Business English: three platforms, three taxonomies

Chalk Index has spent the first two issues mapping the supply side of three online English-tutor marketplaces — Preply, italki, and Wyzant — through the test-prep verticals (TOEFL in Issue 1, IELTS in Issue 2). The structural finding has held twice: which platform a tutor lists on matters more than which test they teach, because each platform aggregates a different buyer pool with a different willingness to pay. This issue extends the framework to a different specialty entirely — Business English, the vertical with the most direct corporate-buyer story — and finds that the platforms don't just price the same vertical differently. They classify it differently. The taxonomy itself is the strategic decision.

Preply lists 16,481 Business English tutors. italki has 1,653 tutors with Business as their primary Specialty — and 4,034 who offer at least one Business-category lesson. Wyzant doesn't have a Business English category at all — its closest structured equivalent is the ESL/ESOL subject, with 2,781 tutors. Same vertical, same week, three platforms whose specialty-signaling architectures are not interchangeable.

The headline finding is the architecture difference, not the numbers. Preply has a single-layer specialty system: roughly fifty fine-grained English checkboxes (Business English, BEC, Business & Work, IELTS, TOEFL, ESOL, plus 40 others), each with a required description field. Tutors check the boxes that fit them; the description requirement makes opt-in mildly costly. italki has a two-layer system: a single primary Specialties tag the tutor picks at profile setup, plus a per-lesson configuration that lets tutors set up multiple lesson types under categories (Business, Conversation, Test Prep, Language Essentials). Adding a Business English lesson costs almost nothing in friction. Wyzant has neither — its taxonomy is organized around K-12 school subjects (English, ESL/ESOL, Business, TOEFL, IELTS each as distinct top-level categories), and Business English emerges only at search-time intersection.

Each platform's structured supply for Business English, expressed as share of its English-tutor base:

The structural insight: a platform's specialty-signaling architecture is the decision that shapes supply-side signals. Preply chose a single-layer checkbox system with description-blurb friction. italki chose a two-layer system with high-friction primary Specialty but near-zero-friction per-lesson configuration. Wyzant chose subject-categorization without Business English at all, leaving the category to emerge at search-time. The supply-side numbers downstream are responses to those three architecture choices.

Before any supply-side analysis, an important methodology aside. Worldwide Google Web Search interest for the phrase "business english" roughly tripled starting in late 2025 — from a baseline of ~20 to a sustained ~70 (peaking near 100 in early 2026) on Google's 0–100 normalized scale. If you stop reading there, you conclude demand is exploding.

YouTube Search for the same exact phrase, same time window, same global scope, has been stable in the 60–100 band for five years. No inflection point. No spike.

Demand for business-English information moved sharply on text search. Demand for business-English video tutorials didn't move at all. The divergence is the actual story. Three mechanisms most likely explain it:

Marketplace tutor listings — what this issue measures — sit downstream of all three of those buyer motions. We can't see HR procurement directly from a listings scrape. We can see how each platform has positioned itself to capture the shift.

Before discussing platform supply, the US buyer pool deserves a number. The most recent BLS data places foreign-born workers at 32.3 million — 19.2% of the US civilian labor force, a record share. That's the headline buyer base.

Three further numbers from primary census and labor sources sharpen the picture:

The Brookings breakdown by sector matters. The 47% LEP figure spans services, manufacturing, and agriculture, where Business English coaching isn't the natural fit. The buyer profile for a marketplace 1:1 Business English tutor is closer to the management/professional slice — non-native English speakers in white-collar roles where written and spoken English is daily currency. Even within that slice, the addressable Business English buyer pool runs into the millions.

The contrast is the finding. Wyzant — the only marketplace of the three that serves a primarily US buyer pool — has 2,781 ESL/ESOL tutors as its supply addressable to this demand base. Even at perfectly-utilized supply (40 hours per week, every week), the entire ESL/ESOL pool could deliver about 5.8 million tutor-hours per year — enough to support roughly 110,000 buyers in weekly one-hour sessions. Against a US LEP-professional buyer pool of roughly 5 million, that's structural undersupply by a factor of about 45×. The narrower keyword intersection (288 tutors whose profile text mentions "business english" specifically) covers about 11,500 weekly buyer-slots — closer to 430× undersupplied against the same buyer pool. Wyzant's top ESL/ESOL tutors clearing $90–$125/hour aren't priced into a niche premium tier by demand sophistication, they're priced by the rational market-clearing rate for a tiny supply against a US LEP buyer pool measured in millions.

