Most blockchain projects we are asked to price should be a database, a good audit log, and a harder conversation about who is allowed to edit the truth.
That sentence loses some rooms and saves years in others. We would rather lose the room.
Chain theater is expensive. It shows up as a token, a public roadmap, and a promise that "the blockchain" will create trust. Trust does not appear because a hash exists. Trust appears because incentives, law, operations, and software agree on what a record means when two parties disagree.
If you can solve the disagreement with a contract, a reputable operator, and a database you can restore from backup, you probably do not need a chain. If you cannot, you might.
The honest filter
Use this filter before you talk to a developer, including us.
Ask: who is not allowed to be the single owner of the record, and what happens if they lie or disappear.
If a single company can own the record, and customers already accept that — your bank, your SaaS, your internal ERP — a normal database with permissions and an audit trail is the grown-up answer. It is cheaper to run, easier to fix, and easier to explain to a regulator who does not want a seminar.
If several parties must share a record, they do not trust one of them to host it, and they need a history that is hard to quietly rewrite, then you are in the small set of problems where a ledger earns its keep.
Everything else is costume.
When a database is enough
These requests come in dressed as innovation. They are still databases.
A customer loyalty point system you fully control.
An internal workflow that needs a status history.
A marketplace where you are the operator and the dispute desk.
A certificate page that could be a signed PDF and a verification URL.
An NFT because the brand team wants a moment.
A signed document plus a public verification page is often the entire "provenance" story a buyer needed. Shipping that as a chain does not make the document truer. It makes the launch louder.
If your fear is "someone on our team might change a row," fix access control, logging, and four-eyes approval. That is governance. A chain will not repair a company that cannot decide who has admin.
The three cases it is not the wrong tool
We will build blockchain products when the problem is one of these. Not because the word is fashionable. Because the architecture matches the politics of the record.
1. Provenance that several strangers must believe
When an asset moves through hands that do not share a boss — a machine, a batch, a title packet, a high-value part — people ask for a history they did not have to take on faith from the last seller.
A ledger can hold "this identifier was attested at this time by this party" in a way that is painful to rewrite. It does not replace inspection. It does not replace law. It makes the attestation portable.
If your provenance story still depends on one operator's admin panel, you have a database with extra steps. If the whole point is that no single operator should be able to tidy the past, you are closer to a real case.
2. Settlement that should not wait on one company's books
Cross-border payment, escrow-like release, and tokenized settlement show up in our work because moving value between parties who do not share a bank is still slower and more political than moving a row in Postgres.
Here the chain is not a brand story. It is rails. The product questions are unglamorous: who pays fees, what happens when a transfer is stuck, how you represent the off-chain event that authorized the on-chain move, and how a human unsticks a Monday morning failure.
If you do not want to answer those questions, you do not want this architecture. You want a payment provider and a status email.
3. Shared records when the fight is about the copy of the truth
Consortia, multi-dealer networks, and multi-party registries sometimes fail for a simple reason: nobody will let a competitor host the system of record.
A shared ledger is one way to stop the argument about whose database is real. Smart contracts can encode the few rules everyone already agreed in a legal schedule: who may write, what must be present, what is final.
Notice the order. The legal schedule comes first. The contract on chain is a narrow machine for a wide human agreement. If the humans have not agreed, the contract will encode the fight, permanently.
What we will not pretend
We will not tell you a token will create a community.
We will not tell you an NFT is a strategy.
We will not tell you immutability is always a feature. Sometimes you need to correct a record because the physical world was wrong. A system that cannot correct, or cannot correct in a visible way, is a liability.
We will not hide fees, keys, and failure modes behind "web3." If your operations team cannot explain the failure mode, the architecture is not ready.
And we will not put a chain under a product to make a pitch deck sound current. That is how you inherit a runtime you did not need and a story you cannot maintain.
Keys are operations
If a chain is justified, keys are not an IT footnote. Who holds them, who can rotate them, and what happens if the person who set this up is on a plane.
Write that before you mint anything, move value, or tell a customer the record is "on chain." A ledger without an operational key story is a locked cabinet with the only key in someone's laptop bag.
This is also why "we will figure out custody later" is not a phase. It is the product, delayed until it is expensive.
How to brief a blockchain project without wasting a quarter
If you still think you are in the three cases, write this one-pager before anyone opens an editor.
The parties who will not accept one company's database, named.
The record they must share, in one sentence.
The event that is not allowed to be silently edited.
The off-chain facts the chain cannot know — inspections, court orders, warehouse receipts — and who attests them.
The exit. How you still operate if a network, a key, or a vendor fails.
If you cannot name the parties, you do not have a multi-party problem. If you cannot name the event, you do not have a ledger problem. If you cannot name the exit, you have a dependency you have not admitted.
That brief is the same discipline we use on AI and marketplaces. Architecture follows the politics of the work.
When the brief is real, blockchain development is a craft: contracts, keys, payments, and the unglamorous edges. When the brief is a vibe, it is a costume.
If you want a second opinion before you fund a chain, start a project with the filter at the top of this page. We will tell you if you need a ledger, a database, or a better process. The useful answer is the true one.