Build Traffic Through Good Site Design (part 1)

Navigation Basics

Many Web site owners concentrate on ways to drive visitors to the site, without giving enough thought to what happens when they get there. The way your site is designed and laid out has a huge impact on the amount of traffic you can build and maintain. In this article, we'll explain how to design a site to maximize traffic. Regardless of whether your site is selling products, content or advertising, and whether it's old or new, an hour spent improving your navigational scheme can be far more productive than an hour spent spamming the search engines.

One thing that people tend to forget is that driving people to the site is a means, not an end. The end goal is either to get them to buy something, or to show them ads. It does little good to bring people to your site if most of them just take a peek at the home page and split.

There's only one sure-fire way to build substantial site traffic over the long run, and that is to make your site as useful as possible. Although content is king, your navigational scheme also has a lot to do with how useful your site is. One of the main reasons people leave a site after only a few seconds is because they can't find what they're looking for, or are forced to click through page after page to get to the content. See ( Designing for Good Navigation) for hints on how to build a logical, user-friendly layout.

Traffic Tips from your Log Files also makes a good companion article to this one. Analyzing your log files can tell you a lot about the paths people take once they reach your site, and point out areas that could be improved.

To some extent, your navigational scheme will depend on the purpose of your Web site. Back in the good old days of course (2 or 3 years ago), the only purpose a site needed was to share information, but nowadays, in any discussion of site goals and purposes, the question of a "business model" will rear its head. This trendy term basically means "How do you expect to make your money?" Now, there may be endless ways to make money on the Internet, but for the purposes of this article, we'll assume that you're either selling a product or service, or that you're selling ads on the site. Of course, some sites do both, but the navigational implications of these two business models differ, so it makes sense to discuss them separately.

If you're selling a product or service, then your goal is to get the visitor to click on the button that authorizes a transaction, in other words, to buy something. Your whole site exists to guide the visitor to that point, so you don't really care much what other sections of the site they visit, how many pages they see, or whatever. If they come straight from a search engine to the ordering page, hand over their bucks and split, that's fine.

Of course, this doesn't mean you want to herd them blindly to the order button. There is such a thing as informing your customers, and in many cases, the longer they spend perusing your product information, the more likely they'll be to choose just the right model, see how much better your product is than the competition, be impressed with your professional-looking Web site, etc. etc. So include plenty of good-quality content on your site (well-organized, of course), but make sure that order button is right there for those who are ready to click it. Don't make visitors click through a million pages to get to the checkout counter. From every page on your site, the ordering page should never be more than one click away.

To sum up, owners of e-commerce sites are more concerned with their "conversion rate" (the percentage of visitors who are converted to customers) than with the number of page impressions generated.

For an ad-supported site, the traffic goals are a bit different. Since ads are sold by the impression (Each time an ad image is shown to a visitor is one "impression.), every page view may be a couple of pennies in the bank. While an e-commerce site wants to send everyone to the order page, an ad-supported site wants to send each visitor to as many different pages as possible. In this sense, an e-commerce site should be designed like a funnel, and an ad-supported site like a maze.

Now, by invoking a maze, I don't mean to imply that it should be hard for people to find what they want. On the contrary, not only should it be easy for visitors to zero in on exactly the type of content they're looking for, but they should be presented with lots of opportunities to check out closely-related content, so they keep clicking off to explore ever-more-fascinating new worlds, the page impressions keep racking up, and the cyber-cash register rings!

Later in this article, we'll discuss how to maximize traffic with internal links.

Navigation that Works

Nothing chases away visitors faster than a dysfunctional site. Dead links, errors, long page load times, fancy scripts that don't work as they're supposed to...Each of these incidents is a penny out of your pocket. Nay more, because once someone has encountered more than one or two errors on your site, they will probably never return.

No matter how careful you are when working on pages, bad links can creep in, so use a tool like LinkBot ( see WebDevelopersJournal Linkbot Pro Review), which automatically searches your site and tells you of any bad links. Also, just in case, you should have a custom error page on your site. Set up your Web server so that if anyone requests a page that doesn't exist, they are automatically redirected to your error page. The error page contains links to your home page and major sections, your search feature, and the Webmaster's email address. When a visitor arrives at the error page, chances are good they'll stay on your site and try to find what they were looking for. If they just get a generic "404 - Page not found" message, they'll disappear.

One of the biggest problems for e-commerce sites is ordering scripts that simply don't work. Okay, online retail is still fairly new, and a certain amount of chaos is to be expected, but it's appalling to see how many sites, including some of the big names, use fancy-shmancy scripting that crashes half the time. Again, when a script fails to work (or worse yet, locks up the visitor's whole system), not only do you lose that particular sale, but you may permanently lose the respect of a potential customer. Make sure all your scripts have been thoroughly tested, with a variety of browser versions and operating systems.

Navigation and the Search Engines

Just being listed in the search engines isn't good enough - you want to try to rank as high as possible in searches. This means optimizing the appropriate keywords, and your navigational scheme can be a part of this.

Some search engines will assign more weight to a keyword that appears as a heading than one that is plain body text. So you may want to make sure that your section headings contain plenty of keywords. This may have a chilling effect on creativity - no more cute chapter titles - but everybody's doing it. You can't prevent the English language from going to Hell (it's already crossed over the Styx), so you may as well profit from its demise.

