Biznology Blog: May 2005

« April 2005 | Main | June 2005 »

May 31, 2005

What Will Personalized Search Mean for Rankings?

Andrew Goodman weighs in with some interesting thoughts on what personalized search will mean to search marketers. I agree with most of those thoughts, and want to amplify one that Bill Hunt and I mention in our book, Search Marketing, Inc.—what will become of rank checking?

Andrew believes that the old organic search marketing task of checking the rank of your pages for various search queries will simply disappear in a personalized search world. He has a point. If most searchers are receiving personalized results, there won't be any such thing as a #1 result anymore—each person would get their own #1 result. Andrew expects that we'll have to settle for metrics such as search referrals and conversions.

He may be right. But I can see it going a different way as well. Would search marketers pay search engines to get better metrics? I think the answer is yes.

If search engines know how to personalize search results, they probably have information about the searcher that would allow searcher activities to be broken down by interesting demographic categories. Would search marketers want to know how many of their referrals were from men? From women under age 25? People from France? I think they would, and would pay for this information, just as they want that information for other media.

But what about rankings? Will the concept of rankings just wither away in a personalized world? Not necessarily. Rankings still have value—if your referrals are low, you need to know whether your rankings have dropped or your clickthrough rate is tumbling. You take different actions to correct each problem.

Google and other search engines could sell search marketers the ranking information they crave. You could learn your average ranking for a particular keyword on a certain day. So while every searcher might get a different #1 result from a search, search marketers could find the average ranking for their pages for that keyword.

Moreover, search marketers could find the average ranking for any particular demographic group. Would this be valuable enough information for search marketers to pay Google? We may soon find out, since personalized search is on its way.

Posted by mikemoran at 11:34 PM | Comments (0) | TrackBack

May 21, 2005

Measuring Corporate Search

Having trouble measuring the success of your corporate Web site's search engine? Avi Rappoport of Search Tools Consulting told the attendees at the Enterprise Search Summit what to do, in this speech on Wednesday.

Avi provided several tips about the right things to measure for your Web site's search engine:

• Pages indexed: Check a sudden decrease in pages for an error that is blocking the spider, but sudden unexplained increases can also be problematic—pages that should be password protected may have been inadvertantly exposed the spider.


• Keyword popularity: What are your popular keywords? Keywords tend to follow a Zipf curve, with a few very popular keywords with a large number of keywords used only once or twice a month. Make sure that your most popular keywords are providing good results.


• Keyword trends: Which words are becoming more popular? Which are declining? Your trends sometimes result from changes to your sites navigation—watch for navigation problems that drive people to search.


• No-match keywords: Anytime a keyword yields no matches, you know the searcher was not satisfied. Find the most popular no-match queries and figure out what is wrong. Does the content not exist on the site? Or does it use different keywords? Misspellings? Analyze the problem and take action.

Avi reminded her audience that search metrics are critical to meeting your searchers' needs. You can find what is working and what is not, and you can fix what;s wrong to improve your site search.

Posted by mikemoran at 12:05 AM | Comments (0) | TrackBack

May 20, 2005

What's Multifaceted Search?

If you've never heard of multifaceted search, it's time to learn what it is and what it can do for your corporate Web site's search engine. On Wednesday, I spoke to the attendees at the Enterprise Search Summit, and I've posted a synopsis of that speech with its accompanying charts, called Inside Multifaceted Search.

Posted by mikemoran at 10:01 PM | Comments (0) | TrackBack

May 17, 2005

What is Your Web Site's Goal?

Read May's Biznology newsletter and think about what the basic goals are for your Web site. Learn how to measure its success. Don't be satisfied with counting traffic or clickthrough rates—you need to know if visitors to your site succeed.

Posted by mikemoran at 9:22 PM | Comments (0) | TrackBack

May 15, 2005

Counting What Counts

Attendees learned the basics of search metrics at the recent "Measuring Search Success" session at the Toronto Search Engine Strategies conference, moderated by Chris Sherman, of Search Engine Watch. What should you be counting on your Web site? Read on.

Jeffrey Eisenberg stole the show on this panel. The CEO of FutureNow, a Web consultancy, delivered a message chock full of sage advice. Jeffrey listed the four basic objectives of a customer-facing Web site:

1. Commerce: Increase orders and order size.


2. Media: Attract eyeballs to increase ad impressions and subscriptions.


3. Self Service: Decrease service costs.


4. Lead Generation: Improve lead identification while lowering acquisition costs.

Once you've figured out what your site is about, you need to come to terms with the basic fact that all metrics programs can measure is a click—but they can measure every click that every visitor makes, which comprises a wealth of information. Jeffrey posits that people follow clicks the way a bloodhound follows a scent, citing that 63% of visitors abandon within two pages because they "lose the scent" (OneStat 3/8/04).

So you you need to align your site to meet searchers' goals—you can't just look at end goals (buying your consulting service), but at micro actions (downloading a white paper). Jeffrey discussed several ways to measure conversion rates:

• Overall: Total conversions divided by total visits


• Over time: When it covers multiple visits


• By scenario: Visitors that start a scenario vs. those that complete it

(This last conversion rate, scenario conversion rate, is what I speak of as Web conversion rate in the latest Biznology newsletter.)

