A good token model is simple and it creates a wealth generation effect.
Even if your token has few functions, it can still have huge value as long as it is delivered to the right people as soon as possible.
The problem today is that Sybil attacks make it extremely difficult for projects to implement this, which is obviously one of the biggest problems our industry has not yet solved.
In response, projects are increasingly using creative points schemes to incentivize the right long-term holders and maintain network distribution.
Low circulation / high fully diluted valuation (FDV) is a real problem, as projects often cannot maintain enough buying pressure to offset token issuance, causing token prices to fall, which is bad for community and team morale.
This problem can be alleviated if newly issued tokens are distributed to the right long-term holders, but few projects today are so data-driven.
A simpler solution is to issue tokens at a lower fully diluted valuation and allow more people to participate fairly in the global secondary market.
Again, token distribution and issuance are critical, unlike "token utility", you only have one chance to get it right.
In terms of token utility, the current scalable use solutions are basically staking/locking and buyback/burning/redistribution (but the latter has regulatory risks, please consult professionals such as lawyers).
Staking/locking can be reframed as stake (thanks @NTmoney for the suggestion - works for more than prediction markets!), which is a relatively powerful lever that incentivizes specific behaviors on both the supply and demand sides.
Besides re-staking and lock-based issuance, there hasn’t been much new in the last few years.
Also, utility tokens that facilitate digital resource allocation (liquidity and hardware) and risk transfer (insurance/penalties) are good consumption pools.
Payment tokens still suck for end-user experience, so avoid them unless you’re a gaming project where users are already used to buying in-game currency.
Overall, token utility works at larger scales and over longer timeframes, and it’s a living thing that should adapt over time (unless you’re Bitcoin).