Skip to main content
20% off your first payment$5.99 for your first week, then $7.49 · Ends November 15, 2026See plans

For schools, academies and university programmes

The full evaluation, on one page

Written by the Reach120 research teamReviewed

Reach120 is TOEFL practice software a teacher can put in front of a class today. It is not an institutional platform, and this is written so you can work out which one you need before anyone books a call.

What follows is the build as it stands: what a teacher can do this afternoon, and the list of things institutions routinely require that this product does not have. The second list is longer than the first.

The shape of the workspace your teachers would use

Diagram

This is a diagram of the screens, not a screenshot of them. Every label in it is the wording the shipped screen uses — a test reads the screen off disk and fails if one is renamed — and every value is left blank.

The cohort view a teacher opens

One read across every class that teacher runs, ordered by who needs them first rather than alphabetically.

All classes

Overview

StudentStatusLatestActive
  • At risk
  • Watch
  • On track

This week

  • scored responses
  • active learners
  • mocks completed
  • assignments overdue

The columns carrying a score are left out of this diagram, not out of the screen. The teacher who runs the class reads them behind a learner’s row — the panel below is where that learner agreed to it — and no surface above that teacher shows a score at all.

Setting a piece of work

The teacher picks the task and the question, writes the instructions in their own words, and chooses who receives it.

  1. Choose one or more tasks
  2. Pick a question
  3. Instructions (optional)
  4. Assign to
  5. Publishing

Whole classSelected studentsDue date

The due date is optional and the recipients are chosen per piece of work, so this fits a syllabus you already run rather than replacing it with ours.

What a learner reads before joining

The same screen on every route in — an invite link, a class code typed on their own dashboard, or that code scanned from a projected slide. Accepting it is the only way through.

Your tutor will see

  • Scores and rubric feedback on practice you complete
  • Submission history and PDF feedback reports
  • Assignment progress and activity
  • The email address on your account, which labels you on their class roster

Your tutor will never see

  • Your password, email inbox, or login details
  • Billing, subscription, or payment information
  • Your chat history with the Reach120 assistant

DeclineAccept and join

Neither list is written for this page. Both are the arrays the consent screen itself renders, imported from the module it reads them from, so changing what a teacher can see rewrites this diagram in the same commit.

Who can submit, and who is blocked

Each learner holds their own access, and the teacher is told which of their learners can submit scored work and which cannot.

  • Can submit AI-scored work
  • Blocked — no paid access

It is shown to the teacher because the alternative is worse: a cohort read taken over only the learners who could afford to submit is not a cohort read, and nothing on the screen would have said so.

Nothing here is anyone’s data. No school, class or person is depicted, real or invented, and no figure is shown, because none has been measured and an invented one would be a claim about a customer we do not have.Described in words: the cohort view is a wide table with a right-hand column of week totals, and a filter above it for a single class or all of them. Each row stands for one learner and carries, in order, who they are, a state of At risk, Watch or On track, the latest thing that happened, and when. Setting a piece of work runs as an ordered strip of steps from choosing the task to publishing it, with the recipients and the due date decided at the end. The joining screen is a pair of facing lists — what the teacher will see, and what they will never see — above a decline and an accept. The access panel is a row of labelled slots showing who can submit scored work and who is blocked. Every slot that would hold a figure is drawn as an empty bar.The diagram has no screen above a teacher, because the product has none: what an administrator over several teachers would get, and what it does not, is the list below.

What a teacher can do today

Classes a teacher creates without waiting on us

A teacher makes a class in the workspace and students join it themselves — by a private invite link, by a class code they type on their dashboard, or by scanning the code from a projected slide. There is no roster to send us and nothing for your IT team to install.

Assignments that follow your syllabus, not ours

The teacher picks the task type, writes the instructions and sets the due date. There is no fixed course a cohort has to be marched through, so this fits alongside a programme you already run rather than replacing it.

The response behind every score

A teacher opens any submission and reads the response itself, the rubric reasoning, the line-level corrections, and what the feedback does not cover. A moderator checking a colleague’s judgement gets the same view.

A per-student record a teacher can open mid-term

Assigned, completed and outstanding work sits in one class queue, with each student’s history behind their name. It answers the question a programme lead asks in week six, which is who has stopped working.

