Possible way to classify projects by type of end-result value (not very serious).
1. Projects to earn money
In this category are projects with direct impact on revenue, profit, cost. For instance, change to more productive technology or cutting expenses.
Value is described in money as a direct result.
2. Projects to mitigate or prevent some risks
Projects about better manageability fall into this category. For example, improving release or incident management.
Changes about risks (probability * possible impact) should be outcome of a such projects, so it is to be estimated.
3. Projects to gain more loyalty of customers
These should be projects about quality perceived by customer. First of all, reducing number of failures or % of defects. But also, making faster service, providing more communication paths and any other changes clients can feel as positive.
Value can be estimated in money based on logic of influence.
Let assume, you know relationship between NPS (Net Promoter Score) and customer churn. For example, you found that increase of NPS for 10 correlates to decline of churn for 5%, and it's clear that decline of churn for 1% = saving 1% of revenue. If by managing some quality aspect we change proportion of detractors and promoters such way that NPS grows, for instance, from 50 to 75, saving of revenue should be 12,5%.
To formally prove value of such project, you need to prove NPS / churn and NPS / quality aspect correlation.
4. Projects for passion and pleasure
Value can not be estimated. You are engaged, probably, to someone else's loyalty project (type 3). But, anyway, project could be the most valuable for you.