"You can still have problems even if you installed new batteries and all the voltages check out. If your game is still giving "Factory Setting Restored" or "Set Time and Date" errors, you may have a bad CPU U8 RAM chip. But make sure you double check that battery holder. Even minor corrosion can cause this problem. The voltages may all check out, but the corrosion may be enough to limit CURRENT, and cause this problem. The U8 RAM chip is a 6264-L or 2064 RAM chip."
http://techniek.flipperwinkel.nl/wpc/index3.htm

My suggestion is that if you are absolutely certain that the current and voltage are sufficient, you probably need to replace U8

Another thing to consider is that on WPC-S (not positive WPC-95 is the same) the date and time are calculated by the ASIC. VBATT is provided to the RAM chip and the ASIC to keep the date time calculation going. I have found that the ASIC is much more sensitive to lower voltages than the RAM and will not keep accurate time when VBATT is below 3.2V or so.

Is it not keeping the time at all or just doing a poor job of keeping an accurate time?

If you set the time, will it be accurately kept for an hour or more while the machine is left on?

Good point.

ASIC VCC powered by VBATT is pins 1, 22, 43, 64.

Before you mess with trying to test those tiny pins on the ASIC, I would check is the 32 kHz oscillator that times the RTC (its a bit easier to find). It is a basic inverter (one of the 6 inverters in U21) timed with a 32 kHz crystal connected to pin 70 of the ASIC.

Check that VBATT is making it to U21 VCC on pin 14. Measure the machine off voltage there, it should be the same as the voltage you measured at the RAM chip.

If the voltage checks out,verify the oscillator is working by checking pin 12 of U21 with your logic probe with the machine on and again with the machine off to see if it is oscillating. If you get oscillations on pin 12 in both cases, check that the oscillating signal is making it to pin 70 of the ASIC.

Then I would check that VBATT is getting to the ASIC pins (it pretty much has to as the machine would not run without it)

If you have an PLCC puller, you could try re-seating the ASIC.

If this all checks out, I think you should look to the RAM chip as wayout pointed out...

Does the board have any alkaline corrosion damage?
--
Chris Hibler - CARGPB #31
http://www.Team-EM.com
http://webpages.charter.net/chibler/Pinball/index.htm
http://www.PinWiki.com - The new place for pinball repair info

Yes, the battery power is getting there - but FWIW, this does not tell you if the current is actually sufficient. I discovered early on with Williams sytem 11s that you can get memory hold errors if the battery circuit and its available current is not 100% perfect, even if you measure the proper voltage. I've seen this specifically in a case where a memory capacitor was used in place of the 3 AAAs and couldn't quite cut it, and I've seen it where the combination of a splice and too long of leads for an offboard battery back were causing just enough of a loss of current to keep the memory backup circuit happy.

I'm not saying that the hold voltage/current is definitely your problem, I'm just saying that as your troubleshooting progresses, if some other root cause is not found, to keep this in mind on your list of possibilities....especially because you are dealing with a board that has a history of corrosion.