Biznology Blog: April 2005

« March 2005 | Main | May 2005 »

April 27, 2005

20 Silver Bullets for Site Optimization

Jim Sterne is a prolific author on Web metrics and the organizer of the well-respected Web Analytics Conference held each June in Santa Barbara. Jim is a well-known speaker who used this session to explain his top tips for helping your Web site meet its goals. I had heard a lot about Jim but this was my first time hearing him speak—he is both wise and entertaining. I thought that I wrote down all 20, but I missed a couple—sorry.

Jim explained each of the 20 silver bullets:

1. Start with your site's goals. In many ways this is the key bullet—it won't matter what else you do if you don't understand the purpose of your Web site. Is it online sales? Offline sales? Leads? Signing up volunteers for your charity? Donations? If you know what you want your visitors to do, then you can start the process of optimization.


2. Focus on the buying process, not the selling process. Stop thinking about the sales process and start thinking about things from your customer's perspective. If you take your visitors' point of view, then you will more easily pesuade them to do what you want.


3. Create personas. Use personas of your customers—specific personalities that are archetypes of your real customers (your most profitable customers, preferably). Think about your Web site one person at a time—each of them has a task or a goal to accomplish. Their expectations are set by Amazon and Dell and other companies that really do a good job meeting visitor needs. That's what you need to live up to.


4. Be a customer. Try using your own Web site using mystery shopper techniques. Send in an e-mail on the contact form. What music is playing when you're on hold? You'd be surprised what you find that you were completely unaware of.


5. Get organized. Who's responsible? Who is literally responsible for every part (even every page) of your site? You need to know who is accountable for everything you do so that you know who to ask to fix things that are wrong. You probably have several different functions you need accountability for:
   • Content: Corporate Communications handles company history, HR handles jobs, marketing handles press releases/trade shows/white papers, product management handles all product-related information, and customer service people handle problems.
   
   
   • Design: Designers and Information Architects should handle all of this, but do they or is your company not so well-organized?
     
   • Technology: Your IT department and your hosting company handle this, but the browser wars are back! IE vs. Firefox, yes, but PDAs, cell phones, Blackberry—do all these devices work on your site?
   • Schedule: It might be a project manager or maybe its your Webmaster, but someone is in charge of keeping your projects running on time. Who is it?
6. Define a process. Someone must know how the site can be updated and be responsible for getting it done.


7. Make it functional. Forbid flash. People will spend only so much time on your Web site. IBM has an especially good 404 page because it apologizes for the error and helps you get where you're going—you should do that, too. Do usability testing and make sure that your customers can do what they want. If you think you can't afford expensive testing, just pick out a few people from outside your company and set them loose on your site.


8. Offer a site map and site search. They are essential, and they must be done well. (How about checking to see what people are clicking on from the site map and putting that at the top or making it part of the main navigation?) Boring is better—sacrifice everything for clarity.


9. Trim the trail. Kinko's shortened their navigation process and raised conversions by 25%. Is yours too hard to follow?


10. Crisp up the copy. Describe what you do in clear terms. Don't take the paper copy and put it on the screen. Short and sweet works on the Web. Personal voice. Liberally sprinkled with calls to action.


11. Optimize the source of traffic. Test and measure. Over and over again. Go for constant incremental improvement. Amazon says: "Data trumps intuition" Where do visitors come from and do they convert?


12. Optimize the landing page. Don't send them to the home page. Use A/B tests to see which versions work better.


13. Optimize duration and depth. For some sites it's better to have long duration (like Yahoo! for advertising), but for customer service, longer can be worse.


14. Optimize the path. Check recency, frequency, and abandonment—increase your offers right before they would normally abandon, but don't offer the people that you are 98% confident they will purchase anything—they will buy anyway.


15. Optimize the copy. InterContinental Hotel used to do the big bang—they addressed the surveyed issues, they usability tested, and then they spent $10M and—nothing. They saw no improvement. Then they started to tweak their copy and re-tested and suddenly—$20M in revenue. Analyze the things that people put in their cart and see which ones purchase or abandon and make offers to increase conversions. Don't overlook the basics.


16. Optimize the conversion. See when people are abandoning. Are you attracting the wrong customers? Describing your site wrong? Prices too high? Check to see where in the Web site visitors are abandoning and address the problem to increase conversions.


