Messaging dashboards are unusually flattering. Read rates north of 70% are ordinary on WhatsApp, and next to email’s open rates they look like a triumph. They are mostly a property of the channel, not of your campaign.
Here is what each number can actually carry.
Sent
The count your system attempted. It is a measure of your own behaviour and nothing else.
Worth tracking only as the denominator, and worth watching for one thing: if sent is materially below what you scheduled, you are hitting a messaging limit, and that is an account-health problem rather than a campaign problem.
Delivered
Reached the device. The gap between sent and delivered is the interesting part — a persistent delivery shortfall usually means bad numbers in the list, not a platform issue.
If delivery drops on one segment and not others, the segment is the story. Numbers collected from a single source going stale together is a very common pattern, and it is invisible if you only look at the aggregate.
Read
The recipient opened the conversation. This is the number that looks spectacular and means least.
Two reasons to distrust it as a measure of interest:
The notification is the message. WhatsApp shows the first couple of lines on the lock screen. A recipient can absorb “your order shipped” entirely without opening anything — which registers as unread despite the message having done its whole job.
Opening is not reading. Clearing a notification badge opens the conversation. Read receipts also depend on the recipient’s own privacy settings, which you cannot see and did not choose.
Read rate is a reasonable proxy for whether your messages are unwelcome — a collapsing read rate is a real signal. It is a poor proxy for whether they worked.
Clicked
The first number tied to an action the recipient chose to take. It is the most meaningful of the four and the most commonly misread, for one reason: click rate is a function of whether you gave them something to click.
A utility template with no button cannot have a click rate. Comparing it to a marketing template with a CTA is not a comparison. Segment by template structure before you compare anything.
The comparison that quietly misleads
Utility and marketing templates get compared to each other constantly, and they should not be.
A dispatch notification goes to someone who just bought something, is expecting it, and wants it. A promotion goes to someone doing something else. These populations differ in intent, timing and receptivity. The utility template will win every metric, every time, and the win tells you nothing you did not already know.
Compare marketing to marketing, utility to utility, and prefer comparing a template to its own history over comparing it to a different template.
What we would actually track
Four things, none of which appear on a default dashboard:
Delivery shortfall by source. Delivered ÷ sent, split by how the contact was acquired. This finds list rot early, and list rot is the cheapest problem to fix.
Response rate for anything expecting a response. If a template invites a reply, the reply rate is the only honest measure of it. It also opens a 24-hour window, which has operational consequences you should be staffing for.
Round trips per resolution. For service conversations: how many messages to reach an outcome. This is the number that governs agent cost today, and — if Meta’s announced October 2026 service-message pricing lands as described — the number that governs message cost too.
Block and opt-out rate. The only metric on this list that is unambiguously bad when it rises, and the one that feeds account quality signals and therefore your daily sending limits. Everything else is contextual. This one is not.
The honest summary
A 78% read rate is not evidence that a campaign succeeded. It is evidence that WhatsApp is a channel people look at.
Whether the campaign succeeded is a question about what happened next — replies, clicks, orders, resolved tickets — and answering it requires connecting the message to an outcome outside the messaging system. That connection is more work than reading a dashboard, and it is the only part that tells you anything.