Are average user acquisition costs relevant or actionable?

No, external average UA-related metrics aren't actionable.

Your own proprietary, average UA-related metrics can be actioanble for the purposes of modeling out cash flow. But I wouldn't use any general, universal, average metrics for the purposes of optimizing UA campaigns.

You should strive to use dimensionalized metrics wherever possible; more here: The "average user" doesn't exist in freemium gaming (and a diagram for user segmentation)

in my experience, these averages arent super-accurate or actionable - so your estimates can be somewhat off. I think the CPI(~$3.60) and cost per free trial(~$7) you mention are somewhat in the ballpark of what we've seen, but the conversion to premium is significantly higher. here are rough ballparks for a couple of subscription apps we've seen:

CPI -> ~$2.50 - $4(this is a function of audience and targeting)

Cost per trial -> ~$7-~$8(so ~40-50% conversion from install to trial)

Cost per paying subscriber -> ~$17-~$25(~33% conversion from trial to subscription)

outside of tier 1 geos, we see CPI and CPT to be roughly 40-50% of the above -> but cost per subscriber to be roughly similar(so trial to subscription % tends to be lower outside of tier 1).

this generally seems to work out for many subscription based apps that we've worked with. obvi: each app is different and so is their LTV curve - but the above funnel hopefully offers ballpark numbers in which UA can be profitable.