Do Online English Lessons Work? | Hours and Plateaus

You take a lesson every day. You have kept it up for six months, maybe a year. And still, what you are able to say does not feel different from last year. This article is written for people in that state.

The conclusion first. There is a published figure for how long it takes, and it comes from an examining body rather than from a school or someone’s personal account. Measured against it, most people are still short of the hours. But piling up hours does not always get you through, because there is a stretch where adding time stops working. This article covers those two things in that order.

This article will not tell you to take more lessons, or to switch to a different service. It is only about what to change before and after the lessons you already take.

Key takeaways

• Moving from CEFR B1 to B2 works out to 100–250 hours from the published table. In 25-minute lessons that is 240–600 lessons, which is one to two years even if you take one every day.
• Those hours count "lessons and supervised study". Self-study outside lessons, and time spent around English, are not in them. Counting things that are not included makes you read your own position as further along than it is.
• You can stall even when the hours are there. A wrong form can become the one that comes out first, and once it settles, simply speaking more does not undo it.
• Whether you are stalled is not something the feeling of progress tells you. "How many times has this same mistake come up" tells you. If you are not counting, you have nothing to judge with.
• Three things get you out: noticing what you could not say, asking your teacher to correct you differently, and deliberately using something slightly above what you can already do.
• Before adding lessons, change what happens around them. For the same hours, recall-shaped review with spacing leaves more behind.

There is a published figure for how long it takes

"How long until it works" is the question people searching this most often ask. The answers differ from article to article: two weeks, one month, three months, three to six months, a year. Most of them cite nothing.

A public figure does exist. Cambridge English, which runs English qualifications, publishes the study hours it takes to reach each CEFR level. The numbers are cumulative from a complete beginner.

The page states plainly that "these figures are intended as a guideline only". It also says the time you need changes with your learning background, the intensity of your study, your age, and how much you are exposed to English outside lesson times. So the table is a ruler for checking whether you are moving toward it, not a quota to hit.

What most people taking online lessons care about is the stretch from B1 to B2. From the table, B1 sits at 350–400 hours and B2 at 500–600, so the gap is 100–250 hours. Converted into 25-minute lessons, that is 240–600 lessons. ★That conversion is ours, not something Cambridge says.

One lesson a day puts 240 lessons at about eight months and 600 at about one year and eight months. Three times a week takes more than twice as long. "Six months and nothing has changed" is, against this figure, still partway.

Three times a week is more realistic for many people. At 25 minutes, three times a week, a full year comes to 65 hours. With B1 to B2 at 100–250 hours, a year does not even reach halfway. Concluding "maybe this is not for me" at that point is early.

Does that make the articles saying "three months" wrong? No, they are pointing at something else. What changes in three months is how used to listening you are, how much resistance you feel before speaking, how fast you react to a teacher’s question. Those genuinely move early. Moving up a level means the range of sentence patterns you can use has grown and your accuracy has come with it. The first is weeks to months; the second is hundreds of hours. Both live inside the same word.

Separating the two makes your own position easier to read. Feeling less resistance while feeling no real progress is not a contradiction. You have reached the first and are still partway through the second.

That is the time side of it. But hours alone do not guarantee the climb. There are two conditions under which the table stops applying, and that is the next section.

What those hours actually count (get this wrong and you are out by double)

The figure is published as guided learning hours. The official explanation defines it as lessons and supervised study. Separately from that, it lists "the amount of study and exposure outside of lesson times" as one of the things that changes how long you need.

So the English you listen to outside lessons, the podcast on your commute, the drama you watch, are not inside the hours in that table. They are treated as a separate factor. Mixing them in makes you read your own position as further along than it is.

Two things break this reading. One is that you may not be at B1 in the first place; the table describes moves between levels, so a different starting point changes the hours. The other is that, depending on how you take them, the 25 minutes may not all be supervised study. A session spent on how your week went, one where the teacher does most of the talking, one you got through using only phrases you already had — the time accumulates, but the substance does not.

This is what "feeling like you have taken a lesson" really is. The number of lessons stays on your calendar and is easy to count; progress is hard to count. Look only at what is easy to count and the sense of moving grows first.

When the hours are there and nothing changes, what is happening

It does happen that you pass the figure and still feel no different. Second language research has long described a learner’s language stopping its development and settling with wrong forms in place; it is called fossilization. The paper that first used the term wrote that only a small share of learners reach the level of a native speaker and most stop somewhere.