One read across every class a teacher runs

Above the individual classes sits a cohort view: where the group sits on the practice rubric, which rubric dimension is weakest across all of them, how much assigned work has been completed, and who has stopped submitting. It answers the planning question — what should the next lesson be about — rather than the pastoral one.

The person who pays sees unseated learners across every teacher

Seat candidates are drawn from every class run by anyone staffed to the account, not only the classes the buyer teaches themselves — so a colleague’s student can be handed a seat without asking the colleague to do it. It stops at that one action: it does not open a colleague’s roster, reassign their class, or put one teacher’s account above another’s.

A consent screen the student reads before anything is shared

Whatever route a student arrives by, they see the same screen naming exactly what their teacher will and will not see, and accepting it is the only way into the class. They can leave from their own settings, and the teacher can remove them.

Practice that matches the 2026 test rather than the old one

Every task type across the reading, listening, speaking and writing sections follows the current format, and the map behind it links to the ETS specification with the date it was read. Retired formats are labelled as retired instead of quietly left in.

A feedback report that leaves the platform

Any scored response exports as a PDF the student keeps and the teacher can attach to a tutorial record. Nothing a learner has produced is locked to an account that stops being paid for.

What a student in your programme does

Your teachers work in the workspace; your learners work in the same product any individual buyer uses. Nothing about a class changes what a student sees when they sit down to practice, which is deliberate — a cohort build that diverges from the main product is a cohort build that rots.

  1. They join the class themselves

    A learner follows the invite link, types the class code on their dashboard, or scans it from a projected slide. Before they are in, they read a screen naming exactly what their teacher will and will not be able to see, and accepting it is the only way through. Nobody is enrolled on their behalf.

  2. They practice every section of the current test

    Reading, Listening, Speaking and Writing, in the task formats the test uses since the 2026 change. A teacher can assign specific work with a due date, or leave a learner to work through the bank on their own — there is no fixed course either of them has to be marched through.

  3. Written responses come back scored, with the reasoning attached

    An email or a discussion response is scored against the rubric criteria, with the rubric reasoning, line-level corrections, and an explicit note of what the feedback does not cover. The learner sees the same evidence the teacher does — this is not a report written about them behind their back.

  4. Spoken responses are scored on intelligibility, fluency and language use

    A recorded answer is scored on how intelligible it is, how fluently it runs, and the grammar, vocabulary and accuracy of what was said. ETS scores Delivery by the effect pronunciation, intonation and pacing have on overall intelligibility, and intelligibility is what is measured here. What this does NOT do is tell a student which individual sound to correct — that coaching does not exist in the product, and a teacher remains the only source of it.

  5. They can sit a full mock under test conditions

    A combined mock runs the sections end to end with the timing the specification lists, and produces one report rather than four. It is practice software: nothing here supervises a test-taker or verifies who is typing.

  6. Everything is exportable, so it survives leaving

    Any feedback report downloads as a PDF the learner owns. A student who finishes your programme, or leaves it, keeps their record without needing an account here or a teacher to send it to them.

How your learners get access

There is no seat pool and no institutional licence to buy. What exists is simpler, and worth understanding before you plan a cohort around it.

The Educator Workspace costs a teacher nothing

Creating classes, inviting learners, setting assignments, reading the evidence behind every score and looking across your classes are not billed. Your teachers do not need a plan to run this.

Each learner holds their own access

A student practices on their own account, on the same self-serve plans an individual buyer uses, paid by card. Access runs for a dated window and is theirs — it is not assigned by you, and it does not end because a class did.

There is no way to buy access centrally yet

No seat pool, no bulk purchase, no invoice, no purchase order. If your finance process cannot put a card in front of each learner, that is a genuine blocker today and it is better raised in the first message than the last.

A pilot is arranged directly, and deliberately small

For a first cohort we set the arrangement up by hand and talk to you while it runs. That is not a scale story and it is not pretending to be one — it is what a product at this stage can honestly support, and it means the people who built it are the people answering when something is wrong.

What this product does not have

These are the requirements institutions bring, and the status of each one in the build as it stands. None of them is on a promised date, because none has been scheduled.

Single sign-on with our identity provider

Not built

Students and teachers sign in with their own email address or a Google account. There is no SAML or OIDC connection to a university directory, so accounts are not provisioned or deprovisioned by your systems.

