer's attention.

Your buyer is most likely a non-US-bound international student in Europe or Latin America applying to programs that accept TOEFL alongside IELTS. The platform's buyer pool clears at $20–$25/hour for TOEFL prep — roughly 2× Preply's general English median. Preply's commission structure has two layers: a per-trial tax and a tenure-based tier. Trial lessons with every new student are taxed at 100% — Preply keeps the entire fee, and the tutor earns nothing on the first lesson with each new student, indefinitely. Paid lessons after the trial run a five-tier discrete scale based on a tutor's total cumulative hours taught on the platform: 33% for the first 20 hours, 28% from 21–50, 25% from 51–200, 22% from 201–400, and 18% past 400 platform hours. A tutor teaching 10 hours a week reaches the 18% floor in about ten months. Once at the floor, the only durable commission cost is the trial-lesson tax on each new student acquired — and that one never goes away. The rate is lower; the volume can be higher; the math depends on how fast a tutor builds book and how often they replace churned students.

Your buyer is most likely a globally distributed international learner — predominantly Asian (China, Japan) and European — for whom TOEFL is one credential being shopped alongside conversation practice and general English fluency.

italki's commission is not flat — it varies by lesson type: 0% on trials, 21% on single lessons, 19% on 5-lesson packages, 17% on 10-packs, 15% on 15- and 20-packs, 30% on group classes. A teacher selling mostly single lessons keeps 79% of listed rate.

italki also displays trial-lesson rates on search-card listings, not transacted session rates. The $20 figure cited above is the trial-rate display, used in this issue for cross-platform comparison because it's what a buyer sees on the search surface. Transacted TOEFL session rates on italki are set separately on each tutor's profile page and run $25–40 for a 60-minute Professional Teacher lesson. The Preply $22 and Wyzant $56 figures, by contrast, are listed transacted hourly rates. Direct rate comparisons between italki and the other two platforms are therefore noisy at the absolute level; the relative ordering (Wyzant high, Preply and italki cluster lower) holds, but italki's true transacted rate is meaningfully closer to Preply than the $20-vs-$22 gap suggests.

The wrong question to ask: "Which platform pays TOEFL tutors more?" The platforms do not pay tutors at all — tutors set their own rates, and the platform's buyer pool determines whether those rates clear. The right question is: "Which platform's buyer pool am I credible to, and can I close enough of their willingness to pay to earn a living?"

The TOEFL prep market is fragmented along two dimensions — channel and skill. The three marketplaces we scraped represent one slice of the channel dimension. They serve generalist TOEFL tutors offering all four test sections. They do not capture the channels operating outside marketplace structure, and they do not capture the section-specific tools that compete for a different buyer need entirely.

Several large TOEFL teaching channels are absent from our scan:

The TOEFL iBT has four sections: Reading, Listening, Speaking, and Writing. The 2026 update — launched January 2026 per the ETS Technical Manual — restructured the test in three meaningful ways:

This is the second axis of market fragmentation. All four sections contribute equally to the reported overall score (25% each, by averaging). Test-takers do not need equal preparation across all four — a typical student walks in with three strong sections and one weak one. A student weak only in Speaking does not need a generalist TOEFL tutor; they need targeted Speaking feedback.

The marketplaces we scraped surface generalist TOEFL tutors who offer all four sections. They do not surface tools optimized for one section.

My Speaking Score is the example of how specialized that other axis gets. It is a tool that measures only TOEFL Speaking. Not the whole TOEFL exam. Not even multiple sections. Just Speaking — the single section that ETS itself has standardized on AI-driven evaluation.

The structural alignment matters. A specialized AI Speaking tool is not competing against ETS's scoring approach; it is using the same kind of technology ETS uses to score the actual test. The marketplace generalist tutor, by contrast, is offering human coaching across four sections, only one of which a typical student actually needs help with — and on that one section, the student's score will ultimately come from an AI engine, not a human.