17. Optimize customer service. This has a huge impact on your brand and should be measured by a lot more than cost avoidance. What do you measure? Are you solving problems? Accurately? Are your customers happy?


18. Optimize customer satisfaction. Do surveys. IBM has a good survey because it is short and easy to understand and it does a before and after—What are you trying to do (before) and did you accomplish your goal (after). If not, why not? Did they like the product? service? The site? The experience? A great question to put on your survey: If you had a magic wand, what would you do to make the site better?

The 20 silver bullets are all about testing, measuring, and trying something new. Is your Web site about your company or your customer? Jim's talk was so down-to-earth and practical that I walked out thinking about IBM's Web site (my day job) and my personal Web site in a new way. If you have a chance to hear Jim speak, do it.

Posted by mikemoran at 9:44 PM | Comments (0) | TrackBack

April 25, 2005

What Are Searchers Thinking?

This interesting session at the AD:TECH conference in San Francisco featured three different research studies highlighting the way searchers behave. The first study featured an eye tracking study of exactly where searchers look on a search results page, and was the most interesting. The other two studies focused on why searchers click on various links and what traffic analysis reveals about purchases from search.

Gord Hotchkiss, President and CEO of Enquiro, moderated the session, reminding attendees that it's easy to do search marketing "halfway right" but this session can help you cover the rest of the way. As more and more search marketers get into the game, it becomes important to understand what searchers are thinking and doing to give you the edge.

Greg Edwards, the CTO of Eyetools, collaborated with Enquiro and Did-It.com on an eye tracking study of searchers using Google. First unveiled at the Search Engine Strategies conference in March, the study used 50 people and generated "heatmaps" of the screen areas that more people looked at, with red areas showing more focus of the eyes (with an "X" for a click and a red line for the bottom of the scrolled area of the page). Eye tracking is a very natural environment for visitors (making the results more believable) and answers a question that is critical for search marketers: What are searchers looking at on a result screen? And what are they not looking at? (They won't click on something they never look at.) This study reveals a "golden triangle" in the upper left of the page where the top two or three organic results are shown, with some additional focus on the top paid search result. Only about 60% of the people scroll below "the fold" (the part of the page that is off the bottom of the screen when first shown).

Greg believes that Google loads up the page so fast that many searchers start out looking at the end of the title of the second search result, which is where the entry box from the search screen is—then they move to the upper left corner. Because of this, the title of the second result may actually have a slight advantage in visibility. (Greg did not say this, but it stands to reason that people using toolbars probably move their eyes to the upper left corner without looking at the middle of the screen first.)

Greg told us that the visibility of the right side results is far lower (30%) than the paid results that are above the organic results (90%) and the top few organic results themselves (80-90%). If there is a "OneBox" (news, book, movie, etc.) at the top of the results, the OneBox is very hot but the top organic results are just as hot—in this case the golden triangle just extends further down the page.

When people return to a search results page, according to Greg, they don't seem to look at the right-side paid results any more than they did the first time, but they do look further down the search results page for more organic results, with results above the fold still getting a big edge. The full eyetracking study report will be available on May 23.

The next speaker was Akhilesh Bajaj from the University of Tulsa, who studied what motivates a searcher to click on a link on the search results page. The study looked at several factors:

• how closely the title matches what the searcher wants


• the rank of each result


• the reputation of the seller/information source


• how closely the snippet matches the what the searchers wants

The goal was to determine which factors were more important than the others, realizing that there might be differences depending on whether the searchers were informational or transactional (looking to buy) searchers and how "involved" each searcher was in the search scenario. "High" involvement means that it is very important to the searcher that the right information is found, perhaps because of importance to family or health, or because an expensive purchase is being contemplated.

The experiment was designed to use examples of different types as searches with results varying according to the various factors. The study is still being analyzed, but preliminary results are that title match drives 50% of the clicks, with the snippet causing 30% and rank 15%. Surprisingly, the reputation of the content source seemed not to matter, most of the time, which is good news for sellers with unknown brand names. Another unexpected result is that the type of query (informational or transactional) does not have much effect, nor does searcher involvement.