We are also told that some search engines assign more weight to keywords that appear as hyperlinks, or even as file names. So make sure that there are plenty of links to the sections that you want people to find, and that they are clearly labeled. Of course, well-thought-out navigational schemes will do this anyway. Can it be that the file name really matters? To some search engines it does, so make your file names descriptive of the page content (no, don't name all your files "sex_sex_sex.html").

This brings up a point that a lot of people seem to miss - the point is not simply to bring as many people as possible to your site any way you can. The point is to bring the right kind of people to the pages that you really want to be seen. If you're selling widgets, then you want widget collectors, not sex-starved adolescents (research shows that adolescents buy few widgets). So bulking up on lascivious keywords will bring you quantity, but not quality. Sales and return visits are what you're after, not just raw traffic. Choose keywords that describe whatever it is you're selling, as descriptively and specifically as possible.

For the latest news on how the various search engines rank sites, see
Search Engine Watch.

Also worth thinking about is the question of which pages tend to come up in searches. Generally speaking, you want first-time visitors to your site to enter through the home page or one of the main section hubs, not through minor content pages. Let's say someone searches for widgets. If your home page comes up at the top of the search, that's good, because there they can see the whole range of widget-related content that you have, and are likely to explore your site. If on the other hand, the page that comes up is a minor page three levels down, which explains how to install some esoteric attachment to one of your many models of widget, they are likely to get confused and split. If your site has a lot of pages, then pages of minor importance can easily crowd out your home page altogether in search results.

This problem exists because automated spiders blindly count keywords. They have no way to know which pages are really the most relevant, although techniques such as assigning extra weight to links and/or headings, as described above, are an attempt to make them smarter in this regard. Therefore, you should make sure that every single page on your site includes a navbar with links to your main sections and your homepage. No matter what you do, some people will enter by the bathroom window, so make sure they can easily find their way to the front door.

Another way to address this problem is to use a robots.txt file to try to exclude certain pages from search engines altogether. In a recent article in The Web Developer's Journal ( 10 Ways Not to Promote Your Web Site), I explained how to set up a robots.txt file, which tells visiting spiders not to index certain pages. Any pages which are not meant to be seen by the public (templates, "test" pages, etc.) should definitely be included in this file. If you use frames, then all pages which represent the content of frames (in other words, all pages that are not framesets) should be here as well. You may also want spiders to exclude pages of marginal significance, such as disclaimers, copyright info, etc.

In extreme cases, some sites want to make sure that people enter only through the home page. They use the robots.txt file to exclude every page other than the home page. Of course, sites with lots of pages of loosely-related content, or sites that sell a lot of different products, should take the opposite approach. The faster someone can find exactly what they want, the better, so every content page can and should be thought of as an entry page in its own right. Analyze your log files to find out which pages are your main entry pages (see There's gold in them there log files! ) and adjust things accordingly.

Maximize Traffic with Internal Links

At the risk of repeating myself, a user-friendly navbar is critical, and it has a lot to do with maximizing traffic. Every page should have a link to your home page, and for an e-commerce site, a link to the ordering page as well. Content-driven sites, however, should have much more. Related content areas should be connected by internal links - as many as you can manage.

The ability to link related content is one of the greatest strengths of the Web, so leverage this to full advantage to keep people on your site. Of course, you already have a navbar on every page that lists all the main sections of your site, and perhaps you even have a few sectional navbars, each of which appears in only one section, and has links to other content related to that section. But the more specific you can make your internal links, the more likely they are to be useful (and used).

The finer points of internal linking are mainly of interest for sites that have a lot of pages of content, or that sell a lot of different products. Internal linking schemes range from the simple to the incredibly complex. First of all, be sure to include internal crosslinks in body text wherever appropriate. For some sites, it may also be worthwhile to have a list of related content that appears at the end of a page, on a sidebar, or wherever.

At the simplest level, you can simply hard-code links to related content. As a site becomes larger, however, this can get hard to maintain. The links may need to be revised as new content is added or old content removed, and careless coders may introduce inconsistencies in type styles and such. For these reasons, you may feel the need to move up to some sort of dynamic solution.

Over at The Web Developer's Journal, we use SSI (server-side include) scripts to include fitting lists of related articles. This is very convenient, but unfortunately costly in terms of page-load time. See my recent article on Content Management Tips and Tricks to learn how to implement semi-dynamic features like this.

Dynamic environments like ASP or Cold Fusion give you various ways to create appropriate links to related content, as of course do high-end content management systems such as Story Server or BroadVision. Some ecommerce packages such as Microsoft Merchant Server include ways to include "premiums" or "special offers," which can be used to guide visitors to related content. Buying a plane ticket to Boston? How about buying a guide book, too?

The ultimate goal, which make marketers slaver and privacy advocates cringe, is to be able to serve up links to related content based on information about the visitors themselves. A certain amount of info about your visitors can be gleaned on the fly, including what browser version and OS they're using, and where they're coming from (or at least, where their ISP is located). Some of the high-end banner ad management software (see Ad Management Software Roundup for a comprehensive, if now somewhat dated, list) can use this information to target ads to specific visitors. With a little imagination, this can be used for content management - simply create appropriate internal links and set them up as "ads."

More personal information could conceivably be gleaned from reader surveys (does anyone really fill out those things?) and used to show visitors targeted content. There have even been various attempts to build databases of such information that are available across multiple sites, such as Engage. Personally, I don't look forward to the day when everyone's age, sex and favorite flavor of ice cream are available to everyone else, but I fear it will come sooner or later.