Heinz estimates the lifetime value of a ketchup customer at ~$1000 ($50/year).

Starbucks is ~$14000; Netflix $290; McDonald’s $5000.
Other than at MegaETH, I’ve rarely heard of a crypto project even attempting to estimate this number for themselves.
Same for cost of customer acquisition.
I think this reflects a low level of curiosity about user base. But perhaps more importantly, it reflects a lack of thinking on the margin.
You always want to be thinking in terms of the NEXT user, dollar of TVL, or transaction.
Consider a DEX with $50m of LPing capital. When designing an incentives program, this DEX cares about attracting new LPers.
Where possible, it is likely best to design incentives such that they are concentrated on new liquidity provision, since the existing $50m is already willing to LP on the basis of existing yield.
This introduces complexity, because the DEX now needs to track the marginal LP increase, and also avoid incentivizing existing LPs from withdrawing and re-entering to capture a windfall yield.
It’s hard, so most projects just throw a yield on top, which ends up targeting the average liquidity provider, rather than the marginal one.
That overpays LPs who were already willing to show up, and leave less money available to pay more yield-sensitive LPs.
Net result is less marginal growth for the same budget.
Interestingly, this is the textbook pricing model of monopolists. The difference is the monopolist captures monopoly profits from this pricing system, while a crypto project bears the cost of the inefficiency itself.
I think a lot of the complaints around incentives programs have their roots in this accidental targeting of average (rather than marginal) user acquisition cost, lifetime value, etc. That in turn seems to stem from a profound lack of curiosity about who relevant stakeholders are, what their needs are, and how to attract one more of them.
Unfortunately, it’s easier said than done to figure this out.