Each blockchain case study on this page covers a public project that anyone can check: its design, what worked and what failed. The eight projects cover trading, lending, NFTs, stablecoins, oracles, tokenized funds, hardware networks and an enterprise consortium. Each section links to the project's own documentation, code or announcement.
How to read a blockchain case study
Many case studies online come from agencies and describe an unnamed client with results nobody can check. A useful case study names the project and links to evidence. For a public blockchain project, the evidence is usually open:
- the contract code on GitHub and the verified contracts on a block explorer;
- the project's documentation and governance forum;
- audit reports and incident reports published by the team;
- on-chain data, such as deposits, trading volume and token supply.
The sections below use that kind of evidence and describe design choices and public events, not private numbers.
Uniswap: an exchange without an order book
Design. Uniswap versions 1 and 2 used a constant-product formula (x × y = k). Each pool holds two tokens, and a trade must leave the product of the two reserves unchanged, which sets the price. Anyone can add liquidity and earn a share of the swap fee. Version 3 (2021) added concentrated liquidity: a provider chooses a price range, and each position is an NFT. Version 4 (January 2025) holds all pools in a single contract and adds "hooks", which are outside contracts that run custom code at set points in a pool's life.
What worked. Anyone can create a pool without permission. The core contracts are small, cannot be changed after deployment, and are easy for other projects to build on. Pools without permission gave markets to small tokens that no market maker would serve.
What failed. Liquidity providers can lose value to impermanent loss, and version 3 range management favors professional providers. Bots can sandwich a swap between their own trades, which is a form of MEV. Sources: Uniswap documentation. The guide to building a Uniswap-style exchange covers the mechanics in depth.
Aave: pooled lending with liquidations
Design. Lenders deposit assets into pools and receive interest-bearing aTokens. Borrowers post collateral worth more than the loan. The interest rate rises as more of a pool is borrowed. When a position's health factor falls below 1, anyone can repay part of the loan and take collateral at a discount. Oracle prices, mostly from Chainlink feeds, drive every risk check. Version 3 added an efficiency mode for assets that move together, an isolation mode for riskier collateral, and caps on supply and borrowing.
What worked. Governance sets risk parameters with help from outside risk advisers. Liquidations are done by outside bots that earn the discount, so the protocol does not run them itself.
What failed. Every new collateral asset is a risk decision. Collateral with little trading is the usual target for oracle manipulation across lending protocols. Sources: Aave documentation and Aave v3 core code. See the Aave-style lending protocol guide for parameter design.
OpenSea and Seaport: a marketplace becomes a protocol
Design. In 2022 OpenSea moved its trading to Seaport, an open-source marketplace protocol. A seller signs an order off-chain and the marketplace stores it, so the chain sees only the trade that fills the order. A Seaport order lists what the maker offers and the "consideration" items that must be paid out, so one order can sell several NFTs, pay several recipients, or accept any NFT from a collection.
What worked. Listing an item costs no gas, because only the fill goes on-chain. Other marketplaces and aggregators use the same public protocol, so they can fill the same orders.
What failed. Creator royalties failed, because an ERC-721 transfer cannot force a payment and royalties depended on each marketplace's policy. Under pressure from rivals, OpenSea announced in August 2023 that it would stop enforcing creator fees. EIP-2981 lets a contract report a royalty, but the standard cannot collect one. Sources: Seaport code and docs. More in the OpenSea-style marketplace guide.
MakerDAO and Sky: a stablecoin under stress
Design. Users mint DAI against collateral locked in vaults. When a vault's collateral ratio falls too low, outside bots called keepers buy the collateral in auctions. A stability fee and a savings rate let governance raise or lower demand for DAI. In 2024 the project renamed itself Sky and introduced the USDS stablecoin and the SKY governance token, and holders can upgrade DAI to USDS and MKR to SKY.
What worked. DAI stayed close to one dollar through several market crashes. The vault model became a template for collateral-backed stablecoins.
What failed. On 12 March 2020, known as Black Thursday, network congestion stopped many keepers from bidding. Some auctions sold collateral for zero, the system was left short, and Maker sold new MKR in a debt auction to cover the gap. Later, Maker held large reserves of centralized stablecoins in its Peg Stability Module to defend the peg. Those reserves made DAI more stable and less decentralized. Sources: Maker core contracts and Sky documentation. The decentralized stablecoin guide covers these trade-offs.
