Back in March, after Ask.com seemed to be pulling away from search to some people, I wrote that Ask "killed the search engine," was no longer an innovator in search, and couldn’t expect to compete against Google. Recently, I spoke with Ask.com CEO Jim Safka, who hoped to change my mind. He failed to do that, in the end.
Safka’s been reaching out to a number of the "digerati" who came out against Ask after the March news. Lisa Barone, who wrote about giving up on Ask, was contacted. Danny Sullivan’s had a call and plans to write a piece from that. My turn was about three weeks ago.
I was privileged to not just have Jim Safka on the call, but also the co-founder and inventor of Ask.com’s core search algorithm, Apostolos Gerasoulis, who’s known to many as AG. Yes, AG is from the old Teoma search engine, which Ask.com bought as an effort to focus more on the core algorithm. Having AG on the call was something I was looking forward to.
Jim spoke with me while he was traveling through Manhattan. I lost Jim several times while he traveled through tunnels. At that point, I was able to ask AG certain questions, questions that I believe he could have answered better than Jim. But to be honest, I was fairly surprised and impressed by the level of detail Safka was able to discuss, when it came to discussing the core algorithm.
I asked AG about freshness and depth of the Ask.com index. AG said the new infrastructure has improved freshness and, in particular, navigational queries. This is the new Edison algorithm, which I wrote about over a year ago. It finally went live at the end of March, AG said. Edison is a compilation of Ask.com’s core search technologies, including Teoma’s subject specific communities with Direct Hit’s click tracking algorithm layered on top of it.
AG told me Edison is six-times as fresh as the previous technology they were using. It brings Ask.com’s best assets all together to build a better algorithm than what they had before.
Jim then spoke of "geo-bias" solutions, "news injection," leveraging blogs using Bloglines, and more. I won’t go into what we talked about with those components because I believe those techniques are valuable but not necessarily as important to the core algorithm, as would be freshness, depth, and returning relevant unstructured data from the web.
In an effort to isolate specifics, I asked if the May 20th Ask.com update had anything to do with this roll out. AG said that the infrastructure continues to crawl and find new pages and maybe that had to do with it. Jim then added that it may have been a more broad roll out around that time. Ask, at any specific time, may have 20 to 50 tests going on. They frequently roll out these tests, so this is what some may have seen - or maybe not.
Jim Safka’s cell connection was solid now and he then went on to discuss the strategic direction they have been going towards. Jim started, with no surprise, to tell me that Ask.com’s core focus is on being a general search engine focused on core search. He added that Ask.com has the best people working on this mission and they have an advantage in that they are small and can push things out quicker.
Clearly, that was a stab at Google, Yahoo, and Microsoft, but I am not sure how he had the nerve (for lack of a better word) to suggest they are quicker at Google in pushing out new core search technology, but I did not get into that with Jim. He went on to discuss that he was upset to see that the Microhoo deal did not work out, as that would have made them number three.
Ready for the good stuff? Jim said that Ask.com "greatly over indexes in certain categories," such as references, health, and entertainment and hobbies. Why? Because most their users look for those categories and they want to excel at what their searchers are looking for. Ask users are three-times as likely to ask a question than the three other search engines combined. Jim explains this is likely due to the old Ask Jeeves brand and because their name is Ask.com. Jim said Ask will continue to focus more on those categories and be the best in the world at that.
I was thinking through his speech that these "indexes" are purchased database and not likely unstructured data via a normal web crawl. Jim added that to accomplish this goal, they will add more smart answers and improve the related search feature. He also wants to leverage user-generated content and they are working on better ways to crawl and integrate that content into the search results.
I did tell Jim that Ask.com always excelled at Smart Answers and leveraging structured sets of data. I hinted at getting back to the core algorithm. Jim then gave me more clues on terminology they used there.
"Triggering" technology is used to determine "what else shows on the search results page." The "PageWise" algorithm integrates the monetization component of the page, so if showing more ads for some searches makes more sense, such as shopping searches, they will do it. He even said that Google AdWords is much more relevant than some of their search results in some categories, such as shopping specific. I then asked, “What is the most amount of ads would you show? He admitted they will now show up to five ads above the search results. He answered that honestly, so I thought I would hit him up with the big question on all of our minds.
"Have you or would you test Google organic search results in place of your organic results?," I asked. Jim initially gave me the PR speech consisting of, "We are completely devoted to our own search results." I then blatantly asked, "Is that a no?" He said it was a no and that no, Ask did not test Google results in the past six month.
That was the bulk of my conversation with Safka with AG as a wing man. But Jim spoke for 95% of the time, when compared to AG.
I ended explaining that the core search is far from where it should be, even with Edison in place. I told AG and Jim that I track a number of search results, typically the longer tail queries, and 80% or more of those results are not even close to as relevant when compared to the Google’s results. AG asked for me to send him some examples, which I will do. They promised to continue improving, and I promised to continue watching.
Do I think Ask.com is focusing on core search after speaking with Ask.com’s new CEO? Honestly, no. To "overly index" in specific categories does not qualify as a general search engine. To call "indexing" buying or partnering for structured content to place in a database and spit out a Smart Answer or aid the searcher in refining their search to something that they have structured content in, is not core search.
Do I think Jim Safka can lead Ask.com in competing against Google in core search? At this point, the answer is no. Does that mean Safka can’t lead the company in growing market share? No, he may do that – but not by focusing on what is core to the algorithm. My goal for our conversation was to not discuss increasing market share, but to discuss competing with Google on core search. I do not report for Wall Street, I report for the search industry.
To end it off, Ask Not What Is Best For Market Share, But What Is Best For Core Search.