Quality rating and messaging limits on WhatsApp

A business that suddenly cannot send as much as it did yesterday is usually running into one of two distinct mechanisms, and telling them apart is the first step to fixing anything. The quality rating is a per-number assessment of how recipients have been reacting to your messages. The messaging limit is a cap on how many unique customers you may start conversations with in a rolling period. They are related — quality influences whether a limit rises or falls — but they are not the same thing, and the remedies differ.

Quality is expressed as green, yellow or red, corresponding to high, medium and low. It is derived from recent recipient feedback: blocks, reports, and signals that the messages were unwanted. It is assessed per phone number, so a business running several numbers can have one in trouble while the others are fine.

The messaging limit is a number, and it moves up a ladder as a business demonstrates sustained, well-received sending. Meta's documented ladder has historically started with a small daily allowance for new and unverified numbers and progressed through several rungs toward unlimited. Because the specific figures at each rung have been revised, the ladder's shape is the durable part and the current values are worth reading from Meta's own documentation rather than from any summary of it.

The rating is built from what recipients do, not from what the business intended. The dominant inputs are blocks and reports, weighed against the volume sent, over a recent rolling period. A handful of blocks against a large, engaged audience reads differently from the same number of blocks against a small one.

This is worth dwelling on because it explains why compliant messages can still damage an account. Nothing in the rating asks whether a template was approved, whether the category was correct, or whether consent was technically on file. Those things matter for other reasons, but the rating responds to reaction. A message that was approved, correctly categorised and sent with a valid consent record can still be unwelcome — because it arrived at a bad time, because the recipient forgot they had opted in, or because it simply was not relevant to them — and the rating will register that.

The rolling nature of the assessment has a useful corollary. Because the window is recent rather than lifetime, a period of good behaviour genuinely improves the picture, and an old bad campaign eventually stops counting. The rating is a description of your current sending, which is what makes recovery possible at all.

A small number of patterns account for most rating damage. Messaging people who do not remember agreeing to hear from you is the largest: consent that is old, vague, or collected for a different purpose produces recipients who experience the message as unsolicited regardless of what the record says. Sending the same promotional content to an entire list without segmentation is next, because relevance falls as reach widens, and irrelevance is what blocks are made of.

Sudden volume increases are a distinct risk. A number with a modest sending history that abruptly dispatches thousands of messages presents a pattern that abuse detection is specifically built to notice, quite apart from how recipients respond. Frequency compounds it: several promotional messages in a week to the same person will produce blocks even from people who genuinely opted in.

Making it hard to leave is the quietest cause of damage. If a recipient cannot find a way to stop the messages, the block button is the available alternative, and a block is materially worse for the account than an opt-out. The same applies to stop requests that are not honoured promptly — a person who asked to be removed and then receives another message has a reason to escalate from leaving to reporting.

A drop to yellow is a warning rather than a penalty: it signals that recipient feedback has deteriorated and the number is being watched. Nothing necessarily changes about what the business can send, which is exactly why yellow is the moment to act. It is the only cheap point of intervention in the sequence.

A drop to red is consequential. It puts the number's messaging limit at risk of being lowered, which restricts how many unique customers the business can start conversations with. The restriction applies to the number, so it affects every kind of outbound messaging that runs through it — including the transactional confirmations and reminders that have nothing to do with whatever campaign caused the problem. This is the most under-appreciated cost of an aggressive marketing send: the collateral damage lands on the messages customers rely on.

Template-level quality is tracked separately from number-level quality. An individual template can accumulate poor feedback and be paused or disabled while the rest of the account continues functioning. That granularity is useful diagnostically — a single badly-performing template is a narrower and more fixable problem than a number-wide decline — and it means the template list is worth reviewing rather than only the account dashboard.

Because the assessment is based on a recent rolling window, a rating recovers by accumulating a period of well-received sending. There is no appeal that substitutes for this and no way to purchase a reset. The mechanism is time plus better behaviour, in that order.

What that means in practice is counter-intuitive for a business under pressure to keep marketing. The effective response to a red rating is to reduce volume, not to maintain it while making the messages nicer. Stopping promotional sending entirely for a period, continuing only the transactional messages people actively want, and resuming marketing later against a properly consented and segmented list is the shape of a recovery. Continuing to send at volume while hoping the average improves keeps feeding the same window that produced the rating.

The rebuild afterwards should be deliberate. Increase volume gradually rather than returning to previous levels at once. Segment so that each message is relevant to the people receiving it. Audit the consent record for method, timestamp and category, and drop contacts where any of the three is missing. This is slower than resuming normal service, and it is the difference between a recovered number and a number that oscillates between yellow and red indefinitely.

The rating is visible in Meta's own management surfaces, per number, and Meta notifies a business when a rating changes. The habit worth building is checking it as part of a routine rather than in response to a crisis, because the useful interventions all live upstream of a red rating.

A few leading indicators are more informative than the rating itself, because they move first. Block and report rates relative to volume tell you where the rating is heading. Read rates falling while delivery stays high suggests messages are arriving and being ignored, which precedes blocks. Reply rates on templates that invite a response indicate whether the audience is still engaged. Any of these turning down is a reason to look at the list and the frequency before recipient feedback formalises the problem.

The underlying discipline is simple to state. Send fewer, more relevant messages to people who genuinely agreed to receive them, make leaving easy, and grow volume gradually. Every element of the quality system rewards that pattern, and no amount of template craft compensates for its absence.