An LMS integration (Moodle, Canvas, Blackboard)

Not built

Nothing here is an LTI tool and no grade passes back to a gradebook. A teacher shares the class code from wherever they already talk to the cohort, and carries results across by hand or by the exported report.

Bulk enrolment from a student information system

Not built

There is no roster upload and no CSV import. Every student joins by their own action, which is also what makes the consent screen meaningful — an account created for somebody has not agreed to anything.

An administrator account over several teachers

Not built

The person who bought seats can see, and hand out, unseated access across every teacher on the account — that reach exists for seat assignment only. There is still no account that reassigns a class, edits a colleague’s roster, or holds authority over another teacher’s classroom; each class stays the teacher’s own to run.

An invoice, a purchase order, or seats bought centrally

Individual plans only

Access is sold as the same self-serve plans an individual learner buys, paid by card. There is no seat pool, no site licence, no invoicing and no procurement paperwork. If that is a blocker, say so in the first message rather than the last.

A signed DPA, or a compliance certification

Not built

No certification is claimed — not FERPA, not GDPR certification, not COPPA, and no SOC audit report. What is actually true about data handling is written in the privacy policy, and a data-protection question goes to the privacy mailbox where it is answered rather than badged.

Supervised or proctored testing for placement

Not built

This is practice software. Nothing here supervises a test-taker, verifies who is typing, or produces a result an admissions office should treat as an assessment.

The 2026 test, in the published wording

Each statement below is quoted from the claims record behind this site, which names the source it was read from. Nothing here is our summary of the specification.

The scale your paperwork refers to

The reporting change is the part that reaches an institution first, because it lands in documents written long before anybody looked at a practice tool.

For TOEFL tests taken on or after January 21, 2026, official score reports use a 1–6 scale in half-band increments.

A requirement your programme publishes in the old numbers now describes a report your applicants no longer receive.

Claims record E-001 · source

During ETS's transition through January 2028, reports also include a comparable overall estimate on the previous 0–120 scale.

For the length of the transition an applicant may hand you a report carrying both, so a reader who knows only one of them can still misread it.

Claims record E-002

Before January 21, 2026, TOEFL iBT reported scores as ‘Reading (0-30) Listening (0-30) Speaking (0-30) Writing (0-30) Total (0-120).’

The wording most existing admissions pages and course handbooks still use, kept here so you can find the sentences that need rewriting.

Claims record E-023 · source

What the test contains

Enough of the specification to plan a course against, rather than to plan a course against a memory of the previous one.

The current Writing section lists 12 items: 10 Build a Sentence items, one Write an Email task, and one Academic Discussion task.

The count matters for pacing: most of the writing section is the short task teachers tend to spend the least class time on.

Claims record E-003 · source

The 2026 specification lists the Writing section at 12 items and 20 maximum raw points.

What a whole writing section is worth in raw points, which is the figure to teach against rather than a converted band.

Claims record E-015

Per the 2026 specification, the maximum points per Build a Sentence item is 1.

Each of those items is worth the same whether it is easy or hard for the student in front of you.

Claims record E-016

Build a Sentence represents 10 of the 20 listed maximum raw points in the Writing specification.

Half the writing section sits in the task type that is quickest to drill, which is where a short course buys back the most.

Claims record E-004

The 2026 Reading and Listening sections use multistage adaptive designs; Speaking and Writing are linear.

Reading and listening adapt as the student works, so students sitting side by side can be answering different item sets, and the habit of comparing raw answers afterwards stops working.

Claims record E-005 · source

The 2026 specification lists the Reading section at 50 items and 35 raw points, with task types Complete the Words, Read in Daily Life, and Read an Academic Passage.

What a reading section costs in class time, and the task names to set homework by.

Claims record E-009 · source

The 2026 specification lists the Listening section at 47 items and 35 raw points, with task types Listen and Choose a Response, Listen to a Conversation, Listen to an Announcement, and Listen to an Academic Talk.

The same for listening, where the task names changed more than the skills did.

Claims record E-010

The 2026 specification lists the Speaking section at 11 items and 55 raw points, with task types Listen and Repeat (7 items) and Take an Interview (4 items).

The same for speaking, which is the section most prep courses under-serve.

Claims record E-011

The 2026 specification lists 120 total items, 145 total raw points, and an estimated total test time between 1 hour 23 minutes and 1 hour 29 minutes.