The 1:1 marketplace channel is not the place this demand mostly lands. Corporate training providers (Berlitz, EF Corporate, GoFluent), in-house corporate learning programs, ESL community college courses, and AI-augmented self-study capture most of it. But the marketplace's small supply is structurally protected from rate compression because the demand-side floor is so high.

A separate dimension worth naming: small-business owners and the self-employed are disproportionately non-native English speakers. The SBA Office of Advocacy's 2024 analysis of 2022 ACS data finds 74.6% of US self-employed individuals speak English at home, compared to 78.1% of the general US population. Self-employment rates are 44%–53% higher among workers with moderate-but-not-native English proficiency than among native English speakers, consistent with the hypothesis that "limited English proficiency pushes workers toward self-employment because traditional employment is less accessible." Spanish (14.4%), Chinese (1.3%), Vietnamese, Korean, Portuguese, and Arabic round out the top languages spoken at home among US business owners. The Business English buyer in the US is, increasingly, the business owner — not the employee.

Microsoft's AI Economy Institute publishes country-level AI adoption rates twice a year. The most recent release covering full-period data — Global AI Adoption in 2025 — A Widening Digital Divide, published January 2026 — is the source for what follows. The methodology is more rigorous than Google Trends: aggregated and anonymized Microsoft telemetry adjusted for OS and device share, internet penetration, and population, published as the share of each country's working-age population (15–64) that has used generative AI in the reporting period. The full country-level table is in the report appendix.

For Chalk Index purposes, the interesting move is cross-walking Microsoft's country AI adoption rates against the buyer-pool geographies the three marketplaces serve (Wyzant 84% US, italki dispersed with strong Asia and Europe shares, Preply European and Latin-American tilt — per Similarweb April 2026 measurement). The buyer pool that determines whether a marketplace TOEFL or Business English tutor clears $20 or $60 is the same buyer pool whose AI adoption Microsoft is now measuring quarterly.

Source: Microsoft AI Economy Institute, Global AI Diffusion Report H2 2025. Buyer-pool mapping from Similarweb April 2026 traffic geography.

Three readings worth taking from this cross-walk:

A pattern the data suggests but doesn't prove: marketplace 1:1 tutoring demand should soften fastest in the highest-AI-adoption markets, and persist longest in the segments AI can't substitute (accent, spoken fluency, credential signaling). The data to confirm this requires a second Chalk Index scan separated by months, ideally with country-level buyer breakdowns.

The cleanest way to read each platform's strategy is the share of its English-tutor pool tagged Business English.

The two italki numbers measure different things in the same teacher pool. 1,653 teachers have Business as their primary Specialty tag — they self-identify as Business English specialists. 4,034 teachers have at least one Business-category lesson configured — they offer Business English alongside other lesson types. Both are real italki supply measurements. Which one a buyer encounters depends on whether they search by Specialty or by lesson keyword.

The 1,653 italki number is the direct comparison to Preply's 16,481 — both are structured-specialty signals where the tutor actively self-identifies. Both shares are essentially the same (38.6% on italki, 38.7% on Preply). The platforms differ on the broader supply layer: italki's per-lesson configuration produces 94.1% opt-in because adding a Business English lesson takes minimal friction; Preply has no equivalent broader layer because its specialty system is single-layer (checkbox-per-topic).

This isn't a survey of buyer demand. It's a survey of tutor-side response to each platform's specialty-signaling architecture. The three platforms produce three patterns:

The supply asymmetry above is downstream of a structural difference in how each platform has positioned itself for corporate buyers. Five publicly observable B2B feature dimensions:

Preply has every dimension. italki has a landing page and basic reimbursement support for individual employees, but no corporate dashboard, no procurement workflow, and no dedicated tutor program. Wyzant has none of it — Wyzant is consumer-direct only.

The most concrete signal of Preply's commitment is that Preply pays Google to appear on the "business english" keyword. A sponsored Preply result appears at the top of Google Search for that exact phrase, routing to a tailored Business English landing page with corporate-tier callouts ("Preply For Companies — Corporate Language Training through 1-on-1 Private Lessons"). The keyword exact-match identifier (kwd-77735651) and the structured ad-group naming (stu_sem_generic_web_0_eng_it_multiplesub_ex) indicate a mature, audience-tested campaign — not a casual buy.

This is the supply-side and product-side response to the Google Web Search trend chart above. While we cannot directly observe Preply's CAC or LTV on this keyword, the fact that the campaign continues to run at scale suggests it pays back — and that the demand spike on Web Search is, at least partly, real procurement-intent demand that the marketplace can monetize.

