So, Live Search is availiable on Facebook. What does that mean from a search visibility perspective? Even though Live Search has a very small take on the search market, this development will bring a lot of search users their way. With over a hundred million users, Facebook represents a great resource for Live Search. Showing organic results within Facebook is certainly going to have an impact on Live Search’s overall performance.

Let’s take for example the UK user base on Facebook, which is about 8.3 million active users per month. If you’re a UK company, looking to show up in organic search results, or sponsored search results, this is a great chance to get exposure to a concentrated audience.

Let’s say the usage rate for web search stabilizes at about 10% of Facebook users, projected on UK users, that’s about 830K active search users per month. If we factor in the percentage take of particular search patterns, we’d get the number of potential visitors to our UK based site. Sounds like a small number, but when you think about it tens of thousands of potential visitors who’d be searching for targeted search patterns is a significant source of visits, and with a moderate to significant probability of conversion. If a company offers products or services globally, the numbers escalate dramatically.

Question is whether Facebook users will indeed adopt Live Search based results on FB as their primary source for search needs, at least if they spend a lot of time on Facebook. If that proves to be the case, then Microsoft may just have found a way to increase their presence on the search market, even if by just a little bit. It will be interesting to see whether organic resutls and sponsored results get higher conversion when shown on Facebook. It’s time for testing, and I have a feeling some interesting results are going to show up.

Also see these posts:
Facebook blog
Livesearch blog

Share This

Cuil has been live (or down apparently) for days now. While I immediately formed an opinion about the interface and result presentation, I wanted to keep using it for my own day-to-day searches in order to get a more complete picture about its search value.

Basically, I expected a lot more from all the information I read about it prior to launch. The results are presented in a grid, which would be ok (easier to scan results) except that the grid is actually a set of columns which resemble a newspaper page much more than an organized set of information. Since grid blocks are not the same size, quick scanning is more difficult, as you have to scroll up and down to see the titles and descriptions. Furthermore, since the results are organized in two dimensions, the usual linear approach of ranking results is absent, so I expect users to have a little trouble determining which result is deemed more relevant compared to others, although this is not necessarily bad.

That aside, they kept the right margin space for advertising options, which aren’t there and probably won’t be for a while. This I think was a bit too quick. Why not offer more results on the page, and introduce paid results later? Since the results are two-dimensional, there is no clear way to distinguish a result that ranks better, and is therefore a desired position over other results. Google (and the rest) give linear results based on ranking priority, so the vertical component is much more important for ranking. Maybe a bit too soon to beyond this, since Cuil has only just started and is probably going to experiment a lot (hence the 2/3 column kayout option in bottom left of the screen).

The “Explore by Category” that appear for certain queries are neat, and actually do return interesting links towards content/results that may help me drill down further on a topic. But that’s pretty much in vain, since many of the queries I performed were answered with “no results found”, which wouldn’t be very strange, if it wasn’t for the fact that they boast billions of pages in the index (120+ billion!). Long tail search doesn’t seem to work, and basically if you type in more than 2-3 words, you’ll get scarce results if at all, but I am seeing improvements regarding this over time. Again, something that will probably change over time, simply too early for a more firm opinion. Sometimes, Cuil reports incorrect number of results compared to what you’re shown - for example “pamp for uiq symbian” returned 3 results, but showed only 2. Another bug?

Some of the queries returned results in multiple languages - for example French and English - all on the same page, side by side. I speak a few languages but if you are researching something, one language is really enough. It would be better if the application offered a choice to see results in other languages, if for example I didn’t find anything useful in English. This, however, didn’t happen too often, so I guess this was either a bug, or an experiment.

Notably, there is a clear absence of rich media results, or at least I didn’t get any, but I’m sure this will also change in the future.

Then there are numerous reports about the service being down due to overload. I haven’t experienced their down time intervals, but have frequently come upon the “no results found” scenario already mentioned. This is probably a start-up fluke, so I don’t expect this to happen too often.

Cuil could be an interesting new search option, especially since the people behind it certainly have the necessary experience and knowledge. I expect it to develop over the next months, and the real picture about its performance compared to ther search engines will be more complete then.

Share This

I have just read the BrowseRank document released by Microsoft regarding their work on a new alternative to PageRank and TrustRank.

