WhatsApp quality rating and messaging limits

Two things decide how much outbound messaging your WhatsApp number can do, and almost every article about them treats them as one thing. They are separate mechanisms with separate inputs, and knowing which one is biting you is the difference between fixing the problem and waiting for it to fix itself.

Quality rating is Meta's read on how the people you message react to you.

Messaging limits are the cap on how many unique customers you may start a conversation with in a rolling 24 hours.

Quality feeds limits. Limits do not feed quality.

Quality rating: what it measures

The rating sits on a phone number and, separately, on individual templates. Its inputs are recipient behaviour: people blocking your number, and people reporting your messages. Not your content, not your industry, not how well written the template was.

That distinction matters because it tells you what a falling rating means. It never means "Meta disliked this message". It means people who received it did not want it. The fastest way to damage a rating is to message people who did not ask to hear from you, and no amount of copywriting compensates for that.

The rating is visible per number in Meta's own WhatsApp management surface. A platform can show it to you, but nobody can change it for you.

What happens as it falls

The sequence is graded rather than sudden. A number in good standing sends normally. A number whose rating drops gets flagged, and the practical consequence is a reduction in what it may do, most visibly a drop in its messaging limit. A number that keeps falling ends up restricted.

Two things about recovery are worth knowing.

First, it recovers. The rating is a rolling read on recent behaviour, so a number that stops doing the thing that damaged it improves over time. Nothing needs to be appealed for the ordinary case.

Second, it recovers only if the behaviour changes. Continuing to send at the same volume to the same list, more carefully worded, does not move it.

Messaging limits: what they cap

The limit counts unique customers you start a conversation with in a rolling 24-hour period. Two facts follow from that definition and both are load-bearing.

Replies do not count. A customer who messaged you first is not part of the cap, no matter how many messages the two of you exchange. A business whose messaging is mostly inbound will rarely meet the limit at all.

It counts people, not messages. Three templates to the same person in a day is one unique customer. One template to three thousand people is three thousand.

Numbers begin at a low tier and step up as sending volume grows while quality holds. Business verification is a prerequisite for the higher tiers, and its timeline belongs to Meta rather than to any vendor. Start it early, before you need it, because the day you need it is the day you have a campaign booked.

The specific tier sizes are Meta's, and Meta has changed both the values and the upgrade mechanics more than once. Read the current limit for your own number in WhatsApp Manager rather than from a comparison article, including this one.

The third thing: rate limits

There is a third mechanism that gets folded into the other two and should not be. Rate limits are about speed, not about permission or reputation, and they show up as distinct send errors.

The first three are worth retrying automatically. The fourth is a signal about your list, and retrying it is the wrong response.

How to keep both healthy

Everything that protects quality also protects the limit, so there is one list rather than two.

Collect real opt-in and record it. What wording was shown, where, and when. When a number gets flagged, this is the record that tells you which segment caused it.

Honour opt-outs immediately and globally. A suppression list checked at send time rather than at list-build time, and applied across every campaign rather than the one that prompted it. Meta also signals opt-outs through send failures, and those should suppress the contact automatically rather than sitting in a report.

Segment before you send. A broadcast to everyone is the highest-risk message a business can send, because it necessarily includes the people least interested in it. Sending the same message to a quarter of the list, selected on behaviour, usually produces similar absolute engagement and a fraction of the complaints.

Pace the send. Releasing a large broadcast in batches with an interval between them does two useful things: it stays under the throughput ceiling, and it gives you a chance to stop after the first batch if the early signals look wrong. A campaign that sends over an hour can be halted. One that sends in ninety seconds cannot.

Watch quality per template, not just per number. The number's rating is a lagging summary. A specific template collecting complaints is the thing you can act on today.

Slow down when the rating dips. This is the counterintuitive one, and it is the one that works. The instinct after a quality drop is to send a better message. The action that recovers the number is to send fewer, to a better-chosen audience, for a while.

Warming up a new number

A number with no history has no reputation, good or bad, and its first few weeks decide which one it gets. The mistake is to treat a fresh number as a clean slate to send from immediately, because a first campaign to a large cold list is exactly the pattern the system is built to catch.

A warm-up that works has three properties.

It starts with inbound. Point your existing entry points at the number first: the website button, the click-to-WhatsApp ad, the link in your order confirmation email. Conversations customers start cost nothing against the limit and produce the engagement signals that support a tier increase.

It grows outbound gradually. The first outbound sends should go to the segment most likely to welcome them: recent customers, people who messaged you in the last month, people who explicitly asked for updates. Widen the audience over several sends rather than in one.

It watches after each step. Check the rating after each increase in volume rather than at the end of the month. A dip caught after one send is a small correction. The same dip caught after five is a flagged number.

There is no fixed schedule that fits every business, because the right pace depends on how much inbound you have. The usable rule is simpler: never let a single send be several times larger than the last one that went well.

What to do when the number is already flagged

• Stop the outbound marketing. Not reduce it. Stop it, for now.
• Keep answering inbound. Replies inside the conversation window do not count against the limit and do not hurt the rating.
• Find the template. Complaints attach to specific templates, so the per-template view usually identifies the campaign within minutes.
• Find the segment. Almost always it is a list that was bought, imported, or built from a consent nobody can point at.
• Suppress everyone the failures already told you to suppress. Codes for opted-out and undeliverable recipients are instructions, and re-sending to those contacts is what turns a dip into a restriction.
• Restart slowly, on an engaged segment, with a template that has a reason to be read.

The part that is easy to miss

The mechanisms above are usually discussed as compliance overhead. They are better understood as the reason the channel works at all.

Open rates on WhatsApp are high because people's WhatsApp inboxes are not full of messages from businesses they did not agree to hear from. Quality rating is the mechanism enforcing that, and it is enforced by recipients rather than by a policy team. Every business that treats the channel as a cheap broadcast list is spending down the property that made it worth using.

Sending less, to people who asked, is not a restriction imposed by Meta. It is the strategy that keeps working after the limits are raised.