Each platform discloses its own Business English rate range on its public marketing surfaces. These are not our scrape; they are the platform's own claims about what buyers will pay.

Adjusting for italki's per-lesson pricing architecture changes the rate comparison. A naive read of search-card displays — italki $9–$25, Preply $3–$40+, Wyzant $35–$60 — would suggest italki is the cheap platform. It isn't. italki's search-card display is the trial-lesson rate; the transacted Business English session rate is closer to $27–$46, materially overlapping Preply's range. Wyzant remains the high-rate platform with a real premium tier; Preply and italki cluster in the $20–45 range once trial-rate distortion is removed.

A standalone look at Preply specifically. The platform's English-tutor supply is dominated by Business English to an extent that doesn't appear in any other English specialty pool we have measured.

16,481 Business English listings is more than three times the size of Preply's TOEFL pool (4,456) and roughly three to four times the size of any other specialty submarket. Preply has decided this is the bet that justifies the corporate-tier product investment.

Two reasonable interpretations of this composition:

The two interpretations aren't mutually exclusive. The comparison to italki's 94.1% number is the strongest available evidence that Preply's product layer creates some tag discipline — without that layer, the Preply share would be higher. With the layer, it's capped near 40%.

The honest position: 16,481 is the size of the supply pool that has positioned itself to capture Business English demand on Preply. It is not a count of credible Business English specialists. The credible-specialist count is materially smaller, and bounded above by something like the count of Preply tutors in the Corporate Tutor Program tier (a separate gated population Preply does not publicly disclose the size of).

The marketplace listings tell us about one slice of Business English buyer demand. They are not the whole picture.

Business English buyer demand has at least six distinct landing channels, only one of which our scrape measures directly:

We can quantify the marketplace slice. The other slices we cannot, from our current data. A reasonable hypothesis — to be tested in a future issue with a different data approach — is that the marketplace 1:1 tutor channel captures around 10% of total Business English buyer demand. The rest is split across the channels above.

That bounds the ambition of the marketplace data: it answers "what does the supply side of the marketplace channel look like," not "how big is the Business English market."

Three platforms, three strategic answers.

If you list on Preply. The platform's Business English supply is enormous (16,481 listings) but a meaningful fraction is tag-inflation. Your real competition is the subset of tutors who can demonstrate credible corporate-buyer credentials and who appear in the algorithmic top-of-sort surface or in the Corporate Tutor Program. Preply Business routes real B2B traffic to its top tier — if you can qualify into that surface, the rate ceiling is materially higher than the platform's $17 average. Outside that surface, you're competing against 16,000 other tutors in a low-rate band. Preply's commission structure has two layers: 100% on every new student's trial lesson (indefinitely), then a five-tier ladder on paid lessons that steps from 33% down to 18% based on the tutor's total cumulative platform hours.

If you list on italki. 4,034 of italki's ~4,287 English teachers (94%) have at least one Business-category lesson configured. 1,653 (38.6%) tag Business as their primary Specialty. Either way, what differentiates one italki Business English tutor from the others is the same as for any italki tutor: reviews, lesson count, per-lesson rate, Professional vs Community Tutor tier, and language pairs. The italki Professional Teacher tier displays $21 trial-rate on test-prep listing cards; transacted Business English session rates run $25–45 per 60-minute lesson per profile-level captures. italki's commission varies by lesson type (0% on trials, 21% on single lessons, sliding to 15% on package sales of 15+ lessons, 30% on group classes), so the listed-vs-cleared gap depends heavily on whether the teacher sells single lessons or longer packages.

If you list on Wyzant. You list under ESL/ESOL (Wyzant has no Business English category). The ESL/ESOL pool is 2,781 tutors — a meaningful 48% of Wyzant's English-tutor base. To differentiate yourself for a business buyer specifically, mention "business english" in your profile text: 288 of the 2,781 ESL/ESOL tutors do, and they show up in keyword search for that phrase. The rate ceiling on the consumer-direct surface is the highest of the three platforms. Top-tier tutors with Ivy or Big Tech credentials clear $90–$125/hour. The ESL/ESOL pool is structurally undersupplied against the US LEP-professional buyer base (~5M, see chart 08); rates aren't a niche-demand price, they're a market-clearing price for a tiny supply. The buyer pool behaves like Wyzant's TOEFL buyer pool: pays US private-tutor rates, rewards specialization, expects fast response time and US time zones.

Pool counts captured 2026-05-15. Direct visual confirmation from each platform's authenticated filter page on May 15, 2026.

Total English-tutor pool sizes captured same date: Preply 42,559; italki 4,287; Wyzant 5,769.

