
Web2 vs Web3: what actually differs, and what it changes on screen
Imran KhanNovember 20, 202616 min read

table of contents
Web2 is the internet you use every day: a company runs the service, holds the accounts and sets the rules. Web3 is a different way to build some of those services, where parts of the job are handed to a blockchain that no single company controls.
We're a design studio, not a blockchain lab. We haven't designed a wallet or an exchange. We have designed Reputable, a product for Web3 developers, and Fedica and Dino, which are ordinary Web2 products, so we can show both sides from the design desk. Everything about outside tools and pages is attributed to the page it came from, and each was read on 20 November 2026. Nothing here is financial or legal advice.
The short version
- Web2 means a company runs the service. It owns the servers, the database and the accounts, and it can change the rules or reset your password.
- Web3 means some of the service runs on a public blockchain. You prove who you are with a wallet's key instead of a password, and some records live on a shared ledger instead of one company's database.
- The labels are informal. PIA says Web1, Web2 and Web3 "aren't official versions with fixed release dates", and the boundaries are approximate.
- Most Web3 products are Web2 around a chain. The website, the servers that read the chain and the support team are usually ordinary. PIA lists five common centralised parts.
- Neither is better by default. Each trades speed, cost, simplicity and recovery against openness, ownership and censorship resistance.
- For a builder, the real difference lands on screen. Sign-in, recovery, waiting and mistakes that can't be undone all work differently.
Where the names come from
The numbering is a story about the web's history, and ethereum.org tells it in three steps.
Web 1.0 was mostly static websites owned by companies, with almost no interaction between users. ethereum.org dates it roughly 1990 to 2004 and calls it the read-only web. Web 2.0 began around 2004 with social media platforms: people could now post and share, so it became read-write. ethereum.org adds that, as more people came online, "a handful of top companies" began to control a disproportionate amount of the traffic and value.
Web3 is the third step. ethereum.org says the idea of Web 3.0 was coined by Ethereum co-founder Gavin Wood shortly after Ethereum launched in 2014, and quotes a 2020 post: Web1 was read-only, Web2 is read-write, Web3 will be read-write-own.
Two cautions. First, these are labels, not releases. PIA says plainly that they're informal, and nobody has to agree on where one ends. Second, "Web 3.0" has also meant something else. PIA points out that it historically referred to the Semantic Web, a way of structuring online data so software can understand how it connects, and notes that this idea never became the dominant model of the web. If a page says Web 3.0, check which one it means. We sorted out the next label in the sequence in our guide to what Web4 is.
What Web2 actually is
Web2 is the interactive web behind most of what you use: social networks, online shops, streaming, work apps. PIA's description is the cleanest: one organisation typically runs a Web2 service and operates both the application and the systems behind it.
That has real advantages, and the pages we read sometimes skip them. NordVPN lists easy communication, user-generated content and simple use: "You can share and update information, make purchases, and reserve services with a few clicks, without having advanced IT skills." PIA adds that Web2 services can handle large volumes of activity on ordinary server infrastructure without recording every action on a shared ledger, which makes them easier to scale.
It also has the problems everyone points at. NordVPN names security (a breach of one central system can expose a lot of user data), data ownership (companies collect and monetise it) and censorship (platforms can block information or users). ethereum.org's examples are blunt: a payment service may decide not to allow payments for certain types of work, and, in its table, Twitter can censor any account or tweet.
What Web3 actually is
Web3 is a way of building applications that uses decentralised protocols for some tasks a single company would otherwise handle. That's PIA's wording, and "some tasks" is the important bit. In practice it usually means a blockchain records activity or runs part of the app's logic.
PIA lists the building blocks, and they're worth knowing because they're what your users will meet:
- Blockchain: a shared record kept in sync by many connected computers, called nodes.
- Wallet: software or hardware that manages the keys used to approve actions.
- Smart contract: code stored on a blockchain that runs by its own rules.
- dApp: an app that uses a decentralised protocol for part of what it does.
- Token: a digital unit recorded on a blockchain. It can stand for value, access, a vote or something the app defines.
PIA also walks through what happens when someone uses one. The dApp prepares an action and sends it to your wallet. If you approve, the wallet signs it with your private key, which never leaves the wallet. The network's nodes check that the action follows the rules, a smart contract runs if the action triggers one, and the blockchain records the result.
