A reliable, low-cost alert system for teams without an on-call platform

Somebody has to be woken up. That is the whole requirement, and it is the one most alerting setups quietly fail.

A two-person IT shop, an MSP with one engineer covering nights, a facilities manager with a walk-in freezer: all of them already own something that notices when things go wrong. Nagios notices. The NAS notices. The temperature sensor notices. Each of them sends an email, and at 3 a.m. an email is a notification to nobody.

The usual answers do not fit a team this size. Tools like PagerDuty and Opsgenie are built around rotations, incident timelines and per-user seats, which is a lot of machinery when the same person is on call every night. (The PagerDuty alternative page goes through that comparison, and what PagerDuty costs per user, tier by tier sets out the published figures with the date they were checked.) Texting has the opposite problem. It is simple, but the carrier email-to-text gateways that monitoring tools used to send through have been closing one by one, and a text that does arrive can still sit unread under a Do Not Disturb setting.

What a small team needs is narrower than an incident platform and more dependable than a text.

What an alert channel has to do

1. Reach a human. Not an inbox, not a lock screen. A person.
2. Confirm that it did. A delivery receipt says the phone got it. It does not say anyone did.
3. Try again if nobody responded. On its own, with no second system watching the first.
4. Cope with a burst. When a switch drops, forty things behind it go down at once. Forty alerts should not become forty interruptions.
5. Leave a record. Afterwards someone will ask what went out and when.
6. Cost what a small team can justify. Usually that means no per-user pricing and no equipment.

A phone call that reads the email out loud meets all six, if the service behind it is built to. Here is how Email2Phone does it, requirement by requirement, using the mechanics of the product rather than the brochure.

Reaching a human: the call

Email2Phone gives your account an address, yourname@email2phone.net. Anything that can send an email can send to it: a monitoring tool’s notification contact, a NAS, a UPS, a sensor gateway, or a mail flow rule in Microsoft 365. If the alerts already land in an existing mailbox, a forwarding rule sends the matching ones on to your Email2Phone address. The email is converted to speech and the phone rings — mobile, landline, desk phone or VoIP. There is no app to install and no email-to-text carrier gateway in the path.

A ringing phone wakes people in a way a badge count does not, and a landline on the nightstand does not depend on a mobile signal.

Confirming it: mandatory acknowledgement

A message is not counted as delivered until someone presses a key. Listening is not enough. The listener has to press 5 (reply), 7 (delete) or 8 (save) on the keypad, and until they do the message stays unacknowledged. That is the difference between “the phone rang” and “a person heard this and dealt with it”, and it is the feature most alerting channels lack.

Trying again: the callback ladder

On a phone with voicemail switched off, Email2Phone calls back on its own at 5, 15, 30, 60 and once more at 60 minutes. Five attempts over about three hours, with no second system needed to notice that the first call was ignored. A message that was never acknowledged is filed in the account’s online Message History, so nothing is lost even if every call is missed.

Ringing more than one number

Most small teams have two or three numbers that could take the call. Email2Phone dials them either one after another or all at once, and that is an account setting. A second setting decides what happens once someone acknowledges: stop there, or carry on until every number has been called.

That is dialling behaviour, not an on-call system. There are no rotations, no schedules of who is primary this week, and no handover. Two numbers and a rule about when to stop is the whole of it — which is the point for a team where the same person is on call every night anyway.

Coping with a burst: one call, not eighty

This is the part that separates a mailbox from a dialer. Email2Phone is a mailbox. When alerts arrive, the call announces how many unacknowledged messages are waiting, “you have 4 messages”, or 80, and you hear them one after another and clear each from the keypad in a single pickup. If more arrive while you are listening, the next call carries the whole backlog rather than starting a fresh call per email.

So a burst of 80 alert emails produces one phone call, not eighty. Services that place one call per message do the opposite, and a flood at 3 a.m. is when you least want your phone behaving that way. There is manual flood control from the handset too: press 9 then 1 to block a sender (Email2Phone creates the filter for you), or 9 then 3 to delete every unread message from that sender.

Deciding what gets through: filters

Not every email deserves a call, and not every call should go to the same person. Filters handle both. Each filter has one condition, on From, To, CC, Subject or Body, using equals, contains, begins, ends or matches, and one action: ignore the message, or call one or more of the phones you have configured. Anything no filter excludes goes to the default phone number.

Two examples that cover most small-team setups:

• Only the serious ones. A filter with Subject contains RECOVERY, set to ignore, means the all-clear emails never ring anyone. The CRITICAL and DOWN notifications fall through to the default number.
• Route by site or by person. Subject contains FREEZER-2, set to call the facilities phone and the manager’s mobile. Subject contains WEB-, set to call the web team’s phone. Put the keyword in the monitoring tool’s subject line and the routing takes care of itself.

