Why alert texts get blocked: spam words, links and other triggers
Mobile networks scan business texts for signs of spam before they reach a phone. A real alarm can get blocked when it holds a word scammers use, a shortened link or a wall of capital letters. Most fixes are small: reword the alarm template, trim the text and send from a number only your company uses.
7-day free trial, 15 message creditsCard required for verificationNo charge for 7 daysEmail reminder before billing
Published:
The problem
Refused as possible spam
- The alarm says a door is locked and a filter is blocked
- A shortened tracking link sits in the footer
- The text goes out from a number many senders share
The fix
Delivered to the on-call phone
- Plain words in the alarm template
- One short text with no footer link
- A dedicated alert number for your company
- Carriers filter texts by words, links, capitals and the sending number's history.
- Ordinary alarm words like locked, blocked, claim and vape can look like spam.
- Fix the wording in the alarm system's template, not on each phone.
- Welco holds texts with blocked phrases before sending, and a held text uses no credits.
- Welco delivery reports tell you in plain English when a network refused a text.
Why do carriers block alert texts?
Carriers block texts that look like spam to protect their customers from scams. Every business text passes through automatic filters on the mobile network before it reaches a phone. This is called carrier filtering, and the filters can't tell a real alarm from a fake one by intent. They only see the words, the links and the number that sent it.
The wireless industry writes this right into its rules. The CTIA Messaging Principles and Best Practices (May 2023, read October 6, 2026) say in section 7.2.3 that providers may block or filter traffic. They don't have to tell the sender what tripped the filter.
When a filter refuses a text, the network sends back a short note to the sender. In plain words, it says the network refused the text as possible spam. The person on call sees nothing at all, a silent failure our other alert texting guides also cover. That silence is the hard part, because nobody knows an alarm went missing.
Words that scam texts also use
Scam texts lean on fear and money. They say an account is locked, a payment is blocked or a prize is ready to claim. Your alarm might use the same words for honest reasons, and the filter can't see the difference.
Links and link shorteners
Short links hide where they lead, so filters treat them with suspicion. Many alert emails add a tracking link or a portal link in the footer. If that link ends up in the text, it raises the risk of a block.
The same CTIA document, in section 5.3.2, asks senders who shorten links to use a shortener dedicated to them alone. A public shortener shared with millions of strangers is the opposite of that.
Bursts of capital letters and symbols
Alarm systems love capitals, like ALARM ALARM ALARM or *** CRITICAL ***. Long runs of caps and repeated symbols look like shouting, which is common in spam. One word in caps is fine. A whole text in caps is a risk.
A sending number with a poor history
Filters also judge the number a text comes from. Some services send texts for many companies from the same few numbers. If one of those senders behaves badly, every company on that number can be filtered with it.
Which alarm words look like spam to a carrier filter?
Carriers don't publish their word lists, so treat these as words to watch, not a full list.
Change the word in the alarm system's own message template so every alert uses it.
Vape
- Where it shows up in alerts
- School restroom air sensors
- A plainer way to say it
- Air sensor alert
Locked, locked out
- Where it shows up in alerts
- Door controllers, tenant access systems
- A plainer way to say it
- Door secured, no entry
Blocked
- Where it shows up in alerts
- HVAC filters, drains, intake screens
- A plainer way to say it
- Clogged, restricted flow
Claim
- Where it shows up in alerts
- Insurance batch jobs and file transfers
- A plainer way to say it
- Batch file, job run
Free, winner, prize
- Where it shows up in alerts
- Disk space checks, contest systems
- A plainer way to say it
- Space left, result
Does Welco block words in alert texts too?
Yes, but before sending, not after. Welco holds any outgoing text that contains a blocked word or phrase, so it never reaches a mobile network. The full list sits on our page of words and phrases Welco holds.
The list covers sales lines, adult content, hate, loan and crypto pitches, and phishing phrases. It also covers public link shorteners, card numbers and passwords. Welco looks for phrases, so an honest alarm word on its own is usually fine.
What happens to a held alert
A held text is not sent, your alert feed shows the reason, and it uses no message credits. Fix the wording in the alarm template and the next alarm goes out normally.
Held by Welco versus refused by a carrier
On the phone, both look the same: nothing arrives.
A held text stops at Welco and is free. A text the network refuses was already handed over, so it uses its credits, as our delivery rates disclosure explains. Your delivery report names which of the two happened.