Because the three platforms decompose Business English differently, the three counts are not strictly equivalent. We use each platform's most-credible structured equivalent for the comparison and disclose the difference. The cross-platform share-of-pool comparison (38.7% / 94.1% / 48.2%) is the most defensible apples-to-apples reading available given the taxonomy differences.

Two earlier methodology errors corrected before publication. (1) An internal draft stated italki Business English listings as 479 rather than 4,034 — that number traced to a third-party blog summary, not a direct fetch, and was wrong by an order of magnitude. (2) An internal draft used the 288 Wyzant keyword-search count as the Wyzant Business English supply — that number measures a narrow text intersection inside ESL/ESOL, not the structural equivalent of Business English on Wyzant. Both errors are the kind Pitfall #1 (phantom numbers) exists to prevent; the methodology lessons are (a) pool counts must trace to direct platform fetches, not search-engine snippets, and (b) platform taxonomies are not interchangeable — each platform's most-equivalent structured category must be identified before cross-platform comparison.

Three specialty-signaling architectures. The most-equivalent-category audit revealed that the platforms structure tutor-side specialty signaling in three different ways, and the differences matter for interpreting the supply numbers. Preply uses a single-layer system of approximately fifty fine-grained English specialty checkboxes (Business English, Conversational English, IELTS, TOEFL, BEC, English for Job Interviews, Business & Work, ESL, ESOL, and roughly 40 others), each with a required description-blurb field. A tutor opts into a specialty by checking the box and writing a paragraph about their experience in it — opt-in cost is small but non-zero. italki uses a two-layer system: a single primary Specialties tag chosen at profile setup (high friction, one per tutor), plus a per-lesson configuration that allows tutors to set up multiple lesson types under categories like Business, Conversation, and Language Essentials (near-zero friction). The 38.6% italki primary-Specialty share is comparable to Preply's 38.7% checkbox share; the 94.1% italki per-lesson-config share has no Preply equivalent because Preply's specialty system is single-layer. Wyzant uses subject categories: English, ESL/ESOL, TOEFL, IELTS, and Business are each separate top-level subjects in the Wyzant taxonomy, and Business English is not one of them. The closest structural equivalent on Wyzant is ESL/ESOL (48.2% of the English pool). All cross-platform Business English share comparisons in this issue use each platform's structured-specialty signal (Preply checkbox / italki primary Specialty / Wyzant ESL/ESOL), not the broader opt-in layer where applicable.

italki per-lesson rate architecture. italki tutors set per-lesson-type prices. A tutor's profile displays separate prices for Trial Lesson, Business English, Conversational English, Interview Preparation, and other configured lesson types — each with its own rate. The price visible on italki search-card listings is the trial-lesson rate, which tutors discount to attract first-time bookings. The transacted Business English session rate is set separately and visible only on the tutor's profile detail page. From two representative italki Business English specialists captured this session: Jason Michel charges €17.20 trial and €43.01 per 60-minute Business English lesson; Paula Kamarados charges €21.50 trial and €24.08–€34.40 across her Business English lesson types. Any italki rate cited here or in prior Chalk Index issues that derives from a search-card display should be read as a trial-rate-anchored measurement, not a transacted session rate. Relative-rate-gap findings between italki and other platforms hold directionally; the absolute italki numbers underestimate transacted rates. A proper per-lesson rate scrape across the italki profile pool is on the work list for a future issue.

Rate disclosures captured from public marketing surfaces. Preply's "$3 – $40+ average $17" claim comes directly from the Business English landing page. italki's $9–$25 trial-rate range is from the filter-page card display; standard hourly rates per tutor are higher and require a profile-page click. Wyzant's "$35 – $60" range is from the Business English subject page's pricing-callout block.

B2B product feature comparison from public surfaces. Preply Business landing page (/en/business-language-training, /en/corporate-english-training), italki Business landing page (/en/business), and Wyzant's lack of a B2B product. The "paid search bidding" row was verified by observing a Preply Google Ads sponsored result on the "business english" keyword (Google Ads keyword identifier kwd-77735651).

Google Trends curves. The asymmetry chart approximates the Web Search and YouTube Search trend curves from screenshots captured directly from Google Trends Worldwide for the search term "business english" (Web Search) and the same exact phrase on YouTube Search, both over the past five years (2021-05 to 2026-05). The curves in the chart are smoothed approximations of the actual Trends output; the qualitative pattern — Web Search inflecting late 2025, YouTube flat throughout — matches the source screenshots.

