A welcome email goes out to 10,000 new signups. 9,840 get it. The other 160 don’t, and nothing in the dashboard explains why: no bounce, no spam complaint, no unsubscribe. The campaign just looks like it skipped them.
Customer.io does not send an incomplete message, it blocks it
Customer.io’s own documentation is direct about this: if a Liquid tag in a message can’t resolve, like a customer who doesn’t have a first_name attribute, the message will not send. That is a deliberate choice, not a bug, built to stop people from receiving a message with a visible hole in it, a “Hi ,” instead of a name. The cost of that protection is that the send fails outright for that recipient, and shows up as Failed in the message log rather than as a warning anyone sees in open or click rates.
Why it looks random instead of broken
The failure is scoped to the recipient, not the campaign. Only the people missing that specific attribute fail to receive the message. Everyone else, with a first_name on file, gets it without a problem. From the outside that reads as a handful of customers randomly not getting an email, not as a platform issue, because 98% of the send worked fine. The only place the pattern shows up is the message log, filtered to Failed, which most teams check after a bounce spike, not after a quiet gap in delivery.
The fix is one filter, not a new tool
- Audit every personalization tag in your active journeys, not just the welcome series. Any tag pulling a customer attribute, first name, company, plan tier, is a candidate for a missing value on some fraction of your base.
- Add a default filter to each one.
{{ customer.first_name | default: "there" }}keeps the send alive with a generic fallback instead of blocking it. Customer.io’s docs note the filter needs their current Liquid version, so confirm which version an older message is running before assuming the filter will apply. - Use an if/else block where a filter isn’t available, same logic, more lines: show the personalized value if it exists, a fallback line if it doesn’t.
- Check the Failed count in message logs on a schedule, not only when something else breaks. It is the only signal this problem leaves, and it never shows up in open rate, click rate, or deliverability reports.
This is the same root cause behind why a CDP can split one customer into two profiles: a gap in the data that the platform assumes will always be there. The fix for both is the same discipline, treating every personalization field as something that can be missing, not something you plan growth automation around assuming it never is.