Chainlink: oracles as a separate layer
Design. A Chainlink price feed is run by a group of independent node operators. Each operator fetches prices from several sources. The operators agree on a report off-chain and publish one combined value on-chain. A feed updates when the price moves past a set threshold or when a set time passes. Applications read the latest value from an aggregator contract. The same operator model now also serves randomness, automation and, since 2023, cross-chain messages through CCIP.
What worked. Lending markets and derivatives use shared feeds and do not each build an oracle.
What to watch. An application can still fail if its code reads a correct feed in the wrong way. Applications must reject stale prices, check whether a rollup's sequencer is down, and avoid feeds for assets with little trading. Oracle handling is a common audit finding. Sources: Chainlink Data Feeds documentation. See the smart contract audit guide.
BlackRock BUIDL: a tokenized fund
Design. BlackRock launched BUIDL on Ethereum in March 2024. The fund holds cash, US Treasury bills and repurchase agreements, and Securitize is the transfer agent and tokenization platform. Only approved, qualified investors can hold the token, so the contract blocks transfers to other wallets. The fund pays its yield as new tokens and has since been issued on several other chains.
What worked. The fund uses a public chain for settlement, while investor checks, custody and legal ownership stay in regulated, traditional structures. Other on-chain products now use BUIDL as collateral and as a reserve asset.
What to learn. The token contract was the small part of the work, while the transfer-agent records, investor eligibility rules and redemption process make up most of the product. Sources: Securitize. See the tokenization platform guide.
Helium: tokens for wireless hotspots
Design. Helium paid tokens to people who installed wireless hotspots. The first network served low-power IoT devices over LoRaWAN, and a "proof of coverage" scheme checked that hotspots were real. Helium ran its own blockchain until it moved to Solana in April 2023, and it later added a mobile phone service.
What worked. Token rewards built a large physical network fast and at low cost to the project. Networks built this way are now called DePIN, short for decentralized physical infrastructure networks.
What failed. The network paid for coverage long before customers needed it. For years, data-transfer revenue was small next to the value of new tokens issued. Some operators faked hotspot locations to collect rewards. Running a custom chain also cost the team time and money. Sources: Helium documentation. See the blockchain IoT guide.
TradeLens: an enterprise consortium that closed
Design. IBM and Maersk announced TradeLens in 2018 as a blockchain-based platform for shipping documents and events. TradeLens ran on a permissioned ledger where carriers, ports and customs agencies could share container events and trade documents on one network.
What worked. The two companies built a working platform, and several large carriers joined.
What failed. In November 2022 Maersk and IBM announced that they would shut TradeLens down. Maersk said the platform was viable but "full global industry collaboration" had not been achieved, so the platform could not reach commercial viability. The platform was due to go offline by the end of the first quarter of 2023. Source: Maersk announcement. The Hyperledger Fabric explainer and the supply chain guide cover consortium design.
Patterns across the eight
| Project | Core pattern | Main lesson |
|---|---|---|
| Uniswap | Liquidity pools, fixed core contracts | Other projects build on small contracts that cannot change |
| Aave | Pooled lending, liquidations | Risk parameters and oracles decide the safety of the protocol |
| Seaport | Off-chain orders, on-chain settlement | A contract cannot enforce a payment that the token standard does not require |
| MakerDAO/Sky | Collateral-backed stablecoin | Liquidations must work when the network is congested |
| Chainlink | Shared oracle network | Each application must still check the data it reads |
| BUIDL | Permissioned fund token | Legal and operational structure is most of the work |
| Helium | Token rewards for hardware | Rewards without customer demand produce a weak network |
| TradeLens | Industry consortium ledger | A shared ledger has value only when most of the industry joins |
Frequently asked questions
What is a blockchain case study?
A blockchain case study describes how a real project used a blockchain, which design choices it made, and what happened after launch. Useful case studies name the project and link to code, documents or on-chain data.
Were any of these projects built by Blockchain App Maker?
No. Each section analyzes a public project from its own documentation and public history. We don't publish client projects without the client's permission.
Can I copy these designs?
Many of these projects publish their code, but check each license. Uniswap v3 core used the Business Source License until it changed to GPL in April 2023. Most Uniswap v4 core files use the Business Source License until 15 June 2027 at the latest, after which they change to MIT.
Which case study matters most for a new DeFi product?
Aave and MakerDAO. Most DeFi failures come from collateral, oracle and liquidation design rather than from the user interface.