Akhilesh believes that as search engines improve, rank will become more correlated with clicks because they favor results with the title and snippet that best matches the query—so ranking correlates to clicks but are not the cause of the clicks. The cause is that the titles and snippets match better, which was shown by searchers avoiding top-ranked results in the study with bad titles and snippets in the study. So, searchers do not click on results because they are #1 for any reason other than that they are better matches.

Akhilesh says that search marketers should use focus groups to find out what they are searching for, and they should design titles and descriptions that match the best—he believes this will lead to higher clickthrough, based on their study. I think using keyword planning tools and ensuring that your titles and snippets reflect those keywords is just common sense which this study confirms as the correct approach. I don't think you need to do focus groups—you just need to follow existing best practices.

Jarvis Mak, a Director at Nielsen/NetRatings, was up next. Jarvis presented Nielsen's results from studying search behavior from searchers in their Web panel. Jarvis explained that government and mass merchandisers are the sites that get the highest percentage of visits from search. Universities and current events sites are also quite high in this metric.

Jarvis said that 50% of search sessions had just one search, and 35% of those sessions looked at just one search result screen. As searchers do more and more searches within a session, they are far more likely to look at a second page of search results, and even a third page. These are the distinct minority of searches, but it does show that when searchers are intently searching for something they use more queries and view more results pages.

Google Desktop, Jarvis noted, already has over a million users and 17% are filtering by files and 16% by e-mail. Jarvis believes that Apple's and Microsoft's coming move of building desktop search into their operating systems could change searcher's behavior dramatically—he said that future surveys will look for these changes.

Jarvis found that the top 20% of online consumers account for 88% of all e-Commerce spending, but they actually search a bit less than others. It could be that searchers are actually spending less than non-searchers (which Jarvis believes), but I think that it could easily be that those searchers are buying offline. Jarvis says that 30% of new customers come from search, but that 70% come from existing brand and merchant loyalty or offline drive to Web campaigns. (It is also possible that they once found the Web site from search and have now returned.)

Jarvis believes that search moving to the toolbar and to the desktop is increasing the number of searches, and can help keep per-click prices lower as supply of searches goes up. This is an interesting viewpoint that I have not seen before—given the rise of prices over the last few years, perhaps supply is not increasing as quickly as demand for search paid placement is rising among advertisers.

Search marketers are increasingly interested in studies of searcher behavior, so expect to see more reports like these three in the future.

Posted by mikemoran at 10:54 PM | Comments (0) | TrackBack

April 12, 2005

How Do You Influence the Search Results Title and Snippet?

In April's Biznology newsletter, find out what you can do to get the title and description (snippet) you want to appear in the search results for your Web page. Where do the search engines get what they display? What can you do to exert some influence over what they decide to show?

Posted by mikemoran at 10:22 PM | Comments (0) | TrackBack

Click Fraud: The Inside Story?

The click fraud news just won't quit. Since I wrote my March Biznology newsletter, we've seen court cases against both the search engines and the alleged defrauders, but now there's something new. Joe Holcomb, a Senior VP at BlowSearch, has written an entry in his personal blog that reveals some scary possibilities.

Joe is an insider in the paid placement industry, and as such his opinion carries more weight, but it is still just one voice. Nevertheless, his post may confirm some of the fears of search marketers that click fraud really is spiraling and may be as hard to stop as I speculated in my March newsletter.

Joe baldly states, as I believe, that click fraud from competitors is not nearly as big a problem as people think. Joe says that automated click fraud is a huge problem and that search engines don't really want to dwell on that.

Joe further charges that search engines can't stop click fraud and know that they can't, and that they handle individual complaints from advertisers but don't issue wholesale credits to others based on reports from a few. It is frankly not in their business interest to do so. He thinks that the estimates of 20% of all clicks being fraudulent may even be low!

Joe may be biased in his role with BlowSearch, but he disagrees with my assertion that the larger search engines do a better job of policing click fraud than smaller players—Joe claims that none of the vendors have a handle on this growing problem. He believes (and it has some logic to it) that clever defrauders may be able to hide their fraudulent clicks more easily in the larger traffic volumes of the most popular engines. But he does say that the smaller players, such as FindWhat and Kanoodle, that syndicate their traffic to other sites may be more susceptible to click fraud. (Blowsearch does not do as much of that, as you might expect.)

