Every WhatsApp message template falls into exactly one of three categories. You choose it at submission. Meta’s review then checks whether the category you chose matches what you actually wrote — and if it does not, it can reclassify the template for you.
That one field decides three things simultaneously: how strictly the template is reviewed, what content rules apply, and which price tier the message falls into. It is the most consequential dropdown in the product.
The three categories
Utility covers messages tied to a transaction: order confirmations, dispatch and delivery notifications, appointment reminders, recurring billing statements, account updates the customer is already expecting. These are generally the easiest to get approved, because the intent is unambiguous.
Authentication covers one thing: verifying a customer during login, registration or account recovery. These use copy-code or one-tap autofill buttons to deliver a passcode. They are held to the strictest format compliance of the three — this is the category where adding brand voice actively hurts you.
Marketing covers anything promotional: offers, new arrivals, re-engagement, product announcements, abandoned-cart nudges. Carousel templates sit under Marketing as a subcategory. Marketing templates cost the most and require opt-in you can evidence.
The rule people get wrong
The dividing line is intent, not tone.
A message can be politely worded, contain no exclamation marks, and still be Marketing — because its purpose is to get someone to buy something. Conversely, a blunt “Your order #4821 has shipped” is Utility no matter how flat it reads.
The failure mode we see most often is a hybrid: a receipt that also carries a discount code, a delivery notification with “and here’s 10% off your next order” stapled to the bottom. Each template must have one clear, singular purpose. Hybrids get rejected, and the rejection is correct — that message is doing two jobs and only one of them is Utility.
Don’t try to arbitrage the price tier
Utility is cheaper than Marketing. This creates an obvious temptation: write the promotion, submit it as Utility, save the difference.
It does not work reliably, and the downside is asymmetric. Meta’s review is specifically built to catch category mismatches. When it catches yours, you do not just lose that template — repeated misclassification feeds into your account’s quality signals, and low quality scores reduce your daily messaging limits.
You are trading a small per-message saving against your ability to send at all. That is a bad trade, and it is a slow one to unwind.
If you disagree with a reclassification
You are not stuck with it. If a template you submitted as Utility comes back categorised as Marketing, you can request a category review — per Meta’s documentation, within 60 days of the category update date.
The request goes through WhatsApp Manager rather than the API: Message Templates → the rejection banner → Go to Business Support → Template Category Updates → select the templates → Request Review. If the review succeeds, the category is updated and a template_category_update webhook fires carrying the new category, which is the signal your own systems should listen for rather than polling.
If it fails, the category stands.
What this means for how you organise
Two habits make this manageable rather than painful.
Name templates by category. If your library is order_shipped_v2 and spring_sale_v1, you already know which review path each one is on. If it is template_17, you will re-derive that fact every time.
Write the category into your brief, before the copy. Deciding “this is Utility” first constrains the writing usefully — it rules out the promotional sentence someone will otherwise add at the end, and it stops the hybrid before it exists.
One more thing worth internalising: once a template is approved, you cannot edit its text. Any change means creating and submitting a new template. That makes the first draft matter more than it does in most channels — and it makes the character budget something to plan against rather than discover.