Microsoft AI Diffusion data. Country-level AI adoption percentages are quoted directly from the Microsoft AI Economy Institute's Global AI Adoption in 2025 — A Widening Digital Divide (H2 2025 AI Diffusion report, published January 2026). Microsoft's methodology measures the share of each country's working-age population (15–64) that has used a generative AI product during the reporting period, derived from aggregated and anonymized Microsoft telemetry adjusted for OS and device share, internet penetration, and country population. The cross-walk to Chalk Index marketplace buyer geographies uses a Similarweb traffic-geography pull from April 2026, captured 2026-05-14. The country-pool color-coding in chart 07 is editorial: countries are assigned to the platform whose buyer pool they most contribute to, based on the Similarweb top-5 traffic shares per platform. We have not directly measured AI substitution effects on tutor bookings; the substitution hypothesis is testable in future Chalk Index scans, not proven in this one.

US labor force and language data. Foreign-born workforce share (32.3 million / 19.2%) is from US Bureau of Labor Statistics published figures on the foreign-born workforce. The 1-in-5 households-speaking-non-English-at-home figure is from the 2022 American Community Survey via the US Census Bureau. The 47% LEP estimate for the foreign-born US workforce is the Brookings Institution's analysis of ACS data; we cite Brookings as the secondary source because their analysis aggregates across metropolitan areas and is the most-cited single estimate. Self-employment language data is from the SBA Office of Advocacy's November 2024 paper "Lost in Translation: The Effects of Language on Business Ownership and Outreach," which works directly from the 2022 ACS microdata. The "professional/management slice" estimate (~11 million foreign-born US professionals) is derived from Brookings's reported one-third figure applied to the BLS 32.3 million; treat as a rounded estimate. The "5 million LEP professional" funnel midpoint is a simplifying assumption — the actual count requires intersecting LEP status with occupational classification at the ACS microdata level, which we have not done for this issue.

Tag-inflation caveat. Multi-specialty box-checking is endemic to specialty tagging on Preply (10.9% of the Preply TOEFL pool tags all four of IELTS, TOEFL, TOEIC, and PTE simultaneously — four tests that serve materially different populations). Business English is the most heavily tag-inflated specialty we have measured on the platforms that have it as a category: italki's 94.1% near-universal tagging is the extreme case. The 16,481 Preply count includes tutors who tag Business English alongside many other specialties and who do not credibly specialize in B2B teaching. The credible-specialist count is materially smaller. Wyzant, which has no Business English category, sidesteps this problem at the taxonomy level — its ESL/ESOL pool is structured by what tutors actually teach rather than what they self-tag, so the 2,781 figure is a tighter measurement of relevant supply than either of the other two. We will measure the exclusive-Business-English tutor count on Preply and italki, and the Wyzant ESL/ESOL rate distribution, at scrape time.

Marketplace scope. This issue measures only the marketplace channel. Corporate training delivered through Berlitz, EF Corporate, GoFluent, Voxy, in-country corporate training providers, and direct enterprise contracts is outside our scrape. The corporate-training layer is the next frontier for Chalk Index to measure; the methodology will require a different data pipeline (sales-page pricing capture, RFP response analysis, LinkedIn job postings for corporate trainers).

Listed rate as cleared rate. Listed rates approximate but do not equal transacted rates. Preply's 100% trial-lesson tax means new-student transactions clear materially below the listed rate; italki's lesson-type commission schedule (0–30%) means the listed-vs-transacted gap depends on package size; Wyzant's buyer-side service fee varies by tutor rate. Relative platform comparisons hold; absolute rates require adjustment.

Single-point measurement. This is a snapshot. We cannot make trend claims about Business English supply growth without multiple measurements over time. The Google Trends demand-side data is time-series; the marketplace supply-side data is not, yet.

Issue 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?

Issue 5: The corporate training layer that doesn't appear on the marketplaces. Berlitz, EF Corporate, GoFluent, Voxy. Public pricing pages, sales-deck pricing leaks, LinkedIn job postings for in-house corporate language trainers. The other 80–90% of Business English demand that this issue could not directly measure.

Data: Specialty filter pool counts captured 2026-05-15 from Preply, italki, and Wyzant filter URLs. Rate disclosures from each platform's public marketing copy, same date. B2B product feature comparison from each platform's public marketing surfaces. Google Trends Web Search and YouTube Search curves for the exact phrase "business english", worldwide, 2021-05 to 2026-05. No tutor-listing scrape yet for Business English; planned for Issue 3's follow-on data refresh once the marketplace scrape methodology has been extended to handle the larger pool.