62% usable at 18 cents is roughly normal for bulk tier. The cheap providers are mostly reselling the same aggregated data, so switching between them at the same price gets you the same records. Moving up a tier, cascading providers or paying 40 to 75 cents for a waterfall search, typically pushes usable rates into the 80s. Your cost per contactable owner goes from about 29 cents to maybe 55 to 70 cents. Whether that's worth it depends entirely on your contract rate per contactable owner, which you didn't give. If you're converting 1 in 400 contactable owners into a signed contract, an extra 35 cents per record is $140 per contract and irrelevant next to your spread. If you're at 1 in 3,000, it starts to matter.
On county records: the recorder's index gives you the owner of record and the tax mailing address, and in most counties that mailing address is more reliable than what an aggregator returns because the county actually sends tax bills there. What it won't give you is phone numbers, and availability and format vary a lot by state and county, with some publishing bulk files free and others charging or restricting access. For mail-only campaigns, pulling the tax roll is often better data than skip tracing and costs you labor instead of money.
The thing your math hides is that the 38% failure isn't random. Trust-held, estate-held and long-vacant properties fail to trace most often, and those are the highest-motivation owners on your route. So your effective list is skewing toward the easier, less motivated names. I'd separate the failures out and work them by hand through probate and county records rather than treating them as waste.
