Perhaps you shouldn't just implement ideas, you should implement technologies. And carefully map out your user flows. And then you should make sure that each conversion step actually converts in the real world.

You probably could have done this with Google search.

Would you love to replace the super slow inaccurate search you are using now with one that delivered accurate results in 1/10th of the time?

[1999] Sure, but I'll believe it when I see it. Look, I've already tried about 10 different search engines, and even with the best of them (altavista) you count on having to click through a few results pages to find what you want. Web servers aren't psychic, you know. You get out what you put into it. There are already a lot of smart, well-funded companies working on this, so good luck.

People who Knew Search back then knew that tedium was part of the bargain, just like people who Know photo sharing today know that tedious sorting/tagging is inevitable if you shoot a lot.

A shot in the dark will sometimes hit something. This doesn't make shooting in the dark a viable strategy.