Tastybuds

Users who come to your site via Google, and other search engines, are much different than users who come directly to your site. Organic users know they are coming to your site and are familiar with it. Search users, on the other hand, are looking for something specific, which they may or may not find on your site.

Because these users are so different from each other, you should optimize your pages for each type of user. You can have different navigation structures, extra information for search users, special deals, or simply rearrange what you do have. It is difficult to know exactly what works best, but keep your key metrics in mind, segregate your audience, and optimize each one.

Even for something like AdSense, which you might believe has an overall “best” style and position should be tested for each type of user. You will often find that what performs better for one group of users will be quite a bit different than what performs for others.

Also, you might have different goals for organic users than for search engine users. For search engine traffic, you may want to drive them to register on your site; or you may want to immediately monetize them as best as possible; or you may want to lead them to a key information page. Whatever your goals, realize that they may differ for each audience segment.

A simple method to do this sort of testing is to cookie users that came directly from www.google.com, or any other sites you would like to segment. Then, for each page you display, and for your tracking software, look at the cookie, and act accordingly. Run the same exact test on both sets of users, and you may find the best result is different for each.

One of the most critical aspects of a testing and optimization is choosing proper segments. You can lump everything together, and run one test for your entire site, but the gains will be far less than if you were to run 3 or 4 tests, one for each segment you are interested.

You can segment on just about anything. One of the most obvious variables is pagetype. For example, if you run a blog, the front page may be structured different from your article pages, which might be different from your links page. Each of these pages might be able to comfortably fit a different amount of ad units (for example, your homepage might get one below each article, while your article pages might only have one on the right side). Even if you do have the same initial layout on all your pages, you will find out that the same exact ad unit will perform different on each page.

Another variable you can segment on is whether a user is logged in. Regular visitors behave differently than casual ones, so you you should treat them as such. Often, the best style for one set of visitors will not perform as well for another set.

Time of day, or day of week, is another thing you can split on. If your site is business oriented, you might see different types of users during regular work hours than on weekends. Weekend users might be better optimized for multiple units, while business users for a single precise unit.

The point is that you have a ton of information about your users, and you should use that information segment. Each of these segments may perform different, and you should optimize accordingly. Other ideas dr segmentation include:
Time on site
Browser type
Country of origin
IP
Users from Google
Category or topic of page

Not all sites will be able to segment on all of these variables easily. Also, some of these variables might or might not have a huge impact. But the point is that you should be testing as much as you can, and when performing your tests, you should break things down as granular as you can.

Optimizing AdSense is a fairly simple procedure to get started on, and you will notice the results very quickly. But as you begin to do more of it, you have to take into consideration the possible negative effects of your optimization. Many sites have multiple revenue streams, including display media, leads, and sponsorship. Is your optimization hurting those other revenue streams?

In order to get a good idea of the effect you are having, it is necessary to segment users into groups, and see how those groups are monetizing with AdSense, along with any other useful metrics you have access to. For example, how many pageviews per user are you getting with each test style? How many leads are these users generating?

Depending on your site structure, and what tools or partners you are using, getting the answer to these questions will be different, but the overall methodology should stay the same. One simple way to look at the overall site performance is to start each page with a cookie. If a user already has a cookie, you check which test group they belong to, and assign that value to a variable. If they don’t yet have a cookie, you randomly put them in one of your test groups, and give them the cookie for later pages in their session.

Once you have established a user in a group, you then make sure to pass this information anywhere it applies. For AdSense you will show the test style associated with that group. If you use a partner like shopping.com, you can pass in the test name for tracking leads from these users. For your analytics software, you likewise segment based on this test group.

Now instead of looking how just AdSense is performing, you can start to see a larger picture. In my experience, often what is best for AdSense often is a net gain for sites, but there are definitely exceptions. Using this framework, you can make sure that your testing gives you results that optimize your entire site.

One reason I decided to start blogging is that I was so impressed by bloggers being able to ask a question, and get their readers to answer in the comments. I have seen Dave Winer do this several times, and I have always wanted the ability to do the same.

The funny thing is that I was just thinking about this very issue when I saw Shel Isreal post about it. It seems that even beyond blogging, simply using twitter is an effective way to get questions answered from the community around you. To take it one step further, TechCrunch then posts about ToAnswer, which seems like the perfect service for someone like me who has no community.

However, although ToAnswer might work well for me, I still would like to build up a base community of people I can learn from and share knowledge with. I have been reading a few blogs for a few years now, but I have never really participated. I have already started posting quite a bit on this blog, but my next steps will be to actually use my Twitter account, and maybe sign up on FriendFeed (which it seems is another branch that needs to be explored).

My main issue right now is finding what blogs to read, and who to follow on Twiter. I don’t think it will be too hard, I just have never tried before, so am a bit confused as to which directions to look. I just signed up on Digg, so I hope I will begin to find interesting stuff over there. I also am trying to be more active about searching Google for information I am interested in, and then commenting when something catches my eye.

Anyway, any suggestions of how I can build a community of people (through blogs, twitter) who I can ask questions to, and give and get answers from would be much appreciated.