mac.column.ted: Searching for good searching

Ted Landau

November 2008

How is it that the Web sites of many of the world's biggest and most technology-oriented companies have such pitiful search capabilities? Searching these Web sites can be a bit like taking off in a jumbo jetliner and discovering that the aircraft's primary navigation tool is an AAA TripTik.

I first added a search capability to MacFixIt back in the 90's. It didn't cost me a penny. I got it from a site that offered the code for free. As crude as it was, it worked better than some of the search engines I have dealt with in recent months. The problems with most of these search engines boil down to one of two extremes: either they provide too many (mostly irrelevant) results or they provide no (or almost no) results at all. Either way, you don't find what you're looking for.

At the one extreme, the search facilities seem incapable of separating wheat from chaff. Typically, these search engines have only a bare-bones set of features, failing even to include Boolean searching or the ability to search for an exact phrase by putting the text in quotes. On such sites, if you searched for "Apple TV," you might wind up with dozens of results. Unfortunately, they would mostly be pages that match only the term "Apple" or only the term "TV" or, at best, both "Apple" and "TV" but in separate unrelated paragraphs. If you're lucky, buried in all of this might be one result that actually talks about the Apple TV device. Finding it, however, could be harder than locating Osama bin Laden.

At the other extreme are search engines that too often come up empty, citing "No results found." Or perhaps they'll grudgingly offer one or two hits, which would be fine except that the results bear no relation to your search. Imagine if you went to a site devoted to current television shows, searched for "CSI" and found no matches. You'd be justifiably shocked. Surely, at least one of the CSI shows must be mentioned on at least one page of the site. In fact, you may recall having been to such a page on a previous visit to the site. As it turns out, your memory is correct. The problem is that the search facility is unable to give you the existing links. This example is made up, but I have had real-life situations that were not much different.

As a troubleshooter, I spend significant time searching the support pages of various technology sites. Somewhat surprisingly, these sites are far from immune from the sorts of search failures just described. Sometimes, the problem is that these sites don't have adequate support information in the first place. However, more often, the search facility is the primary culprit.

For starters, take Microsoft's Mactopia Help and How-To site. It could serve as an example of problems at both extremes. As a semi-random test, I picked Mactopia's Entourage Support page. Things looked bad right from the start. There were no sign of any "advanced" search options. Indeed, I quickly confirmed that putting search terms in quotes had no effect.

My first search was for "mail does not open." The search engine came back with 20 matches. The top five matches were: Set up an e-mail account automatically; Create and send a message; Verify and rebuild the Entourage database; Set up a POP e-mail account manually; and Label a message as junk or not junk. Only the database rebuilding article even vaguely related to what I wanted to know. However, the text of this article did not actually include my search terms. It was hard to figure out why any of the articles turned up in the results.

I tried a second time, using a different variation: "can't open message." The top response now was Entourage and Exchange: powerful partners. Optimistic, but hardly what I was hoping for. Looking down the list, I eventually found one article that dealt with an inability to open/view attachments. Close enough to be taken seriously as a match although not a direct hit. Could it be that Microsoft truly had no articles on my sought symptom? Perhaps. Still, if that's the case, I'd rather accurately get no results than a bunch of useless ones.

For my next test, I made sure in advance that there was a relevant match to be found. As it turns out, Microsoft's main (not Mactopia) support site is not just for Windows; it includes support for Microsoft Office 2008 for Mac. Unfortunately, a Mactopia search apparently doesn't check the main site (which is a problem all by itself!). The main site, however, does have its own Microsoft Support search engine. This is what I now intended to use. Perusing the main site, but not doing any search, I eventually came across a list of Office 2008 for Mac support articles. I picked one of the articles, titled You experience some issues when you use Entourage 2008 for Mac after you use the Time Machine backup tool. The question now was, assuming I didn't know this article existed, could I find it from a search? I entered "Time Machine backup" as the search terms. Given that these words were in the title of the article, I assumed I had placed a good bet for getting a match. I was wrong. Only one article came up in the results. And it was a useless article titled Description of Office 2008 for Mac Service Pack 1 (12.1.0).

I don't mean to pick on Microsoft. Many other corporate Web sites aren't any better. For example, I have had similar headaches at Comcast's Web site. In this case, I searched for an actual symptom I was having: when I attempted to access On Demand, I got a "communications error." I tried Comcast's search engine both with the terms "on demand communications error" and "on demand error." The former term was a dead end; the top result was Comcast and Best Buy Offer The Ultimate HD Movie Experience At Home. The latter term yielded a more promising result: a top item titled Where can I get help for Comcast Games on Demand issues? But even this was ultimately unsuccessful. I eventually wound up at a page with a link to a useful-sounding article; unfortunately, the link didn't work.

Entering other search term variations fared no better. When I tried "On Demand trouble," for example, I got 85 results, none of them relevant. The item that made it to the top this list did so because it contained the following sentence: "Available July 30th On Demand, Harold and Kumar get in trouble trying to sneak a bong onboard." Sounds like fun, but not likely to resolve my communications error.

I finally stumbled across an AskComcast page. Using it, I asked "What is an On Demand communication error?" This, at last, yielded positive results, pointing me to an item that described an "unable to process request" error. Reasonably close to what I wanted to know. I counted this as a success, but it should have been much easier to get here.

Apparently in recognition of the anemic quality of their in-house searches, many Web sites have switched to using a custom version of Google for their search facility. A wise choice. Actually, if you simply go to Google's own site and search from there, you'll likely get better results than from the non-Google search engines on these other sites. For example, when I entered "On Demand' Comcast 'communications error'" at Google, all of the top hits pointed to the solution I sought -- although none of them were from the Comcast site. This can be disappointing if you really want an official statement from the vendor, but is otherwise much better than the alternative.

Do the people responsible for these inferior search engines ever try to use them? Are they aware of how pathetic they are? Do they realize how difficult it is for the typical visitor to actually find the information they seek, even if the information is on the site? Do they even care? Or is it all part of some intentional and cynical plot? I don't have the answers yet. But I do know that I won't find them by searching these vendors' Web sites.

Resources