That sequence is the thing to keep in your head for the rest of this guide. Almost every difference below comes from it: a key instead of a password, a shared record instead of a private database, and a wait while the network agrees.
The differences, side by side
The pages we read mostly give decentralisation as the headline difference. Here's the longer version, in one place. The phrasing follows PIA and ethereum.org, and both stress that these are common patterns, not strict rules.
| Question | Web2 | Web3 |
|---|---|---|
| Who runs it? | A company | A network of independent computers, plus whatever the project keeps centralised |
| Who sets the rules? | The company | Protocol rules, and sometimes a community vote through a DAO |
| How do you sign in? | Account with a password or another provider | Wallet signs a message with a key |
| If you lose access | The company can reset it | Possibly nobody can, depending on the wallet's recovery setup |
| Where does the record live? | The company's database | A shared ledger anyone can inspect |
| How do payments move? | Banks and payment processors | Transfers recorded on the chain, often still with an exchange at the edges |
Control and governance
In Web2, the company decides how the platform works, when to update it and which policies apply. Web3 can spread some of that decision-making. PIA describes DAOs, decentralised autonomous organisations, where members vote on proposed changes, often by holding governance tokens. It adds that Web3 doesn't require that model, and a project can still leave most control with its developers.
ethereum.org's table of centralised and decentralised systems is the fairest summary we've seen. Centralised systems have a single authority that can resolve disputes quickly and push upgrades through with little friction. Decentralised ones have no single point of failure and no gatekeeper, but coordination is harder, and in the worst case a network can fracture over a disagreement about changes. It also stresses that real networks sit on a spectrum, not at either end.
Ownership and data
Web2 platforms usually keep your account data and digital items inside systems they control. ethereum.org's example is a game: buy an item in a Web2 game and it's tied to your account, so if the creators delete the account, the item goes with it. In Web3, ethereum.org says, ownership is recorded on the chain, so not even the game's creators can take it away.
PIA adds the honest limits. Control over a token through your key doesn't mean every app can use it: the other app still has to support the same chain or token standard. And controlling a token isn't the same as owning every piece of data an application stores about you.
Identity and recovery
This is the difference your users feel first. Web2 asks you to sign in to an account, with a password or through another provider. Some Web3 services let you sign in with a wallet instead: the service sends a message, your wallet signs it, and the service checks the signature against your address. Sign-In with Ethereum, a published standard (ERC-4361), describes this as off-chain authentication for Ethereum accounts to establish sessions. PIA notes that it confirms control of a key but doesn't verify your real-world identity.
Recovery is where the two split hard. A Web2 service can reset your password because it controls the account system. ethereum.org's account abstraction page describes the other side: accounts are secured by private keys calculated from seed phrases, and "If you lose your keys they can't be recovered, and stolen keys give thieves instant access to all the funds in an account." The same page describes newer account designs where lost keys can be recovered using multiple backups, but those are a goal being built, not something every wallet does today.
Privacy
Web3 isn't automatically more private, which surprises people. PIA says most wallet addresses are pseudonymous rather than anonymous: your name may not appear, but on many public blockchains anyone can view transactions and follow activity tied to one address over time. If someone links an address to you, they can trace the rest. PIA adds that holding an asset yourself doesn't make the activity around it private.
Web2 has the opposite problem. Services connect your activity to your account and collect personal information under their own data practices. So it's less "which is more private" and more "who can see what".
Security
Neither is safer by default. PIA's line is the one to remember: they introduce different risks in different places. With Web2 the provider handles most of the application's security, though you still have to protect your own account. With Web3 there are two separate jobs: keeping your key safe, and trusting the code you approve. A safe key can't protect you from a flawed smart contract, and secure contract code can't protect you if someone gets your key.
NordVPN's page dates from 2023 and refers to the 2022 crypto hacks. It cites its own study, which found 7 in 10 Americans have security concerns about digital currency transactions, which is the human side of the same point. It's the vendor's own number, from an older page, and we haven't checked the survey, so treat it as a signal, not a measurement.
Payments
Web2 payments usually pass through banks or payment processors. Web3 apps can move blockchain assets by updating the chain's shared record, and a smart contract can enforce a rule, such as releasing money only when a recorded condition is met. PIA's caveat matters for anyone building a product: it doesn't remove every intermediary, because you may still need an exchange to turn crypto into ordinary money.
