The Average LINE Block Rate Does Not Exist, Your Benchmark Does

There is no normal LINE block rate, and the sources that ought to settle the question contradict each other in public. A SocialPLUS survey of the LINE Official Account clients it supports puts the mean at 29.7% and the median at 27.0%. LINE Yahoo for Business, writing on its own note account, publishes 36.12% as the overall average. Most Japanese agency blogs repeat 20% to 30% and name no source at all. Every one of those is a real published figure. Not one of them is your figure, because the number shifts with your market, your account age, your friend count, your industry and the channel your friends arrived through. So take five benchmarks instead of one average.

Agree on what the number is before arguing about it

The calculation is consistent across the Japanese sources. Total blocks divided by total friend additions. The Japanese guide at lme.jp writes it as ブロック数／友だち合計数 and works the example plainly: an account with 100 friends, 20 of whom have blocked, has a block rate of 20%.

You will not find that percentage printed anywhere in LINE Official Account Manager. What the console gives you is a raw block count, under 分析 and then ブロック, with comparisons against yesterday, seven days ago and thirty days ago. The percentage is yours to compute. That is part of why so many different ones circulate.

So track the percentage, and when it climbs past 30% go and fix your delivery. That is the standard advice, it is what the console nudges you towards, and I would hand you the same advice if the fraction behaved like a fraction. It does not. The denominator cannot fall.

Japanese documentation is direct about this. The friend addition count, 友だち追加数, is cumulative and permanent: 友だち追加して、その後ブロックをした場合でも、ユーザーが自身のLINEアカウントを削除した場合でも、友だち追加数は減りません. Somebody who blocks you stays in the denominator. Somebody who deletes their LINE account outright stays in the denominator. The numerator behaves differently, because the block count goes down when a user unblocks and climbs again if they block a second time. You are watching a ratio between a number that only grows and a number that wanders, on an account that gets older every month. Upward drift is structural. Read the level as a diagnosis and you will misread it.

The three published averages, and who each one is measuring

Source	Figure	Who is in the sample
SocialPLUS survey (Japanese, published August 2024)	Mean 29.7%, median 27.0%	Accounts SocialPLUS supports, with accounts under 1,000 friends excluded
LINE Yahoo for Business, official note account	36.12% overall	Accounts under the author’s management, with industries described as roughly 30% to 40% and no extreme gaps between them
General Japanese agency blogs	20% to 30%	Unstated. Written as 言われています, meaning it is what people say

The gap between 29.7% and 36.12% is not a measurement error. The SocialPLUS sample cuts out every account below 1,000 friends, and small accounts sit at the low end of the curve, so that exclusion should push the mean up rather than down. The likelier explanation is composition. Different client rosters, different industries, different acquisition habits.

Japan benchmarks against an average, Taiwan benchmarks against age

This is the split that catches operators running accounts in more than one market. Japanese practice is to compare against a single population average. Taiwanese practice is to compare against how long the account has been running, on the reasoning that blocks accumulate while the friend count is cumulative, so an old account should look worse than a young one.

Market	How the benchmark is framed	Published guidance
Japan	A single population figure	Mean 29.7% and median 27.0% (SocialPLUS), or 36.12% (LINE Yahoo for Business). Agency guidance often treats above 30% as worth investigating
Taiwan	By account age	Within the first year, 10% to 20% is the standard band. Past one year, 20% to 30% is accepted, and up to 35% is described as nothing to worry about. Past two years, staying inside 45% is still called safe
Taiwan (alternative agency framing)	By absolute level	Most brands land at 20% to 30%. Holding under 10% indicates content and segmentation are working. Regularly above 40% means go back and check the data
Thailand	By absolute level	Thai agency guidance describes roughly 20% to 30% as acceptable performance, and notes accounts reaching 50% or higher

LINE Biz-Solutions Taiwan, in its own column on message design, takes a position worth borrowing whichever market you are in. It argues the block rate is a reading of fit between your audience and your content rather than a verdict on your competence, and it recommends setting incremental targets: 如果商家目前的封鎖率為 50%，則可以先以降低至 45% 為目標. If you are at 50%, aim for 45% first. That is a more useful instruction than any industry average.

Bigger accounts are supposed to look worse

Inside the same SocialPLUS dataset, block rate rises with friend count. The article states it directly: 友だちが5,000未満の場合、約22％であるのに対し、友だちが500,000以上の場合40％を超えています. Under 5,000 friends the figure sits around 22%. Above 500,000 friends it passes 40%.

