I wonder how many include the google !bang to go direct to google. I love DDG, and have it as the default search on all my devices and browsers, but more than 50% of the time I go directly to google with !g in the search, and another 20% i guess i start at DDG but then add in !g when i find the results inferior.
For many searches, google's results I find are better - a common issue I've found it DDG's results are just too general for quite specific searches - .e.g the top result is teh homepage of a brand when I search for something more specific like "brand something something" (sorry i can't come up with a great example right now).
I still use DDG as I can use the rest of the bang syntax to get to wikipedia et al, and for certain things the results (e.g. code/technical) are clearly superior to googles.
I hope DDG logs all the times I go to DDG, then add in !g, to get some great analytics on when I found the search results lacking!
I don't think it matters how often you do !g, you're still going through ddg, which suggests ddg has a better relationship with you than google.
In facilitating your use of google for your specific searches, which you apparently can determine the likelihood of beforehand, ddg gets more of your searches for everything else. This is definitely a win for you and ddg, perhaps even for google if it increases your dependence on queries in general.
Also, I don't think ddg is trying to replace google. From your numbers, ddg is getting ~ 30% of your queries without having to implement deep search.
ddg doesn't log ip addresses/track you, so they could only probabilistically identify cases where people think ddg would do better but then fall back to google, but it will likely get harder and more computationally intense as they grow.
If x% of searches go direct to google or another source, that is x% of the "1 million a day" where DDG doesn't have the opportunity to show me adverts, which is DDG's primary revenue source I assume. So the figure is relevant. But notwithstanding that I agree with your comment - DDG has won me over and has a strong relationship with me.
Re. tracking, it's not just searches that I do that go directly to google; it's the times I go to DDG, am unsatisfied with the results, then tack on "!g" to try it on google. This should provide valuable data on when a user deems results not good enough.
I've recently done a double-take on my love for Google. I want to love Duck Duck Go. I really do. I tell all my less tech-savy friends/relatives about it.
Personally, though, I can't live without deep search. I'd say about 99.9999% of my searches are in the "brand something something" category, and there's not much point in adding an extra step to every search.
Unfortunately for DDG, my less tech-savy friends have had a similar experience. Since they're never going to try (or comprehend) the !bangs, after two or three attempts at a deep search they inevitably say "Well this is nice, but Google gives me what I'm looking for exactly."
Me, I was on Scroogle until recently when it's availability started to really suffer. Now, it's all Goosh, all the time!
But higher level, !g roughly accounted for 1% of queries yesterday, which also includes all the cases where we ran out of results and people clicked on our Try google link.
Here's a real example that just occurred. A Search for "jquery select visited anchor" on DDG returns the jquery selectors docs first (okay match) and jquery.com match second (useless). A search on google.com, the first 2 results are SO answers that answer exactly my question.
My point isn't specifically you rank SO lower than it should imo (although i do think this is the case) - its in general, for me, you return roots of homepages or brands at hte top when the query is clearly looking for something very specific.
You're probably going to do a lot better automating data collection based on DDG/!g searches. I really don't know how you'd score for quality, but at least take this as an example of a search whose results were not satisfactory.
Might also be the sort of thing to look at with a specific focus group. I know that the trend is to use a naive audience, but you might want to specifically recruit self-selected power-users.
I'd also design such a study to blind the result sets. I know Microsoft raised a bit of a stir claiming that their search results in blind testing were better than Google's a fair bit of the time. Problem at least personally with those results are that the two things I care about in search are relevance, and credibility in protecting user interests.
Microsoft have a very poor track record in that latter. I don't Bing. Period.
I have an idea, how about you add a feature, opt-in where you track the searches I do, and then do again with an added !g, at which point, store those in a list where I can go back and add feedback to each one where I got improved google searches, Or the ability to leave up a DDG feedback window when I do those searches. If you make it easy for me to give you feedback, I'll give you limitless amounts of it.
I don't understand this mentality... At least by judging from your response, it would appear that Google offers superior service for you. Why not use it then?
Is the promise of no tracking/bubbling a big draw for you?
For many searches, google's results I find are better - a common issue I've found it DDG's results are just too general for quite specific searches - .e.g the top result is teh homepage of a brand when I search for something more specific like "brand something something" (sorry i can't come up with a great example right now).
I still use DDG as I can use the rest of the bang syntax to get to wikipedia et al, and for certain things the results (e.g. code/technical) are clearly superior to googles.
I hope DDG logs all the times I go to DDG, then add in !g, to get some great analytics on when I found the search results lacking!