Four things reliably reduce LINE block rate, and each of them has an account size, a cost or a reverse failure that the advice never mentions. Segmented delivery is locked out entirely below 100 target reach. Cutting frequency stops working past a point and takes revenue with it. Triggered sending outperforms both, and costs developer time you may not have. Changing where your friends come from moves the number further than any of them, and is the one nobody suggests. This is each remedy written as a procedure for a LINE Official Account, with the entry conditions stated first, because a fix you cannot run is worse than no fix at all.
Table of Contents
Segmented delivery works, once you clear two thresholds
This is the most recommended remedy in every market, and LINE’s own manual quietly gates it twice. To filter a broadcast by attribute you need ターゲットリーチ数が100人以上必要です, at least 100 target reach. To actually send after filtering you need 選択後の推計対象ユーザーが50人以上必要です, at least 50 estimated recipients.
So a new account with 80 reachable friends cannot segment. An account with 400 target reach that filters down to women aged 20 to 29 in Osaka may land under 50 and be refused. Both cases fail after you have already built the campaign.
The second problem applies at every size. The attributes you are filtering on are inferred, and LINE says so on the same page: 年齢や性別など、属性で指定できるユーザー情報はLINEの利用動向から推定したみなし属性に基づいています。そのため、実際の属性と異なる場合があります. Age and gender are guesses derived from usage patterns and may differ from the real thing. You can therefore be locked out of the fix at small size and mis-targeted by it at large size, which is an uncomfortable combination for the remedy everyone leads with.
Use it anyway once you clear the thresholds. Just build the segment, check the estimated recipient count before you write the creative, and stop treating the demographic labels as facts about your customers.
Sending less often helps, until it starts hurting
The evidence for cutting frequency is genuinely strong, and it is worth quoting properly rather than as folklore. Mobilus surveyed 655 Japanese consumers aged 20 to 60 and over between 5 and 11 June 2025. Seventy percent had blocked a LINE Official Account at some point, rising to 85.0% among people in their twenties and 74.8% among people in their thirties, falling to 61.4% among the over sixties.
| Reason given for blocking | Share |
|---|---|
| Message frequency too high | 26.5% |
| Tidying up their account list | 25.2% |
| No useful information arrives | 23.3% |
| The campaign or sticker period ended | 20.7% |
| Cannot use the account to ask a question | 8.0% |
On cadence, a 2023 L Step survey of 1,000 people registered to LINE Official Accounts found 39.2% naming about once a week as appropriate, the most common answer.
So send weekly and you have solved it. The survey supports it, the reason ranking supports it, and if you want the safest possible account you would send less than that, and the safest account of all would be the one that never sends anything. That is where the logic breaks. SocialPLUS found that accounts delivering automated behaviour-triggered messages through the API had lower block rates than accounts sending nothing at all: 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. Going quiet is not the floor. It is its own failure mode, and Japanese guidance on cadence warns that irregular sending gets the account forgotten, which is why fixing a send day matters as much as the frequency.
One more caution. Several Japanese blogs publish neat tables of block rate by weekly send count, along the lines of 15% to 25% at weekly and 35% or more at three times a week. The source I traced labels its own table 目安値, a guideline estimate that varies by industry and operation, and recommends measuring your own data instead. Those bands are not measurements. Do not plan against them.
Triggered sends beat both scheduled sends and silence
The strongest finding in the public data is about mechanism rather than volume. SocialPLUS describes API delivery as カゴ落ち配信やお気に入り登録した商品の再入荷通知, cart abandonment and back-in-stock alerts for saved items, automated against site behaviour and user data. Those accounts show lower block rates than accounts broadcasting manually, and manual sends get worse as frequency climbs, with daily the worst case.