What "most stop" points at is whether you reach the level of a native speaker. It is not a figure about which CEFR level you stop at. That said, the intermediate range is a common place to stop, for a simple reason: at intermediate, wrong still gets understood.

A missing article, a dropped third-person s, a tense that slides — the other person still understands you. Because you are understood, no need to change how you say it ever appears. And after a few hundred repetitions, that form is the one that reaches your mouth first.

As skill learning this is a normal process. Moving from knowing something as knowledge to using it without thinking is called automatization, and speaking fluently requires it. The problem is that automatization happens to wrong forms exactly as readily as to right ones. Once a wrong form is automatic, it is hard to correct later.

And a form that has settled does not come loose just from hearing the correct version. A meta-analysis comparing types of correction in the classroom found larger effects for asking the learner to say it again, or explaining in words why it is different, than for a teacher quietly recasting it into the right form.

This is why more lessons do not change things. Most teachers recast quietly so as not to break the flow of the conversation. As conversation it is natural, but it is the weakest way to strip off a settled form. Doubling the number of lessons only doubles the weakest way.

There is a second reason progress gets hard to see. Speaking is usually looked at not as one line but as three: fluency, accuracy and complexity. They do not grow at the same rate. Keep taking lessons and fluency moves first — you stall less often, your silences get shorter. Accuracy and complexity, meanwhile, stop easily unless you work on them deliberately.

That asymmetry is what makes the experience confusing. While fluency is rising you feel you are getting able to speak; the moment it plateaus it looks like you suddenly stopped improving. Nothing stopped suddenly. The one that had been rising hit its ceiling first, and the two that were never moving came into view.

So it is worth looking at which of the three is moving and which is not. If you stall less often, fluency is moving. On top of that, if the number of corrections has not fallen, accuracy is stuck. If the sentence patterns you use are the same as six months ago, complexity is stuck. Do not try to raise all three at once; pick the one that is stuck.

Check whether you are in that state with records, not with how it feels

Judging by the feeling of progress is hard, and that is something research says too. In a learning experiment, learners’ own predictions of how much they had retained barely corresponded to their actual results.

What you can use instead is records. The items below are written so you can check them against your own lesson records. This is not a list saying that people it applies to are taking lessons badly. Use it to decide which of the steps in the next section to start from.

• In your last five lessons, did you write down even one thing your teacher corrected? If not, there is no way to know how many times the same mistake has come up.
• Are you counting how many times the same mistake has been pointed out? Up to the second time it is chance; three in a row is a habit.
• In the last five lessons, how many expressions did you use for the first time? If it is zero, you are running the conversation on what you already had.
• Do you remember where you went silent during a lesson? What you could not say leaves little trace, so it disappears unless you pick it up on purpose.
• Does your teacher ever push you with "tell me more" or "say that another way"? If not, your turns look complete from the outside, so nothing is pushing you upward.
• Do you say the corrected version out loud yourself within the same lesson? If you only hear it, the corrected form has not entered what you actually use.

If you cannot answer the first two, start by keeping records. What to write is settled, and three things are enough.

There is one detail in how you keep them that makes the whole thing pointless if you get it wrong: do not break the record up one page per lesson. Split by date and the same mistake scatters across pages, and you lose the count. Stack the same item in one place, and when it comes up a second time, add the date to the existing line instead of writing a new one.

Do that and within about three months your habits narrow down to two or three. What comes up is fairly settled per person: a dropped third-person s, a missing article, present tense for a past event. Once it narrows, make only that the theme of the next lesson. Do not try to fix everything at once. One is what you can hold in mind during a single lesson.

How to take lessons differently — only three things change

From here on, only things you can do without adding lessons. In other words, changing how you take them. There are three, and the order matters: if the first is not in place, the second and third spin. Other things get recommended as effective ways to take a lesson, but narrowed to the goal of stripping off fossilized forms, it converges on these three.

① Notice what you could not say

Acquisition starts where you notice — that is a line often cited in second language acquisition. What a learner notices in the input is what becomes intake. Put the other way round, what you do not notice passes through however many times you are bathed in it.

In practice: hold on to one place where you went silent during the lesson, and look it up afterwards. Write what you wanted to say in your own language, as it is, and later fit exactly one English expression to it. Do not line up several ways of saying it. Only one reaches your mouth next time, so deciding on one makes it usable.

