TOEFL 2026 guide
How to use this site
Twelve games, timed practice tests and two dashboards. The order you use them in decides how much they are worth.
The five moves
Preparation usually fails in one of three predictable ways. Students practise the tasks they are already comfortable with, because those feel productive. They never work under time, so they never find out that a skill they have at leisure disappears at speed. And they measure themselves on material they have already seen, which tells them what they remember rather than what they can do.
This site is built to make all three harder to do by accident. Everything on it belongs to one of five moves, and the moves run in an order.
- 1UnderstandRead what a task is actually asking before you ever attempt it.
- 2LocateTwelve minutes, fifteen questions, ending with the one task to start on.
- 3TrainPlay the game built for that task until the pattern is automatic.
- 4MeasureTake a timed practice test, with no hints, on material you have never seen.
- 5ReadFind out which task actually moved, and what to fix next.
Then back to Train, with a specific target rather than a vague intention. One pass through this is not the method — the loop is.
Each move fails on its own
Guides without practice is knowing the theory of a language you cannot speak. Practice without tests is comfort. Tests without the record is a number you forget by Thursday. The value is in the cycle, not in any one part of it, which is why the navigation bar is ordered the way the method is.
Why the games and the tests never share material
This is the most important thing on the page, and it is the thing most preparation products get wrong. The passages, questions and audio used by the games are drawn from a completely different bank to the ones used by the timed papers. No item ever appears in both.
The reason is simple. If you study a passage, have your mistakes explained, and are then tested on that same passage, the result measures how well you remember the session you just had. It does not measure reading. It reliably flatters you, and it is the mechanism behind the most common disaster in test prep: strong practice scores followed by a real result twenty points lower.
Practice — the games
- Hints, retries, and an explanation after every mistake
- Difficulty adapts to you; it gets harder as you get better
- Untimed, or timed generously, so you can think
- Rewards effort — the point is to build the habit
- Tells you: how much you have trained
Measurement — timed practice tests
- No hints, no retries, one attempt per question
- Fixed difficulty; the paper does not care how you are doing
- Real section timing, with the clock visible
- Rewards nothing — it is not trying to encourage you
- Tells you: what that training is currently worth
A test that flatters you is worth nothing
Every result the timed papers give you is a band, never a single number, and it is labelled a practice estimate rather than a score. Twelve or twenty questions cannot produce a TOEFL score, and any product that hands you a confident single figure from that many items is guessing at you.
Where everything is
The site has four areas, and each one looks deliberately different so you always know which you are standing in. Bright and editorial means you are reading. Dark and coloured means you are training. Plain and institutional means you are being measured.
Reading the Games tab
The Games tab answers one question: how much have you trained? Every figure on it comes from the games themselves, and none of it is a prediction of your TOEFL result.
- Earthianthe big dial
- The equal-weight average of your mastery across all eleven games. It moves when any game moves, which makes it a fair summary of total training.Not a score, a percentile, or a prediction. A high Earthian with no papers sat means you have practised a lot and measured nothing.
- Masterythe per-game percentage
- How reliably you are getting that game's items right at your current level. It rises as you improve and falls if you stop.Not how much of the game you have finished. That is the next number.
- Sets donebeside each game
- How much of the game's content you have actually worked through — coverage, not quality.Not interchangeable with mastery. Eighty per cent on two of eight sets is not eighty per cent done, and reading only the first number is how students stop far too early.
- Train this nextunder the dial
- Your weakest game by mastery — the cheapest place to gain.Not an instruction to ignore everything else, and not evidence from a timed paper. For that, use the Tests tab.
Reading the Tests tab
The Tests tab answers the harder question: what is that training worth on material you have never seen, against the clock? It is deliberately kept apart from the Games tab, because a strong game record beside a weak paper is the single most useful thing this site can tell you — and averaging the two would hide it.
- Latest estimatethe band, out of 30
- The range your performance on that paper is consistent with. A band, because a short section genuinely cannot narrow it further.Not an official TOEFL score, and not a promise. It is a practice estimate from one sitting.
- Changethe signed number
- The difference against your previous attempt at that section, in percentage points.Not meaningful under ten points. Anything smaller is shown in grey because on sections this short it is noise, not progress.
- Trendthe small line
- Every attempt at that section in order, drawn against a fixed nought-to-hundred scale with a dashed line at fifty.Not scaled to your own range. Fitting the line to your data would turn a three-point wobble into a dramatic climb.
- Thin evidencethe tag beside a task
- You have met fewer than four questions of that task, so the percentage beside it is close to meaningless.Not hidden from you. The figure is still shown; it is simply excluded from what you are told to fix.
- What to fix nextthe three cards
- Your weakest tasks that have enough evidence behind them to act on, each linked to the game that trains it.Not simply your three lowest percentages. A task you have missed twice out of four does not qualify.
- What keeps catching youthe tags at the bottom
- The named wrong answers you chose most often — surface matches, over-broad options, repeated words.Not gaps in your English. These are habits of reading, and they are the fastest thing on the page to fix.
A week that works
If you have an exam date and limited evenings, this is a defensible shape for a week. It assumes about forty minutes a day, and it deliberately puts measurement at the end rather than the beginning.
| When | What | Why |
|---|---|---|
| Day 1 | Read the guide for your weakest section | Most lost marks are misread tasks, not weak English |
| Days 2–4 | One game, twenty minutes, same task each day | A pattern becomes automatic through repetition, not variety |
| Day 5 | A second game from a different section | Keeps the other three sections from decaying while you focus |
| Day 6 | Take one timed practice test | The only honest check, and it must be on unseen material |
| Day 7 | Read the Tests tab and pick next week's target | Evidence chooses the target, not preference |
Take practice tests sparingly
There are a limited number of forms per section, and each one is only unseen once. Sitting them back to back to chase a better number burns the very thing that makes them worth sitting. One a week is plenty.
What the game layer is actually for
The games are not decoration over a question bank, and they are not there to make a dull thing briefly amusing. Each one exists because a specific TOEFL task has a specific failure mode that ordinary practice does not fix.
- Speed. Echo Drop and Signal Drop run against a clock because Speaking Task 1 fails on hesitation, not on grammar. You cannot drill fluency slowly.
- Failure before explanation. Arrivals and Set Text let you get it wrong first, because a trap you have personally fallen into is remembered and one you were warned about is not.
- Range. Open Thread tracks which phrasings you have used and pushes you off the one you keep reaching for, because fossilised phrasing caps a Writing mark no matter how correct it is.
- Repetition without boredom. Second Nature and Open Channel need volume to work, and levels, streaks and rising difficulty are what make volume survivable.
So the game layer is a delivery mechanism for the kind of practice that works but that nobody does voluntarily. Used well, it gets you to do four hundred items instead of forty. Used badly, it becomes the point.
How to tell which is happening
If your Earthian is climbing and your test estimates are not moving, you are getting better at the games rather than at the test. That is exactly the failure the two-tab dashboard exists to make visible. When you see it, stop training and sit a paper, then work on what the breakdown names rather than on what feels productive.