Page language: English
How we pick tutors
We read the public profiles of thousands of tutors on 10 marketplaces every day and turn them into plain facts you can compare. Here is exactly how, and where the limits are.
How we choose top picks
Every language page starts with our top picks. Here is how a tutor gets there:
They need at least 20 reviews and a rating of 4.7 or higher on their marketplace.
We rank by what students say (35%), teaching experience (30%), qualifications (15%), command of the language (10%) and how reliably they reply (10%). Ratings are compared with each marketplace's own average, because a 5.0 is common on some sites and rare on others.
No marketplace gets more than 3 of the top places, so you see the best of several sites, not just the biggest one.
Missing information counts as average, not as zero: a tutor isn't pushed down just because their platform doesn't publish reply times.
Independent: nobody pays to rank
Tutors and marketplaces can't pay to be listed, picked or ranked higher. “Book” links tell the marketplace the visit came from SpeakDuo and may carry a referral tag; that never changes the order.
Updated every day
Our crawler, SpeakDuoBot, re-reads public tutor listings every day and respects each site's robots.txt. Prices are converted to US dollars per 60-minute lesson with European Central Bank rates, and tutors who leave a platform drop out after the next full refresh.
What each fact on a tutor card means
Every card answers six questions learners ask before booking. Behind each answer is a score from 1 to 99 that we use for ranking.
Language level
How well the tutor speaks the language they teach.
Native speakers score 96 (90–95 when native is inferred), plus 2 with a teaching qualification. Other speakers score by CEFR level, highest when the marketplace verified it, lower when self-declared and lowest when inferred.
Experience
How much the tutor has taught, on the marketplace and elsewhere.
Grows with hours taught on the marketplace (or lessons × lesson length) plus the years of experience a tutor states elsewhere, on a logarithmic scale: about 44 for 100 hours, 66 for 1,000 and 88 for 10,000.
Qualifications
Formal teaching qualifications, such as CELTA or a teaching degree.
Uses the tutor's best teaching qualification: DELTA, official examiner, CELTA and teaching degrees rank highest, then TEFL/TESOL and language degrees. Qualifications verified by the marketplace score higher than self-declared ones; a platform's professional-teacher status alone counts as 70.
Can explain in
Other languages the tutor can explain things in.
Counts the other languages a tutor speaks, weighted by level: one extra native or C2 language scores about 65, two about 99. Matters most for beginners and newcomers who need explanations in their own language.
Student reviews
Student reviews, compared with the platform's average.
A tutor's rating smoothed by review count and compared with their own marketplace's average (a 5.0 means less on a platform where most tutors have 5.0). About 70 is platform average; many excellent reviews approach 99.
Replies
How quickly and reliably the tutor responds and shows up.
The average of whatever reliability signals the marketplace shows: response time and rate, cancellations and attendance, and open slots in the next week. Many platforms don't publish these, so this stat is often unknown.
“Not listed” means unknown, not bad
When a profile doesn't give us enough to go on, the card says “Not listed”. Unknown never counts as a low score: it just means the marketplace doesn't publish that information.
“Likely” means inferred
Some values are inferred rather than stated, for example a likely native language or dialect read from a profile. They're marked “likely” and count slightly less than stated or verified information.
Native speaker and other facts
Some things are facts, not scores: whether a tutor is a native speaker (shown as a flag on the card), their accent or dialect, the levels they teach, their price, time zone and specialties. We show these as they are.
Where the data comes from
All data comes from the public profiles tutors publish on marketplaces such as italki, Preply and Cambly. Part of it is extracted from profile text with AI (for example qualifications or dialects) and may be inferred, so treat the stats as a guide and check the profile before you book.
Corrections and removal
Tutors and platforms can ask us to correct or remove a profile at any time. Email support@speakduo.com with the profile link and what's wrong; we fix the data and it updates on the next refresh.
Questions about our method
How we collect this data
We collect public profile facts (name, photo, languages, price, rating) from tutoring marketplaces and refresh them daily. Prices are normalised to a 60-minute lesson in US dollars. Ratings and reviews belong to each marketplace. Subscription platforms have no per-tutor price.
Currency conversion: ECB reference rates.IP geolocation by DB-IPHow we pick tutorsAbout SpeakDuoBotRequest removal