|
Testing and Tweaking Ads and Promotions
Test, test and then test some more. You never know what you can improve on a promotion until you try. If you placed your ads, the money is coming in, and you've left it alone - you are probably leaving money on the table. You should be testing, tweaking and tracking production if you want to maximize your performance. First, variables to consider about testing and tweaking:
Ad Location
Just like real estate, the top three things that determine an ads price and performance are location, location, location. Make sure your visitor's eye is caught by the ad. Put it in a prominent location that fits your layout. The hotspots generally are at the top left of the page, scanning downward and right to the upper center of the page. Always put your primary ads above the fold. Put them near other highlighted areas of your site, such as the header or navigation.
Ad Type
Nine times out of ten, text ads outperform image ads. People read and scan to understand what a page or link is about, or where to click next. Banner-blindness has really set in and most traditional banners are ignored by even some of the least experienced internet users. Some situations call for a graphic ad, or at least a test for a graphic ad. Sites heavy on images, such as sites related to photos and or video, can be good testing grounds. Also, if your page is heavy on text, an image in the middle of the content can work well.
Ad Colors
Sometimes you will want an ad color that blends in well with your site. Users may click without realizing they are clicking on an ad. Other times, you will want a color that stands out from the rest of your page and shouts "CLICK ME". Conventional wisdom does not always apply, so it is sometimes worthwhile to test something that may not make sense at first.
Other Ad Variables
There are other less common characteristics about ads, such as interactivity, that can be tested. Some ads will have great interaction functions built into them, which can help draw a click from an interested user. In other situations, it may be better for the user to clickthrough and get the full interactivity of the merchant's website. You really won't know for sure until you try. Basically if there is an option to do something different on a promotion, consider testing it.
Different Options for the Same Promotion
Many merchants have an affiliate program at several different networks, and maybe their own tracking solution as well. You can test the different tracking mechanisms to see which works best for you. Often, you can find a much better commission through one network or provider than the others, on the same product or service. For contextual advertising, a specific keyword set may have better paying advertisers on a different network. You won't know until you try.
Testing
Before you test anything, be sure you are working with a site that has a steady flow of traffic. You want to normalize variables outside the test variables as best you can. Also, you probably don't want to spend your time squeezing 3 more bucks a day out of a domain. Take an established domain, with established traffic where there is enough potential to make more money for your time.
Track before and after results. Make sure you are dealing with similar incoming traffic from the same or similar sources, or discard your results unless they are highly indicative of the changes. Track your uniques, pageviews, referrers, and ad clickthroughs for a given time period. Use a day, week, or month, depending on topic and volume of traffic. Then use your new ad placement, or color, or type, or whatever. Do not test many changes at once, or you won't be able to measure which change made how much difference. Track the new numbers for the same time period and compare. Note that if your CTR picks up, your pageviews may drop as a correlation. Note your observations and test another tweak on the change you made or tweak a different characteristic of the ad. Continue until you are happy with the numbers, or you're satisfied you have maxed that particular ad's potential.
|