Jeffrey left us with some critical questions to ponder:

• What pages are the ones where people come to one page on your site and then leave (single access pages)


• Which are your top exit pages? Some pages are good exit pages (an order confirmation page) and some are bad (your home page).


• How do you define your metrics?


• What are your benchmarks?


• How are the metrics trending? Trends are far more important than absolutes—which can be inaccurate.

Good measurements, Jeffrey argues, come with definitions that everyone agrees to, expectations for what a good or bad number is, a simple method of presentation (so you can quickly see what is going wrong, for example), and a set of agreed-to actions when things do go wrong. These qualities allow your business to respond quickly and appropriately to the metrics that you collect, so that the problem can be addressed. Your measurements, however, depend on your Web site's goals. Retailers may track sales conversions and shopping cart abandonment whereas media sites might track subscriber conversions and site traffic. A lead generation site might track "Contact Us" forms filled out or phone calls from leads.

Laura Thieme of Web marketing consultancy Bizresearch challenged listeners on whether they are retaining their most valuable customers, based on ROI. Laura says that just 19% of search marketers perform any ROI analysis at all, although success rates vary widely from customer to customer. Even if they want to analyze the data, Laura further explained, they may not have the analytical skills required to do so.

Jeffrey also emphasized the need to track ROI, but added that your best benchmark is your own site yesterday, not your competitors today. Jeffrey's bottom line is that Web analytics can pay off—and they pay off as fast as your company can make the changes pointed to by your metrics. The faster you can make changes, the quicker your payback for your metrics investment.

Posted by mikemoran at 9:23 PM | Comments (2) | TrackBack

May 4, 2005

What Else Are Searchers Thinking?

At Search Engine Strategies in Toronto, Chris Sherman of Search Engine Watch moderated this forum on what searchers think and do while they are searching. Gord Hotchkiss made a similar presentation to what he showed at AD:TECH last week in San Francisco. Remember, the more search marketers learn about searcher behavior, the more efficiently you can target searchers with effective keyword planning and copy.

First up on the panel was Gord Hotchkiss, CEO of Enquiro, a search research firm. Enquiro's research into searcher behavior (performed in conjunction with Did-it.com and Eyetools), had 50 participants but tested only with the Google search engines. The study recorded eye movements and clicks and created graphs that corresponded to where they looked on the screen and what they "fixated" on. As we discussed in a recent Biznology blog entry, Enquiro developed "heat maps" to show where most searchers look at the screen, revealing that searchers start with the title of the first result. If it is interesting, their eyes scan to the end of the title. If not, they move down the page to the next title. They may look at the right-side ads, but are far more likely to see the top-left ads and the organic results. The heat map depicting this pattern is dubbed the "golden triangle."

Gord explained that almost 100% of all searchers scanned the top three organic results, but that dropped to 85% for organic results four and five, and to about 50-60% for the results #6 through #8. The #1 organic result received nearly 30% of the clicks, with 20% for the #2 result and 12% for #3, but less than 5% by the bottom of the page. The top-of-page paid results were seen by nearly 100% of visitors, dropped to 50% visibility for the #1 paid result on the right side of the screen, and fell off further from there to 40% for #2 and 30% for #3. When searchers returned to the results page, the scan activity was much more spread out (there is no golden triangle), even though the researchers expected to see more visibility for sponsored ads on the second viewing. (According to Gord, about 50% of search sessions include a viewing of a second page of results, with some of those searchers returning to view the first results page again.)

Gord explained that the study revealed no distinct patterns broken down by age—searchers tend to exhibit the same patterns no matter how old they are. When the full study is released on May 23, it will show the patterns broken down by other demographics that were significant.

Gord described a particular searcher behavior called "semantic mapping" whereby searchers associate many possible search terms with a concept but enter only one or two words into the search box. For example, a searcher looking for a digital camera thinks about many words (such as reviews, megapixel, Nikon, Canon, Kodak, easy to use, reviews, testimonials, 5 megapixel) but ends up typing in "digital camera." But those other words are not forgotten. The searcher scans for those words in the search results, not just the words that were typed. In this way, a #2 result that contains more occurrences of the words in a searcher's semantic map could attract more clicks than the #1 result.

The second panelist was Debra Jaffe, Product Marketing Manager from Google, who amplified Gord's contention that more searchers click on organic Google results than paid, because they consider organic results more "objective." She said that searchers click on the OneBox (news, product, book, and so on) at the top of the organic results at about the same rate as paid results. Debra suspects that searchers believe that the OneBox results are paid, even though they are not. Perhaps Google is considering changing the appearance of the OneBox to try to get higher clickthrough.

Chris Sherman pointed out that savvy search marketers are spending more and more time understanding searcher behavior. The popularity of this session at Search Engine Strategies on top of the similar one at last week's AD:TECH conference is living proof.

Posted by mikemoran at 8:03 PM | Comments (0) | TrackBack