Tools like that compete with marketplace tutors for the same buyer dollar but solve a structurally different problem. A student who is weak only in Speaking can buy specialized AI scoring and feedback for less than a single hour with a marketplace tutor — and get it 24/7, with the same kind of evaluation method ETS itself uses for the real test. A marketplace tutor offering "TOEFL prep" does not appear in that buyer's search at all.

The implication for a marketplace tutor: you are not just competing with the other TOEFL tutors on your platform. You are competing with specialized section-specific tools for the most addressable slice of TOEFL prep demand — the student who needs targeted help on one section, not generalist coaching on all four.

Cleared rate scrape. 1,135 unique TOEFL-tagged tutors scraped May 14, 2026 from an Italy-IP authenticated browser session. Preply uses ?tags=toefl URL filter; italki uses ?keyword=toefl; Wyzant uses ?keyword=toefl&is_online=true&subject-was-verified=false. Sample sizes: Preply 240, italki 313, Wyzant 582. The scrape captures the rate text rendered on each platform's listing card.

italki rates are per-lesson-type — listing-card display is the trial rate, not the transacted session rate. italki tutors set separate prices for each lesson type they configure on their profile (Trial Lesson, TOEFL Prep, Conversational English, etc.). The rate displayed on italki search-card listings is the trial-lesson rate, which tutors discount to attract first-time bookings. The transacted TOEFL session rate is set separately on the tutor's profile detail page; representative captures put typical Professional Teacher TOEFL rates at $25–40 per 60-minute lesson, materially above the $21 trial-rate display. The italki $20 median in this issue should be read as a trial-rate-anchored listing-card display. Relative-rate-gap findings between italki and the other platforms hold directionally; the absolute italki number underestimates the transacted rate. Preply and Wyzant rates are listed-hourly-rate displays representing the rate a buyer transacts at; the per-lesson distinction is italki-specific. A proper per-lesson rate scrape across italki profile pages is on the work list for a future issue.

Preply specialty system is multi-checkbox-with-blurb. Preply tutors opt into specialties by checking boxes from a ~50-item list of English specialties (TOEFL, Business English, IELTS, BEC, ESL/ESOL, Conversational English, English for Job Interviews, and roughly 40 others), and writing a description-blurb per checked box. A Preply tutor with TOEFL checked is not a "TOEFL specialist" in any gated sense — they marked the checkbox and wrote a paragraph. The 4,456 Preply TOEFL count reflects how many tutors checked the TOEFL specialty box, not how many specialize in TOEFL.

Self-selection on platform skill tags. Anyone can add "TOEFL" to their skill list. The 1,135 tutors in our scan are tutors who marked TOEFL as a skill they offer — not 1,135 TOEFL specialists. On Preply alone, 485 tutors list IELTS, TOEFL, TOEIC, and PTE simultaneously — four tests with materially different scoring rubrics, formats, and test-taker populations. A tutor genuinely teaching all four at a serious test-prep level would be extraordinarily rare; the more likely interpretation is search-surface optimization, where checking every box maximizes the number of buyer queries the tutor's profile appears in. We have no way to separate, from the listings alone, credible TOEFL specialists from generalists who flagged TOEFL among many specialties. The medians in this issue should be read as "the median rate among tutors who list TOEFL" rather than "the median rate among TOEFL specialists." The actual specialist median is almost certainly higher — and likely closer to Wyzant's $56 than to Preply's $22 — because the box-checking generalist pool is concentrated on the price-competitive platforms.

Traffic geography. Similarweb, April 2026, "All Traffic" mode. Includes both buyer and tutor traffic; the directional finding (Wyzant 84% US vs italki 22% vs Preply 17%) holds regardless of supply-side / demand-side split.

