Research on a token is mostly a filtering exercise. There are thousands of assets, the promotional material is abundant and the verifiable information is thin, so the useful discipline is an order of operations: check the things that cannot be spun first, and read the project's own account last, when you have something to test it against.
Begin with the contract and the chain. Is the contract verified on a block explorer, so the code that is running can be read? Who can call privileged functions, and has ownership been renounced? How many holders are there, and what share do the largest ten control? Where does the token trade, and how deep are those pools? None of this requires trusting anyone. It also disposes of a large fraction of candidates within a few minutes.
Second, supply and unlocks, which our piece on tokenomics covers in detail. Circulating against maximum, the emission schedule, insider allocation and the vesting calendar. A large unlock in the near future against a shallow market is a specific, dated, published risk.
Third, usage. Does anyone use the thing? Look for measures that are hard to fake — fee revenue paid by users, addresses transacting rather than addresses in existence, activity that persists between incentive programmes rather than tracking them exactly. Transaction counts and total value locked can both be inflated cheaply, so prefer figures that represent someone paying for something.
Fourth, the people and the history. Are the team identifiable, and do their claimed histories check out independently? Has the project shipped what it previously said it would, on roughly the timeline it gave? Have there been security incidents, and how were they handled — a disclosed and remediated exploit is a better sign than a project with no incidents and no audits. Read the audit reports rather than the badge: audits are scoped, they expire, and the findings section is the part that matters.
Only then read the whitepaper, the documentation and the community channels, and read them for internal consistency with what you already established. Two habits are worth keeping throughout. Look for the argument against the project — someone has usually written it, and the response to it is informative whether or not it convinces. And write down what would make you conclude you were wrong, before you have a position, because after you have one, everything you read will start to agree with you.