Email has a send-cancel window and a forgiving audience. WhatsApp has neither. A message arrives on a personal device with a notification, it cannot be recalled, and mistakes generate blocks rather than shrugs. Ten minutes of checking is cheap.
The audience
- Every recipient has marketing consent — not merely transactional consent.
- Suppression is applied at send time, not from an export made earlier.
- Duplicates are removed on normalised phone numbers.
- The list excludes anyone who received a campaign in the last few days.
- The segment is the one you think it is. Open it and read ten rows.
The message
- The template is approved and in the correct category.
- Every variable has a fallback for missing data.
- Variable data is sanitised — no newlines, no absurd lengths, no junk.
- The business is identifiable from the message alone.
- Links work, are correct, and point at pages that exist. Click every one.
- An opt-out route is present.
Test on a real device, not in a preview pane. Send to five internal numbers, including at least one Android and one iPhone. Line breaks, emoji rendering and link previews differ, and a preview will not show you that.
The operational side
- Someone is available to answer replies. A broadcast to 5,000 people produces inbound, and silence in response is worse than not sending.
- The send is within your messaging limit, or deliberately staggered across days.
- The destination page can take the traffic.
- Any promoted stock or offer is actually available.
The reply problem
This is the check most often skipped. People reply to WhatsApp messages — far more than to email — and every reply opens a service window in which they expect a human. Sending a large campaign on a Friday evening with nobody rostered until Monday converts interest into irritation.
Estimate roughly: a good campaign to an engaged list can produce replies from a low single-digit percentage of recipients. On 5,000 sends that is a meaningful queue.
Start smaller than feels necessary
Send the first campaign to your most engaged few hundred contacts rather than the whole list. You learn the reply volume, the link behaviour and the block rate at a scale where a mistake costs little — and if quality drops, it drops from a small sample rather than your entire audience.
Afterwards
Watch three numbers in the first hour: delivery failures, block rate, and reply volume. A delivery failure spike usually means a data problem; a block spike means a targeting or expectation problem, and is the one that justifies pausing the remainder of a staggered send immediately.