On October 13th, we’re joining Juniper Research, AT&T, and Verizon for a live conversation on mobile intelligence—what carriers and the companies that rely on their data actually know about a phone, and how that knowledge gets used to stop fraud before it happens. SIM swap detection is one of the clearest examples of carrier network intelligence in practice. A recent SIM change can be a meaningful fraud signal, especially when it happens just days before a login, account recovery, transaction, or new-account attempt. So ahead of the panel, we pulled back from any single customer and looked across nearly million SIM swap checks to see how this data actually behaves on our platform.
Across the customers we’ve featured in this series, SIM-change data turns out to be doing a lot of different jobs. A wealth manager checks it before trusting a phone with a high-net-worth client’s account. A global messaging platform checks it as part of verifying a new signup, at a scale of millions of numbers a month. A point-of-sale lender checks it alongside a handful of other signals before extending credit. An identity-intelligence vendor checks it on behalf of the banks and fintechs it serves, applying the same underlying fact into dozens of downstream risk decisions. Different industries, different stakes, same real-time question underneath all of them: has anything about this phone changed recently.
Our customers don’t all ask for this signal the same way, and the difference turns out to be a genuine design choice, not a technical detail. One group asks a direct, binary question: has this SIM changed within a specific window—typically three days—yes or no. If the answer is yes, that’s the trigger for a step-up check, full stop. It’s simple, fast to operationalize, and treats three days as the line between “ordinary” and “worth a second look.”
A second group asks a more open-ended question: how long has this SIM actually been established on this line—days, weeks, months? Rather than a yes/no against a preset threshold, they get back a tenure estimate and build their own risk logic on top of it, often working with our team to calibrate where their own thresholds should sit. Same underlying network fact, two different philosophies for turning it into a decision.
Across nearly a million samples of SIM swap checks drawn from recent production traffic, the two approaches land in strikingly similar territory. Among the threshold-based checks, under 2 percent came back positive for a change at all, and about two-thirds of those positives were caught inside the tightest, three-day window—not stale history. Among the tenure-based checks, independently, just under 3 percent came back with an estimated tenure of seven days or less—the same high-risk band, arrived at through a completely different question.
Two different ways of asking, two different customer philosophies, and the same basic shape in the answer: the overwhelming majority of phones show no recent change at all, and a small, low-single-digit minority show the kind of change that’s recent enough to matter. That consistency is itself a useful signal—it means the underlying fact (recent SIM change is rare, and rarer still in a way that’s actually fresh) isn’t an artifact of how any one customer happens to query it.
This entire picture is, at its root, a carrier-network fact—a SIM change is something that happens on AT&T’s network, or Verizon’s, or T-Mobile’s, before it ever reaches a risk model. Looking at that same traffic, those three carriers alone account for the overwhelming majority of every check we ran. That’s exactly why we wanted carriers in the room for this conversation rather than just the vendors and platforms consuming their data—the signal starts on their network, and understanding what “mobile intelligence” means has to start there too.
If you work in fraud, risk, or identity and want to hear carriers and vendors talk through what mobile intelligence actually looks like in production—not in theory—join us and Juniper Research on October 13th, with AT&T and Verizon on the panel. Registration link
Madhu Vudali is VP, Product Management at Zumigo. Comments or questions? Connect on LinkedIn: @madhuvudali