What you could not say matters more than what the teacher corrected. Corrections stay with you; what you could not say leaves only the sensation of not having said it, and you stop at the same place next time.

② Ask for a different kind of correction

Being asked to say it again, or having the difference explained in words, has a larger effect than a quiet recast — that was the meta-analysis in the earlier section. You can change this by asking your teacher. One sentence at the start of the lesson is enough.

All three are short. You do not need to say them every time; pass on the first one at the start and the teacher will follow. You may worry about the conversation stopping, but the place where it stops is exactly the place you can fix right now.

③ Use something slightly above what you can already do

Run the conversation only inside what you can already say and you reinforce the current shape. Use something deliberately a step up and it gets tested, and what is missing comes into view.

It lasts if you decide the method in advance: go in with exactly three expressions you intend to use in the next lesson, chosen from last time’s notes. Afterwards mark each as used or not. The ones marked not carry over.

There is no evidence behind the number three. It is chosen for being sustainable. At five, carrying them in becomes the burden itself, and before long you stop.

How to use the time around a lesson (before adding more of them)

For the same number of lessons, how you use the time around them changes how much stays. This is an area with accumulated research, so what is known can be used as it is.

One is the shape of review. In an experiment on items already learned once, continued rereading was compared with continued recall testing. Rereading had almost no effect on later recall; recall testing had a large one. Opening your notes and looking at them is the side that did not work.

The other is spacing. A meta-analysis of 48 experiments (3,411 participants in total) on second language learning found medium to large effects for spaced practice. Longer and shorter gaps also looked the same on an immediate test, while the longer gaps left more on a delayed one.

Put together, what you do is "write with the answer hidden" and "open it the next day, three days later and a week later". Three to five minutes each time. Trying to set aside a separate block for review mostly fails. Deciding to open the previous page before you open the next lesson is what lasts.

What to write, how to hide it, and at which intervals to open it are covered with steps and a template in a separate article. This one does not repeat them.

As for preparation, it is better not to overdo it. Read the material closely before the lesson and the lesson turns into checking answers, and the thing you most want to collect — what you could not say — never appears. Stop at choosing three expressions you want to use.

If you use a tool, which one tells you "how many times"

There are many tools for keeping records, but in the context of this article the criterion narrows to one: does it tell you how many times the same mistake has come up. Any of them can hold what you write, so that is not where they differ.

Paper is the strongest tool for the speed of opening it right after a lesson. One line down the middle hides the right-hand side. Its weakness is that you cannot search it; keep it up for about three months and "where did I write that expression" will happen. A spreadsheet lets you hide the right-hand side by hiding a column, and lets you sort afterwards.

If you use Notion, build it as a database rather than a table. Add properties to the items and you can filter by date or by type. If you have a transcript, importing it as CSV means no typing by hand. With Anki, decide the note type and put the cards in a deck, and answering Again or Good for whether you recalled it decides the next interval for you. Not having to manage the intervals yourself is what this tool is for.

You can also have ChatGPT or Claude organize things. With either, putting your instructions inside Projects gets the same shape back each time. But hand the correcting itself over and the chance to notice disappears from your side. As written in the earlier section, what you do not notice is not taken in. If you use them, use them as "have it produce candidates, and you choose".

Choose anything other than the top two and adding the dates stays on your side. That is the part that creates the "how many times", so dropping it removes the point of choosing a tool at all. Put the other way round: paper works too, as long as you add the dates, and you still get the most important information in this article.

Summary — what to do from your next lesson

• Know the figure. B1 to B2 is 100–250 hours, or 240–600 lessons of 25 minutes. Nothing changing in six months means you are still partway.
• Self-study is not in those hours. Mix it in and you read your own position as further along than it is.
• Do not judge by the feeling of progress. Count how many times the same mistake comes up. Three in a row is a habit.
• Ask for one thing at the start of the lesson — Could you stop me when I make a mistake, and ask me to say it again myself?
• Bring back one thing you could not say, and settle on exactly one English expression for it. Do not line several up.
• Go in with three expressions to use in the next lesson. Mark afterwards whether you used them.
• Make review recall-shaped with the answer hidden, and open it the next day, three days later and a week later. Three to five minutes each time.
• Adding lessons comes last. Changing what happens around them first is not too slow.