Why a New York Mobile IP Can Show a High Risk Score
A customer buys a Verizon line, runs the address through a fraud-score site, sees a red number, and messages us asking why they were sold a bad IP. We understand the reaction. The explanation is not a dodge, it is how these scoring services work and what a carrier address is. Read this before you read any score.
What the score is built from
Services like IPQualityScore, Scamalytics and the many clones take an address and return a risk number. The inputs are a mix: whether the address is known to belong to a hosting provider or a VPN, whether abuse reports have named it, how many distinct users have been seen behind it, whether it was recently seen in a proxy list, and the behaviour of previous sessions from it that the service happened to observe through its customers' sites.
A carrier address sits behind carrier-grade NAT. Thousands of real subscribers in New York share it over a day. Among thousands of people, some will do something a site reports. The address inherits that history. That is not your history, and it is not ours, but the score cannot tell the difference, so it reports what the pool as a whole has done.
Why shared is also why trusted
Here is the piece people miss. The same sharing that drags the score up is what makes real sites reluctant to act on it. A platform that blocked every carrier address with a high score would lock out a large slice of New York phone users, including its own paying customers. So the big platforms treat carrier addresses as low-information: they look past the IP to the account, the device and the behaviour. The scoring site is answering a narrower question than the platform asks.
We have watched lines with ugly scores work flawlessly for months on account work, and lines with clean scores get an account flagged because the account itself behaved badly. The number is a weak predictor of the thing you actually care about.
Scores move with the pool
Because the address is drawn from a regional pool and shared, the score you see today is a snapshot. Rotate and you will see a different number. Come back tomorrow on the same address and it may have changed again as reports age out and new ones arrive. Some scoring services also treat any address in a mobile block as higher risk by default, simply because the category is harder to attribute, which pushes every carrier address up a notch before any behaviour is counted.
When the score matters, and when it does not
The score matters on sites that literally buy a feed from one of these services and act on it at a threshold. Those exist, mostly small shops and some signup flows. It does not matter on the large platforms with their own risk systems, which is most of where our customers work. If you are dealing with a site in the first group, rotate until you land on an address with a lower number, then stay sticky on it; a Verizon line in Brooklyn will usually hold that address for a long time.
If you are dealing with the second group, stop reading scores and fix the things the platform actually evaluates: a coherent browser profile set to New York time and en-US, a device fingerprint that matches your host, a realistic pace, and an account history that makes sense.
- Small sites with a bought risk feed: rotate to a lower score, then stay put
- Large platforms: ignore the score, fix fingerprint, locale and behaviour
- Any site: never judge a line by one number on one day
What we do on our side
We monitor every line for things we can control: that the modem holds a carrier session, that speed is in the range we quote, that the address resolves to the carrier and to New York. We do not and cannot control the reputation of a shared address, and anyone telling you they sell carrier addresses with guaranteed low scores is either rotating for you until they find one or not telling the truth. What we can do is move you to another line or carrier for free if a specific address is giving you trouble on a specific site. Message support with the address and the site and we will sort it.
A quick way to check reality instead of the score
Take the thing you actually plan to do and do a small, legitimate version of it through the line: load the site, create nothing, log in to an account you own, read a few pages. If that works without challenges, the score was noise. If it throws challenges, look at the browser and account first, then try one rotation, then write to us. In our experience the order of likely causes is exactly that order.
Frequently asked
Can you give me a line with a score under a certain number?
No, and we would be wary of anyone who says yes. Scores change hour to hour on shared addresses. We can give you a line that works for your job, and move you free if it does not.
Does 5G score differently from 4G?
Not in any consistent way. Both draw from the carrier's metro pool and inherit whatever the pool has done. The generation changes speed, not reputation.
Why does one checker say low risk and another say high for the same address?
They use different inputs and different weights. One may count a two-year-old abuse report; another may ignore anything older than a month. Disagreement between checkers is itself a sign the number is not measuring anything solid about you.
Is a high score ever caused by the previous customer on my line?
Rarely, because the address was shared with thousands of others and the previous customer's share of it is tiny. If a particular site remembers the address itself, one rotation resolves it.
