You open your bank’s website and type your password. Between your computer and the bank, that information has to travel across a network. How can it make the journey without becoming readable to someone watching the connection?
The protection behind an HTTPS address is designed for that problem. It helps your browser check the service it is connecting to, keeps the conversation private on the way, and detects interference with the protected messages. Understanding that job also explains why an encrypted website can still be dishonest or badly run. 18
The beginning of a web address tells the browser how to communicate. HTTP is the ordinary set of rules for requesting web pages and receiving replies. HTTPS uses that web exchange through a protected connection. The extra letter does not mean the page has been inspected for honesty; it identifies a protected way of carrying the conversation. A page can look identical in either case, because appearance and transport protection are different things. 18
First, a private conversation
A browser is the program you use to open websites. At the other end is a server: a computer system that responds to requests. When you ask for a page, the server sends information back. When you submit a form, information travels in the other direction. With HTTPS, this web conversation uses a protection system called Transport Layer Security, usually shortened to TLS. 18
Before your password is sent, the browser and server exchange setup messages. The browser says which protection methods it understands; the server selects a compatible method. They establish secret information that will be used to protect the conversation. This setup is called a handshake. It happens through messages between programs, not through anything you have to type. In a normal first connection to a public HTTPS website, it also includes checking the server’s identity. 18
What does “encrypt” mean here? Your password is represented as computer data. The browser transforms that readable data into protected data before sending it. The server uses the corresponding secret information to recover the original. That secret information is called a key: it is a value used by the mathematical protection method, not a metal object and not the password you chose for your account. A network observer may see data travelling but should not be able to recover the password from the protected contents. This is more than hiding letters on your screen. The protection applies to what is transmitted. 18
The browser and server must establish secret information without announcing it to everyone watching the network. In one common modern method, each program keeps a private mathematical input and sends a related public value. The browser combines the server’s public value with its own private input. The server does the corresponding calculation with the browser’s public value and its own private input. Their results agree, although neither program sent its private input.
An observer can see the exchanged public values but lacks the private inputs needed to recover that shared result. The programs use the established secret information to derive the keys for their conversation. So seeing the setup messages is not the same as knowing the keys that protect later messages. You do not choose or email these keys; the programs do the work. 18
The server also sends a certificate: a digital document linking the website name to public information used to check its identity. The server proves possession of a corresponding private key, which is kept secret. The browser checks that proof and the certificate’s acceptability. This prevents an observer from passing the identity check merely by copying the publicly available certificate. The certificate-related key has an identity-checking job; the separate conversation keys protect the messages exchanged in the connection. 18
A certificate is not accepted just because the server calls it genuine. An issuing organisation, called a certificate authority, signs its information. A digital signature is a mathematical result that software can check against the signed information using a public key.
The browser first checks the website certificate using its issuer’s public key. It also checks any supporting issuer certificates back to an authority already accepted in its trust settings. That starting authority is not accepted merely because the website says whom to trust.
These issuer checks support the certificate’s statement about the website name. The server’s separate proof that it has the corresponding private key checks the system answering now. Both are needed: a signed document alone does not show that the current server holds the secret belonging with it. 5 18
After successful setup, you submit the login form. The browser protects the request; the receiving system recovers it and checks your account password. The server’s answer is protected on the way back, and the browser recovers it to display the result. Notice the two identity questions: the connection setup checks the service, while the login checks your account. A certificate does not sign you into the bank. 18
Protection also checks whether received messages are genuine for that connection. If someone changes the protected data on the way, the receiving side should reject it rather than quietly treat the altered contents as your original request. This is why TLS protects against both reading and undetected tampering. It cannot stop someone cutting off the connection altogether; an interrupted journey and a secretly modified message are different problems. 18
Think of a protected delivery route with an identity check at the destination. Protecting the route and checking the destination are both important. A private route to an impostor would not solve the problem. The browser therefore checks whether the certificate fits the website name it is trying to reach, as well as the certificate’s trust conditions. 3
That last detail is easy to miss. The browser checks the name in the address, not the company you had in mind. If you follow a misleading link to a different domain, a valid certificate for that different domain does not correct your choice. Encryption can work perfectly while you are talking to the wrong business. 3
What stays protected, and what does not
During the protected conversation, someone observing the network should not be able to read the contents or secretly alter them without detection. The systems at either end can still read the information: otherwise the bank could not process your request and your browser could not display the answer. TLS also does not hide every feature of the traffic, such as the length of transmitted data. 18
Return to the delivery analogy. A protected route does not determine what happens to a document after arrival. Similarly, TLS does not decide which employee may open your invoice, how a saved copy is protected, or whether a backup can be restored. Those are separate tasks for the service receiving the data. This is a limit of what TLS is designed to do, not a reason to stop using it. 18
For example, an invoice portal could use a properly protected connection but mistakenly let one customer view another customer’s documents. The connection and the permission check are different mechanisms. The first asks whether communication is protected; the second asks whether this person may receive this document. A good answer to one cannot stand in for an answer to the other. 18
There may also be more than one connection along the route. A provider might have an intermediary receive the browser’s connection and forward the request to another computer. The browser has checked its connection to that intermediary. It has not independently checked the separate onward connection. Ask where the protected connection ends if that distinction matters to the service you are buying. 18
Why people still say “SSL”
A hosting offer may call this an “SSL certificate.” SSL is the older technology whose name has survived in everyday sales language. Modern secure web connections use TLS. The wording in an offer is therefore not enough to tell you which technology a service actually accepts. Security guidance rules out obsolete SSL versions and TLS 1.0 and 1.1. 2
Reports may mention TLS 1.2 or TLS 1.3: these are different versions of the connection technology. For newly designed sets of communication rules, called protocols, current guidance requires TLS 1.3 as the supported default and permits TLS 1.2 as an additional, non-default option for deployment reasons. A version number tells you about the technology in use, not whether someone has broken into the service. 16
Use the right address, then ask the right question
For everyday work, use a service address you have confirmed through an established contact or saved as a trusted bookmark. If a certificate warning appears, stop rather than treating it as an inconvenience to click past. Report the address and warning to the operator through a contact route you already know. The warning needs an explanation; it does not, on its own, identify the cause.
When buying a service, one useful question is: “Does this answer concern our login portal, or only your public website?” An invoicing provider can use different systems for its brochure website, its customer accounts and its invoice delivery. A working connection to one does not measure the others. That follows from the connection-by-connection nature of TLS. 18
HTTPS is valuable because it does a specific job. Keep using it, check where you are connecting, and leave questions about stored documents and access permissions on the agenda after the connection has been secured.
Sources
[2] IETF RFC 9325 — Secure TLS Recommendations
[3] IETF RFC 9525 — Service Identity in TLS
[16] IETF RFC 9852 — New Protocols Using TLS Must Require TLS 1.3 (July 2026)
[18] IETF RFC 9846 — TLS 1.3 (July 2026; obsoletes RFC 8446)