Speed, cost and scale
Here Web2 has the edge. ethereum.org lists scalability (changes to state, like a payment, need to be processed and spread through the network), UX, accessibility and cost as current limitations. PIA says Web2 services are easier to scale on ordinary servers, and that a Web3 transaction may take time to be recorded and may carry a fee, and that heavy network use can raise confirmation times or fees. It also says blockchain technology isn't inherently energy intensive, and that energy use depends on the chain's design.
Reads and writes also differ: PIA points out that checking existing blockchain data doesn't have the same requirements as submitting a transaction that changes it.
Openness and censorship resistance
This is Web3's strongest card. A Web2 platform can require you to meet its access rules and suspend activity under its policies. A public, permissionless protocol allows all valid transactions without approval from a central operator. PIA is careful, though: calling something Web3 doesn't guarantee either property, and the real level depends on how distributed control is in practice.
Why most Web3 products aren't fully decentralised
This is the part the shorter explainers leave out, and it changes how you should read the table above. PIA lists the parts of a typical Web3 app that are often still centralised:
- The front end. The website you see can be hosted by a company like any other web app.
- The RPC provider. A service that lets the app talk to blockchain nodes.
- The indexer. Raw blockchain data isn't practical for searches, so services organise it.
- Oracles. Smart contracts can't fetch real-world information, so something supplies it, such as a price.
- Off-chain storage. Information the app keeps outside the chain.
PIA's conclusion is the right one: decentralisation works more like a spectrum than an on/off switch, since using a blockchain for one function doesn't decentralise the rest. ethereum.org makes a related point from the other side: "This doesn't mean that all services need to be turned into a dapp."
The pages also disagree with each other here. ethereum.org's practical comparison says Web3 servers "can't go down" because they use a network of thousands of computers as the backend. PIA's list above is the counterweight: the front end and the RPC provider can still be one company's servers. On whether Web3 will replace Web2, PIA calls coexistence more realistic than a complete shift, and NordVPN calls an integration of the two more likely. Our view: the useful question isn't which web wins, it's which jobs in your product deserve which side.
Examples of each
Web2 is whatever you used today. NordVPN's table names the payment methods most people know: credit and debit cards, bank transfers, Google Pay, Apple Pay, Amazon Pay, PayPal, Venmo and Zelle. Add social networks, online shops and streaming and you've covered most of it.
Web3 examples are mostly categories, and the sources lean on their own chains. PIA lists blockchain-based lending, social networks on decentralised protocols and blockchain-based naming or identity systems. Algorand's page names projects on its own chain, as a vendor would: the lending protocols Folks Finance and Tinyman, the decentralised identity project NFDomains and the supply chain tracker Wholechain. ethereum.org points to DAOs, where token holders vote on how shared resources are spent, and to Sign-In with Ethereum as a feature available to anyone at no cost. We haven't tested any of these products, so read them as examples the sources gave, not as recommendations.
What the difference changes on screen
The pages we read touch on this (ethereum.org lists UX hurdles, PIA covers wallet signing and recovery), but none goes deep on the design side, which is the part we know best. Every difference above shows up as something a person has to do or understand. Here are the five places we'd look first, drawn from the sources above and from our own work, which we mark as ours.
1. Sign-in. A password is something people already know how to do. A wallet signature isn't. ethereum.org itself lists extra steps, software and education as a hurdle for Web3 apps, and says the whole space asks people to learn different mental models. Our view: decide what your first-time visitor needs to do before they've decided to care, and don't ask for a wallet until the product has earned it. We wrote about this in Web3 onboarding UX.
2. Recovery. In Web2 "forgot password" is a screen. In a self-custodied wallet it may be nothing at all, as the account abstraction page says. Your interface has to say, at the right moment and in plain words, what happens if the key is lost, and who (if anyone) can help.
3. The wait. A Web2 button often answers instantly. A chain transaction takes time to be recorded and can cost a fee, and that wait needs a visible state: sent, pending, done, failed. Our advice: show sent, pending, done and failed, because people can't tell whether a silent wait is working.
4. Mistakes that can't be undone. PIA's point about approving a transaction that interacts with flawed or malicious contract code is a design problem as well as a security one. Say what an approval does before the person signs it, in words they'd use, not the code's.
5. Numbers you can check. A score or a balance is only believable if it leads somewhere. This is the lesson we took from Reputable, below.