TOEFL test volume. ETS does not publish current annual TOEFL test-taker totals. The ETS 2024 Test and Score Data Summary reports score distributions and country breakdowns but no aggregate count. The widely-circulated "2.3 million annual test-takers" figure appears in secondary sources but we could not verify it from any ETS publication. Historical sources from the 2010s placed the iBT figure at roughly one million annually; the test has faced material competition from IELTS, Duolingo English Test, and other alternatives since then. This issue uses verifiable ETS figures only (12,500+ accepting institutions in 160+ countries; 4,500+ test centers in 190+ countries) and avoids citing a specific test-taker total.

TOEFL has been restructured. The 2026 TOEFL iBT — launched January 2026 per the ETS Technical Manual — replaces the legacy 0–30 per section / 0–120 composite with a 1–6 CEFR-aligned band scale per section. The overall score is the average of the four section bands. Raw points (Reading 35, Listening 35, Writing 20, Speaking 55, total 145) are internal scoring infrastructure that gets converted to the 1–6 band via IRT equating — they are not the reported score. Speaking is now scored entirely by ETS's AI engine; both Speaking task types ("Listen and Repeat" and "Take an Interview") are evaluated by natural-language-processing and machine-learning models, with human raters providing training data and oversight rather than per-response scoring. Reading and Listening are two-stage adaptive (router + lower/upper module). Reported section weights in the overall composite are equal (25% each, by averaging). This issue references the 2026 framework where section-scoring and AI-evaluation details matter; the marketplace rate data is independent of the test format change (tutors charge by the hour, not by the section being prepped).

What we did not measure. TOEFL prep delivered through ETS-affiliated products, US test prep companies (Magoosh, Kaplan, Princeton Review), university IEP programs, in-country prep schools, agency placements, direct word-of-mouth tutoring, and section-specific tools (My Speaking Score, similar AI-scoring products). The data covers the online-marketplace generalist-tutor slice only.

italki application status and tier structure. italki classifies every tutor as Professional Teacher or Community Tutor at application, and gates supply by language application windows. As of the May 11–17 2026 window covering this issue's reference date, English-tutor applications were closed on both the Professional Teacher and Community Tutor tracks. Our scrape captures the addressable English supply at the time of measurement; that supply is not actively expanding via new applicants. Other italki language tracks (German, Catalan, several smaller languages) were open during the same window. The two tutor types have different pricing floors ($10–$120 for Professional, $5–$120 for Community), different credentialing requirements, and different listed-rate distributions — the blended platform median masks this segmentation.

Personalization. All three platforms personalize their default surfaces by IP, account history, and locale. Italy-IP results may differ from US- / India- / China-IP results. Multi-region replication from US, UK, Spain, Vietnam, India, and Brazil VPN endpoints is on the Issue 2 work list.

Single-point measurement. This is a snapshot. We cannot make trend claims ("rates rising," "supply growing") without multiple measurements over time. Similarweb's month-over-month change numbers are noted but not used as findings.

Issue 2 (May 21): TOEFL vs IELTS, head-to-head across the same three platforms. Two tests, two buyer populations, different national admissions footprints. Where do the rates align, where do they diverge, and how does the box-checking overlap from Issue 1 cut once we measure both tests at once?

Issue 3 (May 28): Business English. The vertical with the most direct corporate-buyer story. Does the platform-anchored segmentation hold when the buyer is an HR department instead of a student?

Issue 4 (June 4): Overseas EFL contracts in Japan and Korea. The two anchor markets for international English teaching — JET, EPIK, eikaiwa, hagwon. Listings, compensation packages, requirements. What does the full first-year package actually pay once housing, airfare, and severance are counted, and how does that compare to clearing rates online?

Data: 1,135 unique TOEFL tutors scraped May 14, 2026 across Preply (n=240), italki (n=313), Wyzant (n=582), from an Italy-IP authenticated browser session. Platform supply counts and traffic geography pulled from each platform's filter UI and Similarweb (April 2026) the same day. Unified dataset and analysis scripts available on request.