Your SMS provider says “delivered.” How many real people clicked?

Message delivery tracking for teams tired of fake delivery reports.

One link in your campaign. Every click is judged human, machine or unclear, and you read the evidence.

Start freeFirst 3,000 verdicts free. No card.

smstrace measures clicks, not delivery. A confirmed human click proves the message reached a person. Missing clicks alone don’t prove non-delivery — compare providers on the same audience or use control numbers.

Simulation: the clicks of one campaign, judged one by one

Simulation · sample data
Provider A reports
9,812 of 10,000 delivered 98.1%
Clicks, in order of arrivalhover a square, or use ← → keystap a square
humans
412
scanners
96
bots
143
unclear
51
clicks, as others count them
702
Click 702 of 702 T+25:36:03
iPhone · Safari
Mobile network · Munich, DE
  • HeadersHeader set matches Safari on iOS — points to a human
  • NetworkMobile carrier address — points to a human
  • DeviceApple GPU · 393×852 @3x · iPhone — points to a human
  • BehaviourTouch, then scroll — points to a human
Human97%judged by engine v2026.11.21

Same list, same text, same hour. Provider A: 41 confirmed humans per 1,000 sent. Provider B: 12. That gap is a question to ask, not a verdict.

smstrace measures clicks, not delivery. A confirmed human click proves the message reached a person. Missing clicks alone don’t prove non-delivery — compare providers on the same audience or use control numbers.

What a link sees about you, right now.

Six of the signals a verdict is built from, read in your browser by this panel a moment ago. The panel sends nothing.

No verdict is shown here. The engine runs on our server and also weighs what a browser can’t see about itself: request headers and the network a click comes from.

GPU
—
A phone by name with a desktop GPU is an emulator. A software renderer points to a server.
Screen
—
Headless browsers keep telltale default window sizes.
Time zone
—
Checked against the country of your address. A mismatch reads as a VPN, which means a person.
Languages
—
An empty list points to a script.
Touch
—
Declared for free, so it proves nothing. Only a performed touch counts.
First action
—
The strongest evidence there is. An action can’t be declared, only performed.

“Delivered” is a status. It isn’t proof.

The report is written by the party you pay for delivery. Clicks alone don’t settle it either.

“A Fake DLR shows that an SMS message has been delivered correctly, while the opposite is true.”
CM.com (opens in a new tab)
“Some scanners, such as carrier security systems, open every shortened link as soon as a message is delivered and use the same user agents as regular web browsers.”
Braze, bot click filtering (opens in a new tab)

How it works

  1. Put a link in the message

    One smstrace link per provider or campaign, in the SMS you already send. Nothing to integrate. For a first test we suggest about 1,000 contacts per provider.

  2. Every click is judged

    In a fraction of a second, before the recipient reaches your page: human, machine or unclear, with the evidence and the version of the rules on record.

  3. Read the report

    Confirmed humans per 1,000 sent for each provider, with the range the number can move in. Machines, unclear clicks and clicks with no verdict are counted apart. Open any click to see why, or export the CSV.

What it won’t claim

  • It measures clicks, not delivery.

    A confirmed human click proves a message arrived. A missing click proves nothing by itself: the offer may simply be weak.

  • It doesn’t call anyone a fraud.

    A gap between providers is a reason to ask for logs. Most machine clicks are innocent: link previews and security scanners.

  • It doesn’t catch every fake.

    Data-center requests, headless browsers and emulated phones are caught. A farm of real phones can pass.

    But real phones, SIM cards and hands to tap them cost money on every click. That eats exactly what faking a delivery was meant to save.

Questions a careful buyer asks

You see clicks, not delivery. What do I learn?

Every confirmed human is a message that certainly arrived: a floor under real delivery. The rest comes from comparison. One audience, one text, two providers, and a gap in human clicks can’t be explained by the offer.

My platform already filters bot clicks.

That is the counter of the party you are checking, and it usually subtracts known bots by a list. Here a click counts as human only when there is evidence for it, and you can read that evidence.

Will the extra hop hurt conversion or delivery?

The check takes a fraction of a second. Some carriers filter shared short domains, especially in the US, so start with part of your list and watch the numbers.

What do you collect about my recipients?

No cookies, no browser storage. Device parameters and the IP address, for one purpose: telling a person from a machine. In the EU fingerprinting falls under consent rules, so check what your privacy notice needs to say; we don’t provide a ready legal package yet.

How mature is it?

Early. Campaigns, the provider comparison and the detection engine are in production, and every change to the engine’s rules is published in your dashboard. Not there yet: your own link domain, control numbers, a link per recipient. Accuracy on large SMS campaigns is still being measured with the first teams.

Send your next campaign through a link you can check.

First 3,000 verdicts
free
After that, unlimited
$90 a month

A verdict is every judged click on your link: human, machine or unclear. Not a message sent.

Start freeNo card needed to start.
smstrace — proof your SMS reached real people