That is nearly a doubling across the size range, and it has a plain cause. Accounts do not reach half a million friends through posters in a shop window. They get there through sticker campaigns and ad menus that pull in large numbers of people indiscriminately, and those people were never interested in the brand. Judging a 300,000 friend account and a 3,000 friend account against the same percentage is the most common benchmarking mistake I see in English writing about LINE.

Your industry moves the number by nearly 17 points

The SocialPLUS breakdown by product category is the most granular public data available, and the spread between the best and worst category is 16.9 percentage points on the mean.

Category	Mean	Median
Recruitment and HR	38.4%	35.8%
Cosmetics	36.6%	34.8%
Health supplements	33.9%	29.2%
Medical and beauty	33.8%	37.0%
Real estate	31.8%	30.1%
Apparel	30.7%	27.6%
Food	27.8%	22.8%
Insurance and finance	22.7%	24.0%
Daily necessities	22.3%	19.1%
Hobby, leisure and entertainment	21.7%	18.8%
Member services	21.5%	15.8%

Read the median column as carefully as the mean. In most rows the median sits below the mean, which tells you a minority of badly performing accounts is dragging the average up and that the typical account in that category is doing better than the headline. Two rows invert that. Medical and beauty has a median of 37.0% against a mean of 33.8%, and insurance and finance has a median of 24.0% against a mean of 22.7%. In those two, the typical account really is worse than the average suggests, and a handful of strong performers are pulling the mean down. If you are in either, benchmark against the median.

Worth noting that LINE Yahoo’s own analysis reaches the opposite conclusion about industry, saying 各業種でおおよそ30～40％となっており極端な差はありません. Roughly 30% to 40% across the board with no extreme differences. Two datasets, two verdicts. I would use the SocialPLUS table because it publishes its numbers, while the LINE Yahoo industry breakdown is presented only as a chart.

Where the friends came from matters more than anything you send

This is the benchmark that reframes the whole exercise. SocialPLUS publishes block rates by acquisition channel, and the spread dwarfs everything above.

Acquisition channel	Reported block rate	Window
Promotion and sponsored stickers	Around 60%	Within one month of adding
LINE Points friend ads	Around 70%	Not stated
Friend-add ads (CPF)	Around 25%	Not stated

The reasoning is unsurprising once you see the numbers. Sticker and Points users add the account to collect a reward, and they leave once they have it. Friend-add ads reach people with some genuine interest in the brand. An account that ran one sticker campaign has a mathematically different ceiling from an account that never did, and no amount of content improvement will move it back.

Accounts that automate beat accounts that stay quiet

The fifth branch is how you send. SocialPLUS found that accounts delivering through the Messaging API at behaviour-relevant moments, things like cart abandonment and back-in-stock alerts, show lower block rates than accounts doing manual broadcasts, and lower than accounts sending nothing at all: API配信を通してユーザーにとって必要なタイミングで必要な情報を配信しているLINE公式アカウントでは、全く配信をしていないケースよりブロック率が低くなっている. Manual sends get worse as frequency rises, with daily broadcasts the worst case.

Silence is not the safe option. That is the single most useful thing in the dataset.

Track these instead of the percentage

Because the denominator is frozen, the ratio is a poor operational metric. Four alternatives that move for reasons you control:

• Absolute blocks per period. The console already shows this against yesterday, seven days and thirty days. It responds to what you sent this week.
• Target reach trend. Target reach is the population you can actually bill against, and it falls when blocks rise. Watch the line, not the level.
• Cohort block rate by acquisition month. Take the friends added in March, count how many of them blocked by June. This is the only version of the metric that is comparable between a two-year-old account and a two-month-old one.
• Your own last quarter. Every Taiwanese and Japanese source that thinks about this carefully ends up saying the same thing. Compare the account to itself.

Pick your benchmark

If your account is	Compare against
Under 1,000 friends, in any market	Nothing published. Every dataset here excludes you. Use your own cohort trend
Under 5,000 friends, Japan	Around 22%
Over 500,000 friends, Japan	Over 40%, and treat that as expected rather than as failure
Under one year old, Taiwan	10% to 20%
Over two years old, Taiwan	Inside 45%
Any age, Thailand	20% to 30% acceptable, 50% is the level Thai agencies flag
Built substantially from a sticker campaign	Assume around 60% of that cohort is gone within a month and benchmark the remainder separately
Recruitment, cosmetics or supplements, Japan	The high thirties is normal for your category
Member services or leisure, Japan	The low twenties, and a median as low as 15.8%

If somebody has handed you a number and told you it is bad, the first question is not how to reduce it. It is which of those rows you are in.