Filters double as a second layer of spam control on top of the account’s Spam Protection setting, since a filter can ignore mail from a sender, or with a subject, that you know to be noise.

Quiet hours and planned downtime

Schedules attach to a phone number, not to a filter. A schedule says when a particular phone should not be called: the desk phone outside office hours, the personal mobile during a week off. Because routing is per filter and quiet windows are per phone, the same alert can reach the desk phone by day and the mobile overnight, with no scheduling platform in between. A quiet window drops what arrives during it rather than holding it for later, so put one on a phone that has another number behind it.

For maintenance windows there is Suspended Delivery, an account switch that stops every call until you turn it back on. Nobody gets rung about a server you are rebooting on purpose. Nothing is held while it is on, so messages that arrive during the window are not delivered once you switch it off.

Leaving a record

A message nobody acknowledged is filed in the account’s online Message History, and anything the Spam Protection setting stops is filed there under SPAM, so the two categories you are most likely to go looking for afterwards — what was missed, and what was blocked — are both recorded. Any past message can also be replayed from any phone by dialling in; the access numbers in the US, Canada and the UK are free on every plan.

The listener can also answer from the keypad. Press 5, speak after the tone, press #, and the sender receives an email with a link to play the recording. For “on it, twenty minutes” that is faster than finding a laptop.

Two setups, end to end

The pieces above are easier to judge assembled. Both of these take about twenty minutes, most of it inside the tool that already sends the email.

An MSP with one engineer covering nights

The monitoring tool already emails a shared mailbox that nobody watches after six. Add the Email2Phone address as a second notification contact and attach it to the states worth waking someone for — DOWN and CRITICAL, not WARNING, and not the recovery notices. In Email2Phone, set the engineer’s mobile as the default number, add the owner’s mobile as a second, and set dialling to sequential: the engineer first, the owner if the engineer does not acknowledge.

Then one filter, Subject contains RECOVERY, set to ignore, so the all-clears stay silent. And a schedule on the owner’s number so it is never called before the engineer has had the first two callbacks.

What happens at 03:14 when a host drops: the phone rings, a voice reads the subject and the first part of the body, the engineer presses 8 to save it and gets up. Nothing else in the chain had to be bought, and if the engineer sleeps through it, the owner’s phone rings at the third callback.

A facilities manager with a walk-in freezer

The sensor gateway emails when a probe crosses a threshold. Point it at the Email2Phone address, put the freezer’s name in the subject line — FREEZER-2 — and add two numbers: the site phone and the manager’s mobile, dialled simultaneously, because at −18 °C the first person to pick up is the one who matters.

One filter, Subject contains FREEZER-2, calling both phones. A second sensor gets its own keyword and its own filter, and the routing stays readable a year later when somebody else inherits it.

The part that matters here is the acknowledgement rule. A freezer alarm that rings once and gets missed is a five-figure loss; a freezer alarm that keeps calling back for three hours until a person presses a key is a bad evening. That difference is the entire argument for this kind of channel, and it is why “did the message get delivered” is the wrong question and “did a human confirm it” is the right one.

What it costs

Plans start at $10 a month for 40 call credits, then $100 for 500 and $400 for 3,000. One credit is up to 60 seconds of a call to a US or Canadian number; other countries run 1 to 6 credits a minute. Credits beyond the plan cost $0.25, $0.20 or $0.15 each depending on the tier. No setup fee, no contract, no per-user pricing, no hardware. Full details are on the pricing page.

The mailbox model makes the arithmetic kinder than it first looks: a bad night with 80 alerts is one call billed by its length, not 80 calls.

What Email2Phone does not do

Voice calls only. No SMS, no MMS, no fax. No API and no webhook: the integration is email, which is exactly why the list of things that can trigger it is so long. No mobile app. No on-call rotation management, no incident timeline, no MTTA or MTTR reporting. If your team needs those, an incident platform is the right purchase and Email2Phone is not a substitute for it. If what you need is a phone that rings, and keeps ringing, until a person confirms they heard the alert, that is the whole product.

Trying it

Email2Phone has been running since 2008. The guide to how email to phone call alerts work covers the mechanics in more depth, including what the call sounds like and how a call compares with SMS and push for critical alerts. The free trial runs 30 days, delivers up to three emails a day as calls, includes dial-in, and does not ask for a credit card. Point your monitoring tool’s email notification at the address and let the phone ring.