It all depends on the penetration rate of the category that the product is in. For example, toilet paper has high penetration, with 90%+ of all households buying (its not 100%, weird but another story). So a new product in that category might get 10% penetration which is quite low if almost every household is buying that category. Instant noodles though might have 30% penetration, so a launch penetration rate of 10% is relatively much better. The trick with new products is what market its targeting - niche v mass for instance. You wouldn't expect that same household penetration for a niche noodle product versus a bog standard bog roll. The key to success is really the repeat rate, i.e. %age of purchasers who buy again. As a guide 30% is pretty good.

I agree with Patrick in general, what I would is that there a different ways to quantify success (second part of the question).

In terms of pure innovation (a new product) the success rate is low. If you look at the number of products still in market 4 years after launch anything above 10% is pretty good.

For incremental or replacement innovation (a new flavour, or having an updated version of the old product) it is much higher. This makes sense though as you are forced to buy the new product as the it is replacing the older one, hence it only fails if the established brand disappears which is relatively rare.

Also note that some incremental innovation is only designed to last for a short amount of time, if you look at snacks the big brands always bring out new flavours to keep consumers interested, but these only stay in market for a season, or maybe a year at best. Supermarkets don't have space to stock everything so they keep the innovation cycling through.