Searcher Behavior Research Update

Apr 25, 2006 • 3:42 pm | comments (1) by twitter Google+ | Filed Under Search Engine Strategies 2006 Toronto
 

Chris Sherman is the moderator of this session. He introduces this as a unique panel, compared to the others. Recently, he said, we have real metrics on search behavior. That is the whole focus of this panel.

Gord Hotchkiss is up first. He is going to talk about his two stage eye tracking study. The first study he showed the "Golden Triangle" in Google. So a search on the blog for "Golden Triangle" for our past coverage of this. Phase two involved, 54 participants, recruited MSN and Yahoo users, applied 5 scenarios (free to completely scripted), interactions on MSN, Yahoo & Google, the study will be out later in May. He showed the eye tracking heat maps of all three engines. People had to go deeper on the page, to look for better results. The ideal User's Search Engine; if it was up to the user, the 1st top most result would be the only result I want and that is it. When you compare Google to MSN, it seems Google is more ideal. They had the same control users for the second study, so they redid the Google test with the same users to be sure. The 2nd study was remarkably consistent with the first study. So they sliced it differently, male versus female, by engine. It is not the searcher that is the variable, but it is the searcher... Are Google more results more relevant then? He said they did an internal study with his employees, that showed MSN was more relevant. He brought up my study, "RustySearch" where Yahoo was most relevant, but now Google is more relevant. The bolding in the title, bolding the keywords in the title and description, makes the listings more relevant. MSN didn't bold the keywords at the time of the study, hence the deeper and longer eye scans at MSN. That may affect this eye tracking study when comparing Google to Yahoo to MSN. He then discusses "Semantic Mapping" where if you do a search for "Digital Camera" and you see brands or attributes of digital cameras, it will resonate with the searcher as either relevant or not. 6.5 seconds was the time it took to click, and 4.5 seconds of scanning the page, to read 6 to 7 search results. Then he discusses "thin slicing" when you view the bolded terms on the search results page. Findings; much more scanning on MSN and Yahoo than Google, Even MSN and Yahoo users seem to find what they're looking for faster on Google, Google appears to be perceived as the more relevant and other....

Lance Jones is next up from Keynote Systems. He will be talking about Google's brand. They did a recent study, to study the brand of the top search sites. First time they are talking about this publicly. Google is now in the Webster dictionary, now that is something, when your company name turns into a verb. He shows a Reuters story of Google's increase market share, from March 28th. Top three brands, Google, Yahoo then Ask.com. Usage, Google, Yahoo and than Ask.com. Satisfaction; Google Yahoo and than Ask.com. So is Google that much better, really? Brand study goals. (1) How important is the brand? (2) Which engine produces best results (3) Which engines produce best paid results (4) Who has the best presentation and layout? He will answer the first Q. How do they take the brand out of the question? He explained the Pepsi versus Coke challenge, blind taste test. They asked people to interact with one search engine. They used an unbranded search interface, and a branded search interface. Pages are live and screen-scrape Google results. Page design is held constant. The Google results were the same on both sites, only difference, was if the name Google was on the page (not even the logo). No paid results on the page. Feedback based on 1,600 queries. 12 distinct satisfaction metrics. Keynote calculates scores from 0 to 1,000 scale. For the unbranded condition, Google achieved 737 score. Some of the panel comments; "why would anyone ever leave Google for this? and many of these type of comments. The branded versions scored an 800, the unbranded 737 - there is a brand factor. Part 2, Google Web and sponsored results., and the same layout, same type of test, just added sponsored results. Unbranded scored 763 and 809 for branded version. Part 3 was Google Full Featured design. They stripped out all references to Google logos and images, but left text and bolding and you see A 753 score for unbranded and 806 for branded version. Google brand acts like a magnifying glass. Google's brand opens doors (google desktop, gmail, finance, analytics, earth, pack, base, etc.). Google's brand keeps the competition thinking.

Michael Ferguson from Ask.com is next up to talk about some of the innovation they went through. He is a senior user experience analysis from Ask.com. He explained that people love Jeeves but they used it for a very fixed reason, very infrequently. So they decided to take the risk to change that perception. Gives the stats run down on Ask.com and shows the sponsored listing program.

Changing Behavior on Ask.com then Trends in Search Use and show the opportunities for search marketing. He shows % of queries by category; 20.8% general reference and education, 15.2% other, 12.1% science and technology, 9.3% entertainment, 8.6% local and so on. Re-branding to change Behavior; Mental model - more occasional usage than competitors, iconic butler locked up with mental model, low awareness of huge suite of search tools on site, it was a risk.

Ethnography; soccer dad, office concierge, urban hipster, academic writer, creative professional, hydro-geologist and recruiter. He shows the topics of those types people listed before, searched on (from widow spiders, foot in mouth, and so on). Users spend much more time than they expected in their search time investment.

Focus on Ask: How can change behavior? So they pulled the butler off, replaced with a bar on the right, says "search tools". What this did, they looked and saw this, it was about the same height as the butler, so they saw it, when the butler was missing. He showed a sample search on "toronto" which has a the "search tools" at the top with a wikipedia listing and image search result. They also have "narrow your search" and "expand your search" (zoom feature, they show up 60% of the time and then clicked on 20% of the time) in place of where most engines have sponsored results. He also shows the binoculars feature. Results; usage of several tools up 10 times, binocular use up 5x, maps up 7x, and related search up 15%. Frequency, retention is changing and query mix and characteristics changing. Google's Test: They are testing expanded results (the blue arrows on the left side). He shows how Google tends to try things that Ask has been doing for a while, innovate here and there.

Opportunities for Marketing: - Search Tools: More Qualified Clicks -- Following the stream through these may change strategy and pricing -- Information you offer to SE will become more rich - Expanding Vertical Use -- Images, maps, local, etc... - Contexts -- Mobile, position in the buying cycle -- Pricing based on time of day, season, buying cycle stage, data from personal and social search will play in. - He then quotes Danny's keynote, "remember, complication is nice job security!"

Previous story: Search Engine Friendly Design
 

Comments:

SEARCH ENGINES WEB

04/26/2006 03:00 am

///The bolding in the title, bolding the keywords in the title and description, makes the listings more relevant _________________________ The problem with this study is that it is a focus on an Attention Getting attribute (bold) --- there is certainly more to "relevance" than the natural tendency of the eyes to navigate towards the "eyecatching" And of course, KEYWORDS, without CONCEPTS are no real dependable measure of Relevance. GYM appears to individually have certain strengths and weaknesses with certain TYPES of searches than with others. Depending on the query, you can find MSN or Yahoo or ASK more relavant than Google and Visa Versa. Once Concept searching evolves - it is possible that SYNONYMS as well as ACRONYMS as well as alternate Spellings will be Bolded. The search intelligence would know that WEB SITE & WEBSITE are the same ....as well as SEARCH ENGINE OPTIMIZATION & SEO could be the same depending on the context.

blog comments powered by Disqus