Comparison
How reverse phone lookup really works, and where the data comes from
| Source | What it supplies | How fresh it tends to be | Where it fails |
|---|---|---|---|
| Numbering plan records | Area code, exchange, region, sometimes the original carrier or rate center | Kept current by the plan administrator | Never names a person |
| Carrier subscriber files | The real account holder | Current, inside the carrier | Private. Lookup sites do not have them |
| Public records | Deeds, court filings, some voter files, business registrations, licenses | As old as the filing | Only some records carry a phone number |
| Marketing and retail data | What you typed on a warranty card, contest entry or loyalty form | Unknown, because sites rarely show an entry's age | Fake numbers on purpose, shared family plans |
| Telephone directories | Old white pages and business listings | Often years old | Skews to landlines and to people who wanted to be listed |
| Web scraping | Social profiles, classified ads, small business pages, forum posts | As old as the page | Anyone can type a fake name next to a number |
| Licensed broker feeds | Most of what many lookup sites actually show | As old as the supplier's last refresh | Same data under different brand names |
That is the whole pipeline in one table, and the rows are worth reading in order, because the first two are the ones people assume are behind the search box. They are not. I spent a few weeks reading broker privacy notices, regulator pages and the fine print on a dozen lookup sites, and the picture was more mundane than the advertising.
Row by row
Numbering plan records. This is the one genuinely official layer. The North American Numbering Plan Administrator hands blocks of numbers to carriers, and the public records at nationalnanpa.com tell you which area codes exist and where they apply. From there you can say a line was first set up in a certain region. You cannot say who holds it now. Our number reader works at exactly this layer and stops there on purpose.
Carrier subscriber files. After a block is handed out, each carrier assigns single lines to its own customers and keeps those records private. A carrier deals with the account holder, a court order or a law enforcement request, and nobody else. So when a website shows a name beside your number, it did not ask the phone company. It matched the digits against a file that somebody compiled for another reason. The limits around this are the subject of our piece on who can find a cell number's owner.
Public records and directories. Property deeds, court filings, voter files in states that publish them, business registrations and professional licenses all turn up in lookup files, but only the ones that happen to include a phone number. Old white pages are the same story: useful for landlines and for people who chose to be listed, thin for everyone else.
Marketing and retail data. Information you gave a store, a warranty card or a contest entry can be sold or licensed later. People also give retailers a made-up number on purpose, and families share plans, so the bill may be in a parent's name. Matching on a field like that is shaky from the start.
Web scraping. Social profiles, classified ads, small business pages and forum posts show up wherever someone typed a number next to a name. The scraper has no way to tell a real name from a joke.
Licensed broker feeds. This is the row that matters most. Many lookup sites own very little data. They license a feed from a larger broker, which is why different brands show near-identical reports. I would not read much into a claim of "billions of records." A billion records can still hold a single stale entry for your number, and the size of the database says little about when your number was last confirmed against a living person. If you want the longer story of why two sites then disagree, see why one number gets several names.
What the matrix cannot show
A table of sources hides the way data ages. Phone numbers get recycled: when someone cancels a line, the carrier eventually reissues it, sometimes after a quiet period of a few months. The old owner's name can sit in broker files long after that. Cell numbers are the hardest case. Landlines once appeared in a phone book next to an address, but mobile numbers have no equivalent, and prepaid lines often have no named subscriber in any commercial file. If an unknown caller uses a prepaid phone or an internet-based number, the best a lookup can usually do is guess the region.
Two things do hold up without a paid report. The area code and exchange point to a place. And certain codes have reliable meanings: toll-free codes such as 800, 888, 877, 866, 855, 844 and 833 do not map to a city, a 900 number is premium-rate, and several Caribbean codes, for example 876 for Jamaica, look domestic but are international calls.
It is also worth separating lookup from spam labels, because people mix them up. "Scam Likely" from T-Mobile, "Spam Risk" from Verizon's call filter and "Potential Spam" in Google's Phone app are produced by carriers or apps from calling patterns and user reports. They are estimates. A real business can be flagged after a burst of short calls, and a scammer on a fresh spoofed number may carry no flag at all. The STIR/SHAKEN framework, which carriers must now use under the TRACED Act of 2019, signs caller ID information so that spoofing is harder to pass off, as the FCC's consumer pages explain. A signed call can still be dishonest. The signature says the originating carrier probably authorized the number, not that the caller is telling the truth.
Run the matrix on your own numbers
You do not have to take any of this on faith. Pick three numbers whose owners you know for certain: your own cell, a friend who agreed to be a guinea pig, and a business that prints its number on its door. Run each through a free lookup and a second site, then write down what each returned and how close it came. In my experience the business number does best, because companies publish their numbers on purpose. The personal cell does worse, and a prepaid line or a number that changed hands recently does worst of all.
The exercise teaches one more habit. Data ages unevenly, and a record built from a store sign-up three years ago can look fresh on the page, because sites rarely show the age of each entry. Ask of any report when the line was last confirmed. If the site cannot say, assume the answer is long ago, and read our glossary of report fields before you act on anything in it. If you find your own number in those files and mind it, the routine for getting it removed is free, and the comparison of free and paid lookups explains why the paid reports mostly repackage the same feeds.