Basically, the whole ranking algorithm comes down to giving more value to visitor behavior, as opposed to simply counting and valuating the number of links pointing towards a web page. Visitor clicks and time spent on a web page are key factors in determining its ranking potential.

The document explains the mathematical theories behind BrowseRank, and although this is more useful for search engineers, I got a pretty good picture of where they are going with this. The point is this is just the beginning, and there is a lot more work that’s going to happen before BrowseRank gains real validity.

Since it relies on recording user behavior, there are small obstacles in front of BrowseRank.

First of all there is the privacy issue (users have to opt-in for data to be gathered), and although their document states they used data from millions of visitors who opted in for the purpose of research, the question remains whether BrowseRank can actually work in the long term if it has to rely on users opting in to be tracked all the time. Not that we are not tracked already without permission by numerous web sites already, but that’s another story. I suppose social networking data can also be used to this effect (social bookmarking perhaps), but the other dimension is still critical - how will they manage to record time spent on pages in an easy and unobtrusive way?

What the document also mentions is that BrowseRank is a more valid measure of a page’s importance since it takes into account clicks and time spent on a page, and is thus more immune to manipulation techniques already present with PageRank. It is believed that a person will not click to a page and spend time there unless there is good content and their need for information has been answered. I think there is room for manipulation here as well, since you can theoretically engage an army of users to use a search engine, click on a specified result and go spend some time there, and gather “points” for a particular page to “help” it rank better for related search queries. Naturally, this is a very naive scenario, but quite plausible.

Nevertheless, BrowseRank is an interesting evolution of search, and is quite welcome in my opinion. Is it going to give Microsoft (Live Search) the edge to ride the same waves with Google and Yahoo? Well, I am just not sure. I think it’s going to take a lot more than that.

A really independent and safe way to rank content on the internet has yet to be found. Basically, if it involves people and their behavior, it’s going to be open to manipulation. Maybe we’ll get there some day, but not just yet… Anyway, my two cents.

Share This

So, Live Search has rolled out a new look for its home page. Well, it’s different, even interesting, but hopefully not the only thing they are going to change in their quest to improve their presence in the search arena.

In addition to this, Microsoft have published a document explaining their approach to ranking pages with a new rank criterion called BrowseRank (as opposed to Google’s PageRank, which they say is inferior to BrowseRank). Basically, BrowseRank is all about searcher online behavior (clicks and time spent on page) as a more accurate measure of a page’s importance, rather than simply relying on number of incoming links and on-page ranking factors. The BrowseRank document is available here, so check it out.

Share This

Viewzi, the search engine that allows its users to choose with what kind of view they would like the results to be presented to them, is open to the public for use as of today. Basically, you get to decide what kind of results you’d like to see depending on what you are looking for (if your search query is “Bob Dilan” then you choose whether you want to see text results, or music files, biography, video). Results are aggregated from various sources on the Internet. So far, only those who have registered to be beta testers/users could use the engine, but now anonymous usage is open (no login needed).

It’s an interested search experience to say the least. I’ve used Viewzi during restricted beta for my own search needs to find things I’d most likely be using Google for for comparison’s sake, and found that majority of results shown to me in various views are quite satisfactory. Then again, if you are a Google SERP nut, you’ll probably have a hard time adjusting to a completely new way of seeing search results on this site, as the interface, although easy enough, does require some getting used to, especially, with years of [enter search engine name here] SERPs. Basically, after a small leap of faith, everything comes together just fine. I think Viewzi, as a fresh approach to searching, is definitely going to stay on the search horizon in the future.

Share This

For years now, every now and then when checking for a domain name, I tried using Google just to see if it will serve me a direct result pertaining to basic domain registration (registration and expiration date for starters). Somehow it seemed logical that a simple “whois …..” query in Google should do that. If queries like “define: …” return definitions for words, why not do the same for domains. Well, Google now returns basic domain information if you type “whois [domainname]” into the search box. I realized this a couple of days ago.

Today, I saw a post on the domain information query on Google by Danny Sullivan on Search Engine Land, and from the comments it seems other people are seeing this as well, but the feature is still sporadic, as some people are saying they aren’t getting domain info in their results. I’ve tried queries for .com and .co.uk and they worked alright. It doesn’t seem to work for .yu/.co.yu (now .rs/.co.rs), which are Serbia’s domain extensions for example. I wasn’t expecting this to work but thought I’d try anyway. Others are reporting they are still not seeing anything for domains ending with their own country extension. A post on Matt Cutts’ blog showed up a couple of days ago, which points out this new feature as well. It seems Google is using Domain Tools for domain information queries, and since this is still kinda fresh, bugs and limited availability are expected, until the addition is rolled out worldwide.

