He's outed me: I began an experiment about four weeks ago, changing my home page from Google to Yahoo Search for the first time in ten years. The (admittedly imprecise) verdict? If you swapped in Google branding, and added an "I'm feeling lucky button" (the only thing I truly miss), I wouldn't notice the difference. (I'd wager Microsoft Live would yield similar results.)
Google leads now by branding. The gap between the major search engines has narrowed to indistinguishability.
I started searching on Yahoo when I started working with a very passionate colleague from yahoo. It is truly brilliant that for a lot of long tail queries yahoo does a great job. If they can crawl as frequently as google then they are definitely superior in terms of ranking.
Google really got its edge being the default on many browsers. Every time I use a new box or a new browser installation its a long time (months) before i change to Y! search.
I have an old personal home page that I haven't updated in a couple of years. Yahoo crawls that page almost daily, while it has weeks since the last visit from Google.
even if it is, very few people will ever cross search between them. i.e. for me if I can't find something with Google for a specific keyword, I don't go to yahoo to search for that same keyword, I just modify my existing keyword string until I find what I'm looking for
And therein lies the rub for competitors: to even get a tryout, they are competing not with a first search at Google, but against a refined second search at Google, when the user is already staring at a page of results there. And even if disastified with that page, they're probably good enough to hint how the query should be changed.
Great insight. Search used to seem such a fragile business, because where's the sustainable competitive advantage when there're no switching costs? When better search comes along, everyone switches!
But... unless the competitor is significantly superior, why switch? So Google just needs to be good enough such that no one is significantly superior in a way that makes a difference to users. If you're technically better, but it doesn't really help anyone, it doesn't harm Google.
Actually, for what it's worth, that's exactly what I do, (except I start with Yahoo, not Google) especially if I'm not sure what form the information is going to take--that is, I'm in "discovery" mode. I may look a couple of times on Yahoo first, but often I just try Google.
Testing only on your own searches is not a good measure of overall quality.
A good measure of quality can be found in your own webserver logs. In my experience with mass-market sites, page-views-per-user from Google traffic is much higher, indicating that they do a better job of matching user intent with content. Googlebot tends to be much faster and smarter than Slurp and it's fairly easy to get good ranking on long-tail stuff in Yahoo.
A caveat: the userbase of Yahoo versus Google is different in character, so that may account for the difference in behavior. For whatever reason the Yahoo traffic tends to be celebrity gossip-type stuff and basic noun searches.
I don't think that is a good test, simply because Google has many more users than Yahoo. Maybe one thing you could see from the server logs is how well the search keywords matched your site, though.
I would offer conversions as a better test of match, indicating that the search actually returned what someone was looking for. But even that is not very good, as there are many other things involved in making a sale (or other conversion), as I have learned the hard way.
For developers, it's way better, because they've got a much more open API at this point. I used their search API for langpop.com and am quite happy with it.
I won't (and shouldn't) go into specifics, but people in the search engine biz do a lot of comparison and contrast between search engines. They rate their own results and their competitor's results, both in a vacuum and in groups.
Yahoo!'s search was not one of the best sets of numbers I've seen. Google consistently came out on top. Particularly for longer queries, Google is notoriously good at these by comparison to other keyword search engines.
But it might be that was for our dataset bias. You can get surprising boosts in relevance for specific datasets by biasing towards them in your index and ranking algorithms.
I've had the same experience as Searls: frustrated that Google is not returning a page that I know exists, I then have success with Yahoo.
What perplexes me about Yahoo is that they haven't replicated some relatively simple things that would make Yahoo a drop-in replacement for Google. Why no full calculator with unit conversions? Why no 'define:' operator?
They've invested so much in the big hard problem of web-wide search; can't they throw a few person-months at filling some gaping holes at the periphery? I'm sure I'm not the only person for whom these gaps disqualify Yahoo from my default setting.
I get the feeling that Yahoo! doesn't want to be seen as copying Google. They have the unenviable task of trying to innovate without looking like a me-too competitor of a very innovative company.
Yahoo has been slow to adopt other practices, too (their hosting still doesn't support php 5; python is unavailable, etc.) --seems like the turmoil there may have some effect on where they are putting their strongest efforts.
I was up to switch, but immediately ran into a problem - a yahoo image search for "Beijing architecture glass box" returned 0 results. Google has pages and pages.
i wonder if this is due to the nature of PageRank. eg when you search for some words that you intend as direct, Google sends them through PageRank where they're stretched, weighted, and pulled apart by the Graph of the internet, whereas Yahoo's algorithms might be rawer (but probably not as raw as they were in the early days)
Google leads now by branding. The gap between the major search engines has narrowed to indistinguishability.