The whole test is shorter than the timetable slot many programmes still reserve for it.

Claims record E-012

The 2026 specification lists estimated section times of 18–21 min router plus a 9 min module (Reading), an 18 min router plus a 7 or 11 min module (Listening), 23 min (Writing), and 8 min (Speaking), in a two-stage router-plus-module adaptive design for Reading and Listening.

Section timings, for anyone building a mock schedule that matches the current test rather than the retired one.

Claims record E-014

How responses are marked

The distinction to teach before a student argues with any practice result, including one produced here.

Reading and Listening items are machine-scored selected responses; the Write an Email, Write for an Academic Discussion, and all Speaking tasks are AI-scored constructed responses, per the specification's item-type notes.

Which parts of the test are marked against a key and which are marked against a rubric. A class that treats the second kind as arithmetic will argue with every mark it gets.

Claims record E-013

ETS states that its mock-test scores come from a system it calls TOEFL AI Lite — ‘built for speed and immediate feedback, helping you identify strengths and weaknesses in real-time’ — and that ‘when you take the real test, it is scored by a different system called TOEFL AI Deep’, which ‘takes a more comprehensive approach’ and ‘integrates with advanced security protocols with a human in the loop, to ensure your official score is accurate, fair, and recognized by institutions worldwide.’ ETS also states: ‘You might see some small differences between your mock test scores and your real test scores, and that's totally normal.’

Worth reading in full before any practice tool, this one included, is treated by a programme as a predictor of a real result.

Claims record E-030 · source

What this page does not claim

  • No customer list, and no logos. Naming a school implies a reference we have not asked permission for, and a logo wall on a page like this is usually a list of trials.
  • No write-up of a school that piloted this, and no count of institutions. Neither has been run and measured, so there is nothing to report — a page describing one would be describing something that did not happen.
  • No quoted teacher, programme lead or student, and no star badge. Nothing published here is a person’s words unless that person agreed to it for this surface, and nobody has.
  • No score-gain figure and no improvement claim, for your students or anyone else’s. None has been evidenced, so none appears anywhere on this site.
  • No comparison to an official rater. Reach120 produces a practice estimate. It is not examiner-level, ETS-trained or ETS-calibrated, and how the scoring works is written up rather than asserted.
  • No compliance badge. Certifications are listed above as things this product does not have, which is the honest place for them.

Questions programmes ask first

NoCan we buy one monthly plan covering a small group of students?
No. There is no group plan, no seat pool and no central purchase — each learner holds their own access, bought on their own card. This is the most common reason an institutional enquiry stops here, so it is worth confirming against your finance process before anything else.
YesDoes it evaluate speaking and writing automatically?
Yes. Written responses are scored against the rubric criteria with the reasoning and line-level corrections attached; spoken responses are scored on intelligibility, fluency, grammar, vocabulary and accuracy. Reach120 produces a practice estimate. It is not examiner-level, ETS-trained or ETS-calibrated, and it does not tell a student which individual sound to correct.
YesCan students sit full mock exams?
Yes. A combined mock runs the sections end to end with the timing the specification lists and produces a single report. It is practice software: nothing supervises the test-taker or verifies who is typing, so a result here is not an assessment an admissions office should act on.
YesIs there a dashboard for the teacher?
Yes. A teacher sees their class roster, what is assigned and outstanding, each learner’s history, and the full evidence behind any score. Above that sits a view across every class they run: where the group sits on the practice rubric, which rubric dimension is weakest, and who has stopped submitting.
NoCan a head of department see across several teachers?
No. There is one teacher role and it sees the classes that teacher runs. There is no administrator account above them, no way to reassign a class, and no cross-staff reporting.
What happens to a student’s work when they finish or leave?
It stays theirs. A learner can download any feedback report as a PDF and can leave a class from their own settings at any time, which immediately ends their teacher’s visibility. Their account and their practice history are not held by your institution and do not end with the course.

Rule us in or out this week

Reach120 is one person, so support is not outsourced to anyone else. We have not answered enough messages to measure a reply time, so we do not state one. Write to partnerships@reach120.com or use the contact form, which has a field for your organisation.

Reach120 is an independent practice tool. It is not affiliated with, endorsed by, or approved by ETS, and it does not provide official TOEFL® test scores.

TOEFL® is a registered trademark of ETS. This product is not endorsed or approved by ETS.