Blog

How to write a fraud alert text members trust

Scam texts copy bank alerts. A real one earns trust by doing the few things a scam can't.

A plain grey card and a phone on a kitchen table, a group of people laughing in the bright kitchen behind.

Search for "fraud alert text message" and most of what comes back is advice for consumers on spotting fake ones. That tells you the problem a real alert has. Criminals send texts that look like they come from banks and credit unions, and members have been told, rightly, to be suspicious.

So a fraud alert has two jobs. It has to reach the member fast, and it has to look nothing like the fakes. Most scam texts need something a real alert never asks for.

What a scam needs that you don't

A smishing text wants the member to do something that hands over access: click a link to a look-alike login page, type in a one-time code, call a number the scammer controls, or reply with a card number. Your alert needs none of that. You already know the transaction. You only need a yes or a no.

That gives you a clear rule set:

  • No links. Not to "review the charge," not to "verify your identity." Once your alerts never carry a link, any alert with a link is suspect, and you can tell members so.
  • Never ask for credentials. No password, PIN, full card number, security answers, or one-time codes.
  • No phone number in the text to call back. Point members to the phone number printed on their card or shown in your app. A scammer can put any number in a text.
  • Say it out loud. Somewhere your members will see it, say: "We will never text you a link or ask for your password, PIN, or codes." Then keep that promise in every template.

What a good alert says

A fraud alert should fit in one text segment, 160 characters of plain text, so it arrives whole. Within that, it needs four things:

  1. Your name, first. The member should know who it's from before they read anything else.
  2. The specific transaction. Amount, merchant type, and when. Specifics are what make a real alert believable and a vague one alarming.
  3. A single question with a YES or NO answer.
  4. A STOP line.

Here is sample wording for a made-up credit union:

Example

Harbor CU fraud alert: Did you attempt a $1,240.00 purchase at an online electronics store? Reply YES or NO. Reply STOP to cancel.

And a version for a debit card, with a time:

Example

Harbor CU: Was your debit card ending 4471 used for $86.20 at a gas station at 6:42 p.m.? Reply YES or NO. Reply STOP to cancel.

Neither one says "urgent" or "your account will be locked," or sets a countdown. Pressure is a scam's tool, and members have learned that too.

What happens after YES or NO

The reply decides the next text, so write those templates in advance and have compliance approve them with the alert.

  • YES:

    Harbor CU: Thanks for confirming. No action needed. Reply STOP to cancel.

  • NO:

    Harbor CU: Thanks. We've blocked the card ending 4471. Call the phone number printed on your card or in our app to get a new one. Reply STOP to cancel.

  • Anything else, like "who is this?" or a question: route it to a person on your staff, not to another automated text. A member who's unsure needs a human answer, and a canned reply to "is this real?" is exactly what a scammer would send.

Consistency is part of the message

Members learn what normal looks like. If every fraud alert comes from the same registered number, uses the same opening, and follows the same pattern, a fake one stands out because it's different. If alerts arrive from three numbers in four styles, the fakes blend in.

That's an argument for a small set of approved templates and one sending number for alerts. It's also why wording drift is a real risk. A staff member who rewrites an alert to sound friendlier, or adds a link to help, undoes the pattern members rely on.

On timing and consent, the FCC has allowed financial institutions to send certain fraud alerts without prior consent, as long as the messages meet conditions such as being free to the recipient and short. Whether an alert may go out overnight is a policy choice, and your counsel should help set it.

How SndSMS handles it

In SndSMS, fraud alert templates are written once and approved by a named compliance reviewer, and staff can't change the wording when they send. Templates can be set to carry no link. Alerts go out from a number registered in your institution's name, and replies go to the shared inbox, where someone on your staff answers them. No AI writes or answers anything. Every alert, reply, and follow-up is archived.

Choose a planRead the blog