How Reputable shows the mix
Reputable is a reputation platform for Web3 developers. Its case study explains the problem: in Web3 most builders go by a handle, not a name, and a CV proves very little, but what does prove something is public: the code a developer commits, the protocols it ends up in and the value those protocols hold. The founders could already score all of it. What they didn't have was a product that made the score believable.
The design answer, as the case study puts it: every figure had to lead somewhere, to the commits behind it, the protocols it touched and the peers who vouch for it. We worked with the founding team for three months, from a ranking model in a spreadsheet to a design system, and designed 22 screens in dark and light. The case study reports 3.2 times the profiles seen per visit.
Two things in it are useful for this comparison. First, the product is about Web3, but the web app is one developers sign in to with GitHub, an ordinary Web2 sign-in, and a developer claims a profile that already exists. Second, the evidence behind each score is public work: commits, the protocols they ended up in and the value those protocols hold. The case study doesn't say how much of that is read from a chain, so we won't claim it. Our reading is that this is a common shape: a normal web app built on top of what's public in Web3, which is a different thing from a pure dApp. And the design lesson holds either way: a number nobody can trace is just a number, so give every figure a way to check it.
Fedica and Dino, from the other side
Fedica is a social media platform that publishes to every major network, with a shared inbox and audience analytics. We redesigned it over six months into one workspace of 193 screens. We think nothing about its job asks for a blockchain: it needs speed, search and one place for fifteen tools. Dino is a group payments app with a shared card per group. We audited its busiest screens and redesigned 21. Its problem was that every screen showed money and few said whose it was, and the rule we ended on, every number says whose money it is, would, we think, apply just as well in a wallet.
Neither is a Web3 product, and neither is worse for it. We use them because they show the thing the pages we read rarely say: a lot of the design work is the same on both sides. Clear labels, one obvious next step and honest states matter whether the record lives in a database or on a chain.
A short test: does this part of your product need a chain?
Our view, not a standard. For each job in your product, ask:
- Does anyone need to verify this without trusting you? If users, partners or regulators must check the record themselves, a public ledger earns its place. If they'd happily trust you, a database is cheaper and faster.
- Does anyone need to hold it without your permission? An asset a user should keep even if your company disappears is a reason for a chain. A preference or a draft isn't.
- Can you accept the wait and the fee? If the job needs to feel instant and cost nothing, keep it off-chain.
- Can your users handle losing a key? If the honest answer is no, plan recovery before anything else.
- Is a person going to be confused at this step? If so, hide the chain behind a plain screen.
If you're early, put the minimum on-chain and the rest on ordinary web, which is how we'd scope it in a Web3 MVP. We cover the design side in Web3 UX design.
FAQ
In Web2 a company runs the service and controls the accounts and the data. In Web3 some functions run on a public blockchain, and you use a wallet key instead of a password. PIA puts it as where control sits and how much the service depends on a central operator.
Not on the evidence in the pages we read. PIA says coexistence is more realistic than a complete shift, and NordVPN expects the two to integrate. Most Web3 apps already depend on ordinary web parts.
Neither is. PIA says they carry different risks in different places. Web3 adds the risk of your own key and the risk of the contract you approve.
Web2 covers most social networks, online shops and streaming services. PIA's Web3 examples are blockchain-based lending services, social networks on decentralised protocols and blockchain-based naming or identity systems.
People use them interchangeably, but Web 3.0 historically also meant the Semantic Web, a different idea about structured data. Check which one a page means.
No. Fedica and Dino are ordinary products. Add a chain only where someone needs to verify or hold something without trusting you.
We designed Reputable, a web app for Web3 developers that signs people in with GitHub. We haven't designed a wallet or an exchange, so tell us the exact screens you need.
If you're deciding which parts of your product belong on a chain, or you already know and want the screens to make it plain for the people using it, book a call. Our UI/UX design services cover the flows, the screens and the testing with real users, and our pricing page shows how we work.
About the author
Imran Khan, Founder & Creative Head
15 years of industry experienceFormer Head of Design at Day TranslationsFormer Creative Lead at SupersideWork at Superside for brands including Meta, Red Bull and Ray-Ban

related articles
Book a call
Pick a time that suits you
Get in touch
Let’s build something amazing together
Let’s build somethingamazing together
For product, interface and web-system work that needs clear direction and careful execution.