In addition to the vendors that I mentioned in the March Biznology newsletter, Joe also mentions three more vendors that can help monitor click fraud for advertisers (I haven't looked at any of them): TrackingROI, Clicklab, and Alchemist Media.

Read Joe's post for yourself—it is worth your time if you are concerned about click fraud.

Posted by mikemoran at 7:55 AM | Comments (0) | TrackBack

April 7, 2005

Google and MSN Search Experiment with Titles

It's big news when Google changes anything. Google is constantly tinkering with making results appear more relevent—they were the first to extract text from your page to show the snippet under the title on the results screen. And Google is now starting to pull descriptions and even titles from your Open Directory (DMOZ) listing. Garnering less attention, however, is MSN's use of LookSmart for some titles and descriptions.

Some say that Google has used descriptions from Open Directory for a while, but I have to admit that I was unaware of it. But everyone agrees that Google is now taking titles from ODP at times, which is definitely new behvior. The Search Engine Roundtable has a good synopsis of what Google is up to.

Less discussed is the way MSN Search uses LookSmart for titles and descriptions. I thought MSN dropped LookSmart back in 2003, but there must be some relationship still afoot. Check out this query for "IBM" and look at the title and description. They are both taken straight from the LookSmart directory.

Google changes get a lot of attention, but MSN Search carries about 15% of all search traffic, so what it does is important, too.

If you are interested in more information on how search engines construct titles and snippets, check out my Biznology Newsletter this month.

Posted by mikemoran at 7:23 AM | Comments (0)

April 5, 2005

Click Fraud Enters the Courtroom

The words on the March Biznology newsletter are barely dry before we start seeing a whole new side of the click fraud wars—order in the court! Two notable lawsuits have been filed to make case law where there is no written law.

The first story is actually an old story—but it was not widely reported back in November when it happened. CNET reports that Google has filed suit against a company allegedly perpetrating click fraud. Many people predicted this one coming, but it is not clear if click fraud is illegal, or whether it is solid grounds upon to which to win a lawsuit.

Most people would consider click fraud unethical, but despite the word “fraud” in its name, it’s not clear that it breaks any laws. Click fraud artists are clicking on Web pages to cause per-click fees to be charged when they have no intention of ever buying the item being marketed, but it’s not clear that there is any duty for Web surfers to be serious buyers in order to click a hypertext link. You could just as easily argue that radio listeners are defrauding advertisers if they listen to the programs but don’t buy anything advertised. The point is that neither radio listeners nor Web surfers are obligated to ever buy anything advertised within the content they consume. I am no lawyer (I don’t even play one on TV), but I wonder if just because the content provider expects people to buy something whether that is enough for a court to find fraud.

The second case may be the more interesting one for search marketers. A potential class-action suit was filed back in February against Google, Yahoo!, Ask Jeeves, and others on behalf of paid placement advertisers. The suit is currently undergoing certification as a class-action lawsuit, which, if successful, would allow any search marketer that met certain court-prescribed criteria to join the lawsuit as a plaintiff.

This case argues that the defendants (the search engines) should be liable for damages because they have failed to prevent click fraud. From my reading of the news reports, it appears that the engines are being charged as conspirators, which seems like stretch to me. You could imagine a court possibly finding for the plaintiffs on grounds of negligence—where the search engines are found to have been lax in policing click fraud to the detriment of customers to whom they have promised a service.

But conspiracy? Hard to imagine that the search engines all got together and looked the other way as click fraud occurred, or worse, were parties to perpetrate the fraud in the first place. And even if they did (which I don’t believe for a minute), how could anyone prove it in court? Conspiracies are notoriously difficult to prove—you see racketeering cases drag on for months in the face of thousands of mind-numbing details. Can you imagine the e-mails that would be subpoenaed if this case goes forward?

Again, I am no legal eagle, but I think this case is not the one we are waiting for. Let’s see what happens when a class-action negligence case is filed. We probably don’t have to wait very long for it. In fact, it may have been filed already—both of these cases were filed for weeks before they drew wide publicity.

As I said in the March Biznology newsletter, I think that the whole per-click model is the real problem. Maybe changing the market to pay-per-action would take the incentive for fraud away, so that the market can solve the problem instead of the courts.

Posted by mikemoran at 7:33 PM | Comments (0) | TrackBack