The WhatsApp 24 hour window is the real reason your broadcasts keep failing

Free-form messages only exist inside a window the customer opens

WhatsApp's Business Platform runs on one rule that has nothing to do with SMS logic. A free-form message, anything you write yourself with no prior approval, can only be sent inside a 24 hour window that opens the moment a customer messages you. Every time they message again, the clock resets.

The second that window closes, free-form messaging is switched off for that customer. Nothing you type reaches them until they message you first, or until you send a pre-approved template that reopens the conversation. Meta sets this out in its Business Platform documentation, and almost nobody selling WhatsApp broadcasts mentions it.

SMS has no equivalent. You buy a number and send to whoever opted in, subject to consent and quiet hours. WhatsApp was built as a place where a business replies like a person, not a channel that broadcasts to a list. Everything below follows from that.

Why half a broadcast lands and the rest quietly vanishes

A brand builds a decent opt-in list, sees genuinely excellent open rates, and treats the channel exactly like SMS anyway. A WhatsApp open rate is a delivery stat wearing a performance metric's clothes. It tells you the message arrived on a phone that opens the app all day. Not that the channel is working.

Then the promo goes out. Some of it lands instantly, a big chunk lands hours late or not at all, and the business account picks up a warning. The team blames the Klaviyo integration or the send time, because nobody told them the audience was split in two.

A list mixing people three hours into the window with people three weeks outside it gives one send two completely different outcomes. There's rarely a dramatic error. Just partial delivery, and a message the platform expected as a template and refused as free text.

A template is not SMS with a green tick

Outside the window, every message has to be a template Meta has already reviewed and approved. Every template carries a category: utility, marketing or authentication. The category field gets treated as a formality and marketing gets picked for almost everything, because that's what the message commercially is.

Meta keeps judging a template after approval, not just at review. A marketing template sent repeatedly to numbers that haven't engaged recently reads as exactly what it is, a promotional blast through a channel designed for conversation. Do it consistently and the quality rating on your phone number drops. Then Meta caps how many people you can message.

The category is decided by what the message does for the customer, not by what you feel like calling it. A dispatch update with a bit of brand personality is still utility, because the customer expects it as the direct result of something they did. A discount code sent cold to somebody silent for three weeks is marketing, and mislabelling it to dodge the stricter review is the exact pattern that gets accounts flagged.

Let the trigger open the window, then ride inside it

The WhatsApp programmes we build are built around one mechanism, and it's the difference between a channel that performs and one that quietly dies within months. A legitimate, expected trigger event opens the window. The conversational, higher-intent message then rides inside that window as free-form text, with no separate approval needed.

Take a real dispatch flow. The shipping notification is a utility template, expected and useful, and Meta approves those without friction because it genuinely is a service message. The moment it's delivered, the window is open again.

A few hours later, still inside the window, the flow follows up with a free-form note written the way a person would write it. A quick question about whether the size worked, or a pointer to a product that pairs with it. No template approval, no marketing quota spent, and if the customer replies the clock resets and the conversation stays open.

Compare that with the default: a string of marketing templates fired on a schedule regardless of the window, because the automation was originally built for SMS. Every one needs approval. Every one spends quota, and chips at the quality rating whenever it lands on a cold number. The same trigger thinking that runs a properly built flow programme is what keeps WhatsApp deliverable.

Run these checks before any flow goes live on WhatsApp

Catching problems before launch is far cheaper than catching them after a quality rating drop, because ratings fall fast and recover slowly. Five checks cover most of it.

• Map every trigger in the flow and mark it honestly as utility or marketing, never letting the two blur
• Check the last-contact timestamp on the segment before any send, since opted in and inside the window aren't the same thing
• Register every template under its true category, even when the stricter marketing review takes longer to come back
• Watch the messaging limit tier on the business account weekly, because a quality drop shows up fast and the climb back is slow
• Build a genuine two-way moment early in every flow, a question that gives the customer a reason to reply and reopen the window themselves

The last one matters most. A reply is the only thing that keeps a window open without spending a template, and a flow that never invites one is leaving the clock to run out on purpose.

What it looks like when the quality rating goes

The failure compounds instead of staying flat. A brand routes promotions as marketing templates to a list that skews cold, the rating slides from green to amber, and Meta cuts the messaging limit. The team pushes harder with the reduced volume, sends more often to the smaller pool, and the rating drops again.

Within a few weeks a channel that was beating email on opens is reaching a fraction of the list it started with. No better broadcast fixes that. What needs rebuilding is the flow architecture itself, from the trigger events up, which is the work running underneath the accounts we manage.

Most brands never see it coming, because the platform doesn't warn clearly and the symptoms look like a delivery bug rather than a structural one. A system designed around a 24 hour clock will punish anybody who uses it like SMS with a green tick.