What this is
A record of what happened to brand crypto: the NFT collections, loyalty tokens and marketplaces launched by large consumer companies between roughly 2021 and 2023, and what became of them.
Most were switched off. The interesting part is that the contracts were not — a token contract cannot be recalled — so the tokens keep existing, and in some cases keep trading briskly, years after the company that issued them stopped answering emails about it.
Two kinds of claim
Everything here is one or the other, and they are verified differently.
Claims about what a company did
Launch dates, shutdown dates, causes, money raised. Every one of these carries a citation, and the build refuses to publish an entry marked dead or dormant unless at least one source is dated on or after the ending. As it stands there are 31 citations across 12 entries.
Where a company or a regulator stated something itself, that source is marked primary and preferred over press coverage of it.
Claims about what is on-chain
Supply figures, trading activity, whether a contract still exists, whether a website still resolves. These are never copied from a third party. They are measured by calling a public node and fetching the URL directly, and each entry records the date it was measured.
6 entries name a contract. Each address carries a valid EIP-55 checksum — which is the difference between an address being well-formed and being the right one — and links to a block explorer where you can confirm it yourself in about ten seconds.
What the four statuses mean
| Status | Means | Requires |
|---|---|---|
| Dead | Formally shut down or wound up. | A date and a citation dated on or after it. |
| Dormant | Never formally killed, but no operator activity. | A stated reason. No death date, because none was announced. |
| Absorbed | Folded into something else rather than closed. | A stated reason and a citation. |
| Alive | Still actively operated. | Nothing — but it may not carry a death date. |
Dormant is the careful one. It means the evidence is absence: no announcement, no activity, and often a subdomain that no longer resolves. Absence of evidence is weaker than a shutdown notice, so dormant entries say what specifically is missing rather than asserting the project is finished. Nothing is upgraded from dormant to dead without something citable.
Where this could be unfair
Worth stating plainly, because a graveyard is an inherently loaded frame.
- Survivorship works backwards here. A record of endings over-represents endings. There are brand projects still running that are not listed simply because nothing happened to them, which is not newsworthy but is real. The control group exists partly to keep this honest.
- Shutting something down is not a scandal. Most of these were experiments, and ending an experiment that did not work is ordinary corporate behaviour. The criticism implied here is narrower: almost none of them planned for what happens to holders afterwards.
- Trading volume is not evidence of value. A high transfer count means tokens are moving, nothing more. It is recorded because it contradicts the assumption that a shutdown ends the thing, not because it says anything about worth.
- Dormancy is inferred. A company could restart tomorrow. If one does, the entry changes.
Compiled with automation
Research, drafting and the on-chain measurements are done with heavy AI assistance. That is what makes a project this small viable, and it is also exactly why the verification rules are mechanical rather than a matter of judgement.
The failure mode of an automated record is a plausible entry for something that never existed. So the build fails — it does not warn — if an address has a bad checksum, if a source URL does not point at the contract it claims to, if a project is marked dead without a citation for the ending, or if the copy drifts into market commentary. Being unable to publish is the safeguard.
It is not a complete safeguard. Machines are good at consistency and bad at knowing when a source is simply wrong. If you find an error, the correction process below is the real backstop.
Corrections
Every entry here concerns a real company. If anything is wrong we will correct it promptly and without argument — email us. Corrections are logged on the entry.
If you worked on one of these projects and the record is wrong, incomplete, or unfair in framing, please say so — hello@retailtoken.ai. You do not need to justify the request. Corrections from the people involved are the most valuable input this gets, and they are free and prioritised.
Requests to remove an entry entirely are considered, though a documented, cited record of a public product launch is generally something worth keeping — usually the answer is to correct it rather than delete it.
Submitting a project
To go in the ground, an entry needs:
- the brand and the project's own name;
- a launch date, at least to the month;
- a shutdown date and cause, if there was one;
- at least one source documenting the ending;
- a contract address and chain, where a single one exists.
Send it to hello@retailtoken.ai. Contracts are checked against a public node before anything is published.
The boring but necessary part
Nothing here is financial, investment, legal or tax advice. This is a record of what happened, not a view on what anything is worth.
Company names, product names and marks belong to their owners and are used here to identify the projects being recorded. No affiliation, endorsement or sponsorship is claimed in either direction. Where a regulator's findings are described, they are described as findings and linked to the filing.