The entry condition is the catch. This needs the Messaging API and somebody to build against it, which for most operators means a vendor. The cost condition is worth stating precisely, because the billing rules are not intuitive:
| Message type | Counts against quota? |
|---|---|
| Broadcast, segmented and step delivery | Billable |
| Messaging API Push, Multicast, Broadcast and Narrowcast | Billable |
| One to one chat | Free |
| Auto-reply, 応答メッセージ | Free |
| Greeting message, あいさつメッセージ | Free |
| Messaging API Reply | Free |
So triggered sends through the Push and Multicast endpoints cost you the same per recipient as a broadcast. What you buy is relevance, and the block rate data says relevance is what the money is actually for.
Use the window right after somebody adds you, because part of it is free
Look again at that billing table, because it contains a genuine arbitrage. The greeting message and auto-replies cost nothing and are never counted against your plan. Step delivery, which is what most operators reach for when they want a sequence after signup, is billed per recipient like any broadcast.
That means the same onboarding content is free in one place and chargeable in another. Anything you can compress into the greeting message and a set of auto-replies triggered by keywords, put there. Reserve step delivery for the parts that genuinely need to arrive on day three and day seven.
The reason this window matters more than any other is the Mobilus row at 20.7%. One in five people who blocked did so because the campaign or sticker period ended, meaning they were acquired by an incentive and left when the incentive expired. That cohort decides quickly. SocialPLUS reports that around 60% of friends acquired through sticker campaigns block within a month.
One operational detail that catches people. Japanese guides on step delivery report that a block stops the sequence, and if the same user later unblocks, LINE treats it as a fresh friend addition and the scenario restarts from the beginning. Somebody who left at step two and came back six months later gets your welcome message again.
Changing where friends come from moves the number most
Everything above is about what you send. This is about who is receiving it, and the spread is larger than any content improvement will produce.
| Acquisition channel | Reported block rate |
|---|---|
| LINE Points friend ads | Around 70% |
| Promotion and sponsored stickers | Around 60% within one month |
| Friend-add ads (CPF) | Around 25% |
Between the worst and best paid channel there is a 45 point gap. No segmentation strategy produces 45 points. If your account was built on stickers or Points, the block rate you are being asked to fix was largely determined before a single message went out, and the remedy is a budget decision rather than a content one.
Why a falling block rate can be the worse signal
Block rate is total blocks divided by total friend additions, and the friend addition count is cumulative and never decreases, not when somebody blocks and not when they delete their LINE account. Two consequences follow, and both invert the reading.
Run a heavy acquisition push and the denominator jumps immediately while the numerator lags. Your block rate falls that month. If those friends came from stickers, roughly 60% of them will block within about thirty days, so the falling number you reported in March is the leading indicator of the rise in April. A block rate that improved during your biggest acquisition month is usually telling you about the acquisition, not the content.
Now the reverse. Stop acquiring and the denominator freezes while blocks keep accruing. The rate climbs even if your sending improved. A quiet quarter produces a worsening number by arithmetic alone.
So judge the fixes on absolute blocks per period, which the console shows against yesterday, seven days and thirty days, and on cohort block rate by acquisition month. Take the friends added in one month, count how many of them have blocked ninety days later, and compare that cohort against the next. That is the only version of the measurement that responds to what you changed.
The five fixes side by side
| Fix | Entry condition | What breaks it | Message cost |
|---|---|---|---|
| Segmented delivery | 100+ target reach to filter, 50+ estimated recipients to send | Small accounts locked out. Inferred attributes may be wrong at any size | Billable per recipient |
| Lower frequency | None | Irregular or absent sending gets the account forgotten, and fewer sends means less revenue | Reduces spend |
| Behaviour-triggered API sends | Messaging API plus development resource | Push, Multicast and Narrowcast are all billable. Only Reply is free | Billable per recipient |
| Greeting message and auto-replies at add time | None | Cannot be spread across days the way step delivery can | Free |
| Change acquisition channel | Budget reallocation, and the ad products need a verified account | The cheapest adds block hardest, so cost per retained friend has to be computed | Varies |
If you can only do one thing this quarter, move the acquisition budget. If you can only do one thing this week, rewrite the greeting message, because it is the one fix on the list that costs nothing per recipient and lands in the window where the leaving happens.