Share This

Just read this on Webmaster Central Blog - Google is testing the latest approach to content discovery by crawling HTML forms on a selected group of sites deemed to be particularly useful. Googlebot will, upon coming onto an HTMl form, determine whether form method is GET or POST. It will actually proceed only if it’s a GET form, since they want to avoid crawling forms that may require user information input (such as usernames and passwords), which all use POST. The bot will actually “fill in” text fields with words (found on the site), choose options on radio buttons and select menus, and try to crawl resulting content, and index it if it determines it’s useful and hasn’t previously been indexed.

Anyway, the experiment is an effort to try and crawl/index, what has been called “invisible content” or “the invisible internet” for years, which is a previously untapped rich source of information which over the years has stayed hidden behind processes requiring human interaction.

Share This

This is interesting. I came across this post by Cabel (Cabel.name), according to which advertising search seems to be the new direction in advertising, at least in Japan. But wait, not the search advertising you’re probably thinking of…

How about that? Cabel says it makes sense since all the good domain names are gone, and almost no one will go past the first slash after .com or .net anyway. So, why not advertise the search query that will lead the user towards the brand, product or service?

Most people turn to the search engines in order to find something anyway, even if they know the direct URL to a site (it’s still entered in the search engine query field). But there are a few problems with advertising your brand using a search query. Google owns almost two thirds of searches on a daily basis, so your advertising initiative will most likely have the user go to Google. In that case you better own the first result page, which is plausible for brand searches, or searches that are very specific to a product or service directly related to your business, but pretty unstable if you’re advertising generic or unbranded terms. If you rank 1st for a generic term and you are stable on that position, than the risk is worth it. If Google really does record click-through rate for organic search results, than an offline campaign like this could increase the number of searches for a particular term you rank well for and attract even more clicks, maybe thus cementing your number 1 spot even more. Hm, I only wish it was that simple…

Then again, the interface shown on the advertisements may be well known to belong to a local search property used by the local community, in which case it’s easier to make sure you really appear as the main result for the advertised search query. Definitely worth experimenting with. If integrated well with an online campaign, this could actually work well, for example in the UK market. Get a priority listing on a local search engine, make sure you appear on top for that particular query, and then advertise your search query locally. It shouldn’t cost too much, and the experiment result, whatever it may be, will be worth it.

Share This

As part of their initiative to open-up their search platform, Yahoo! Search is opening its search platform to third party content providers and re-designing the way users will be presented with search results. So far the results were the standard title/description combination, but the new result format will soon include user generated data such as reviews and images, as well as other useful information such as map directions, contact information, price range of services and products (if available). They state that sites like Yelp and New York Times are examples of how additional data can be incorporated into search results. Any site that can offer additional information about a particular search result can take part in shaping it for the searcher. This way Yahoo! is hoping to deliver a richer and more useful search experience to the searcher, and higher quality traffic for websites appearing in search results.

Here’s a graphical comparison of what the changes will look like:

before

after

“We believe that combining a free, open platform with structured, semantic content from across the Web is a clear win for all parties involved — site owners, Yahoo! and most importantly, our users. And by the way, users will be in complete control of the experience and will be able to turn off anything related to open search if they so desire. Over the course of the next few months, we’ll be talking more about how this platform will work and what it will enable.”

I think there’s plenty more to be seen before any final conclusion can be made about the new Yahoo! search result concept. It looks promising, but for now, it’s safe to say this is another example of the future of search and why it’s more important than ever to get a site/business listed and reviewed on portals, directories and content sharing sites.

Share This

Live Search introduced some improvements to their search bot, mainly concerning bandwidth usage and crawling issues. Two major changes have been introduced:

1. HTTP Compression - static files and applications responses are now being packed using HTTP compression which reduces network load
2. Conditional Get - the bot will not download a page again unless it has been modified since the last crawl session; this is regulated according to RFC 2616 specifications

The user-agent signature has been updated to show msnbot/1.1.

Share This