Stop your alert texts being blocked in 6 steps
Work through these in order. The first step tells you which alerts to fix, so you don't guess.
Find the texts that were refused
Open your Welco delivery reports and look for texts the network refused as possible spam. Note which system sent each one and what the text said.
Reword the alarm template at the source
Most alarm systems let you edit the message, the point name or the tag description. Swap risky words for plain ones there. Every future alert then goes out with the safer wording, for every phone.
- Change the alarm description, not just one email. The next alarm will use the old words otherwise.
- Keep the meaning clear.
Clogged filtertells the tech the same thing asblocked filter.
Take links and shorteners out of the text
If an alert email carries a portal link or a short link, keep it in email only. The person on call needs the site, the problem and the time. They can open the full email later. Welco holds any text with a public link shortener such as
bit.ly, so it never reaches the network.Trim the alert down to one text
Use one-text alerts to keep the part that matters and drop footers, banners and legal notes. Less text means fewer chances to trip a filter. You can also trim alerts to one text in the alarm template.
Send from a dedicated alert number
An alert number your company alone uses keeps your sending record separate. Its history is yours alone, so another sender's behavior can't drag it down. Welco uses alerts-only numbers that marketing senders never share.
Test the new wording before a real alarm
Turn on test mode to check the reworded alert in the feed without sending a text. Then send one real test to a phone and confirm it arrived.
Get your alarm texts back on the phone
Sign up, verify your business and point one alert email at your Welco address.
7-day free trial, 15 message creditsCard required for verificationNo charge for 7 daysEmail reminder before billing
What a reworded alert looks like as a text
This school air sensor used to send VAPE DETECTED with a tracking link. The facilities team changed the event name in the sensor's settings to plain words. Welco then trimmed the email to one text.
Alert email from alerts@sensors.example.org, subject “Air quality event: Building B, Restroom 2”. Welco trims it to one text of 96 characters: “Welco: [Building B] Air quality sensor alert in Restroom 2 at 10:14 AM. Reply STOP to opt out.”
Text received: Welco: [Building B] Air quality sensor alert in Restroom 2 at 10:14 AM. Reply STOP to opt out.
Characters343961 credit
The text keeps the place, the sensor event and the time. The link and the footer stay in email.
What else can get an alarm text blocked?
Sometimes the words are fine and the pattern is the problem. Filters watch how a number behaves, not just what each text says.
- Floods of the same text. A tripped breaker can fire the same alarm dozens of times, which is what an alarm flood looks like. Hundreds of near-identical texts in an hour can look like bulk spam. Alarm flood control limits the burst, and the alarm flood guide shows how to cut it at the source.
- Duplicates from your mail server. Two forwarding rules can send one alarm twice. Clean up the forwarding so each alarm leaves once.
- Texts nobody agreed to get. Welco only texts phones that replied YES first, which is exactly what networks want to see.
Mistakes to avoid after a blocked alert
Don't send a refused alert again unchanged. A filter that refused it once will usually refuse the copy too.
- Resending the same text by hand. Repeats of a refused text can look even more like spam.
- Spreading alerts across many numbers. CTIA section 5.5.2 names this snowshoe messaging and asks senders not to do it.
- Swapping letters to dodge a filter. Writing
l0ckedorv@pelooks like a scam trick and makes the problem worse. - Fixing one phone's copy. The wording lives in the alarm template, so fix it there for everyone.
- Guessing without the report. Check which carrier refused it before you reword anything.
Check your alarm template before you save it
Run each alert template through this list. It takes a few minutes per system.
0 of 7 done
Frequently asked questions
Which words get alarm texts flagged as spam by carriers?
Can a link inside an alarm text cause a spam block?
How do I know a carrier refused one of my alert texts?
Does a dedicated alert number stop spam filtering?
Why did some phones get the alert while others didn't?
Do I pay for an alert that Welco holds for a blocked phrase?
Is the word free in a disk space alarm a problem?
free gift, not the single word, so 12% free on /var goes out. Carrier filters are harder to predict. If a carrier refuses a disk alarm, switch to 12% left or space left in the monitoring template.Reviewed by Rakesh Patel, Founder, Welco
Send alarm texts that read like alarms, not spam
Reword once at the source and let Welco send each alert as one plain text.
7-day free trial, 15 message creditsCard required for verificationNo charge for 7 daysEmail reminder before billing