If you're new here, you may want to subscribe to my RSS feed. Thanks for visiting!

Note: Although this article does discuss tracking techniques, many technical details that are beyond the scope are intentionally left out. If you don’t understand how to do something suggested here, you can search for a commercial solution, outsource a custom solution, ask for advice on discussion boards and news groups, or ask me for clarification.

Testing and tracking is probably the best thing you can do for your PPC campaigns. Through testing and tracking, you can gain a very important edge over all your competitors: the ability to make more money from the same amount of traffic. Once you gain this edge, many additional opportunities become available to you. You can afford to pay more for traffic, giving you higher placements and more traffic from PPC engines, and you get chances to profit from other traffic sources.

That pretty much covers, in a nutshell, why you should track everything within your PPC campaigns, but how should you do it?

The main idea behind testing and tracking is to track which source a desired action came from. You can track how many clicks (actions) your different ads (sources) generated. You can track the amount of conversions that came from different keyword groups or individual keywords. You can track the amount of conversions that resulted from different landing pages. And on each of those landing pages, there are hundreds of different variables to track.

You should always be testing and tracking something on your website because it isn’t hard to do and it can result in dramatic improvements over time. All you have to do is decide on what to test, spend a few minutes setting up the test, and then let it run to collect the data.

Split Testing Ads

You should be testing different versions of your ads because improvements usually mean that you will get more traffic for the same amount of money. This is blatantly true when you are paying on a CPM (cost per thousand impressions) basis, but is also increasingly true with PPC search engines. AdWords has, since its beginning, been using a quality score that more or less ranks ads by their effective CPM. Yahoo also introduced a similar ranking algorithm back in February. Even though you are paying for each click, when you get more clicks from the same amount of impressions, your ranking increases or your CPC decreases.

Your ads are the easiest things to track because the functionality is built right into the PPC engines (as well as many other ad management systems). All you need to do is create multiple ads, tell AdWords to rotate them evenly, and let them run. When you check back on them after a couple hundred clicks, you will clearly see which one is the winner.

You can split test as many ads as you want, but I like to keep it to two or three. You should always have more than one ad running at any given time for each ad group in your account. When the two ads accumulate 100 or so clicks, decide which is the winner and create another ad in an attempt to beat that winner.

If you repeat this process again and again, you will eventually have ads that perform far better than your initial ads.

Testing Keywords for Conversions

If you’re not testing which keywords in your pay-per-click campaign are making you money and which ones aren’t, you are literally throwing money out the window. It is a widely known fact that not all keywords convert equally, so you should not be wasting your money on those that do not convert well.

Google makes it really easy to track conversion at the campaign, keyword group, and keyword level, but it isn’t always 100% accurate. Additionally, it can be difficult to get your tracking codes placed on a merchant’s completion page, especially if you are testing many, many different merchants. Yet another issue is that some people (myself included) don’t feel comfortable giving Google access to all of their conversion data. Because of these problems (or inconveniences), I usually just do this tracking myself using sub IDs.

All of my campaigns are created with tightly grouped keywords, so I mostly track at the keyword group (ad group) level. I assign unique tracking IDs to each group, apply that ID to destination URL, and then process the tracking on my landing pages (using PHP). For sales and leads, the affiliate network handles the conversion tracking, and the unique IDs should show up in your reports. For other actions (such as clicks on certain links or completion of forms), I store the tracking information to a MySQL database.

If you want to track at the keyword level, you can use dynamic keyword insertion in your ad’s destination URL. You can also combine this keyword level tracking with you unique ad group ID tracking.

At least in AdWords, this will give you the exact keyword that caused your ad to be displayed. Although, I don’t think it will distinguish between broad, phrase, and exact match keywords.

As time goes by and your keywords accumulate clicks, you will collect a large amount of data and will begin to see which keywords are converting like crazy, which are converting at a decent level, which are breaking even, and which are losing money. Adjust your campaigns accordingly, removing losing keywords and possibly increasing bids on the highest converters. If you do this, you will see dramatic increases in your ROI… instantly.

Note: I realize that implementing some of these tracking techniques (using PHP and MySQL) will be beyond skill sets of many readers. I plan on attempting to write some tutorials for implementing these solutions (please give me feedback if you are interested), but in the meantime, I would suggest looking into hiring a freelancer to set it up for you, or purchasing a commercial tracking application that can aid in the process.

Split Testing Landing Pages

The area with the largest potential for improvement in your entire sales process is your website, or in most cases, your landing page. You don’t have absolute control over your ads, and your keywords are just groups of letters, but your landing pages offer the ability to make an infinite numbers of changes.

From the moment you first design a landing page, you should decide on one variable to test. It doesn’t matter what it is (design, colors, headline, introduction, price, pictures, links, font), but you need to create two or three versions and find out which is better. Again, I use PHP to rotate the pages and track which version each action comes from.

The first step is to set up a system that will rotate the landing pages. Let’s say you have three versions of your landing page: A, B, and C. When visitor number one comes to the page, they should see page A. Visitor two should see page B, visitor three should see page C, visitor four should see page A, and so on and so on.

You don’t want this to be obvious to visitors, so you need some way to give a particular visitor the same page every time. In most cases, setting a cookie on the visitors’ computers with the landing page version to display is sufficient. If you need greater reliability, you will need to set up a MySQL database that randomly assigns landing pages versions to IP addresses or host names.

Once you have your landing page rotation system in place, you can track the actions the same way you did when tracking keywords. Each landing page should contain an identifying variable that is either appended to the affiliate link (as a sub ID) or saved to your database. After a sufficient number of actions, you can make an educated decision which version performed the best, delete the other versions, and create new versions to continue tracking other elements on the page.

The amazing advantage of maximizing your landing page’s conversion rate, as I mentioned in the introduction to this article, is that you can afford to pay more money for traffic than most (or all) of your competitors. This is the most important principles for dominating any niche. If visitors are more valuable to you than they are to any of your competitors, you can’t fail.

The End Results (Examples)

All of this testing and tracking talk might seem like a lot of complicated work, but once you get the hang of it, and put systems in place, it is very easy to do. If you make it become like second nature, it will have a profound impact on your business. Here are some examples of what I have achieved through testing and tracking:

I have had text ads on AdWords that started out with a decent ~2% click-through-rate and were split-tested all the way to ~25% CTR, sending over 12x as much traffic for roughly the same cost. Although this isn’t typical (I usually shoot for CTRs of about 6%-8%), it is certainly possible.

My most recent keyword optimization was on a campaign that contained 24 ad groups and was racking up $300/day in PPC costs to make $250-$500 in revenue. After testing which keywords were converting and which weren’t, I was able to consolidate to 8 profitable ad groups and $100 per day in costs. Tracking keywords on this campaign took it from borderline profitable to consistently profitable.

For another affiliate offer I was promoting, I was able to get traffic, but unable to profit from it. I knew there were people making money because I saw plenty of competitors. I ruthlessly split tested everything I could think of on the landing page to take it from 1 lead for every 200 visits (huge losses) to ~12 leads for every 100 visits (huge profits). Again, such dramatic results aren’t typical (the initial landing page was really bad), but improving your conversions enough to increase your profit by 50%-100% is a very common occurrence.

Closing Comments

I hope this article was easy to follow and helped you understand both the benefits and methods of testing and tracking everything about your marketing campaigns. If you aren’t tracking at all, start doing it now. If you are tracking, but not enough, start doing it more. I am as guilty of not tracking enough as anyone else. There are always more things you can test and track to improve your bottom line!

And while were still on the subject, if you want to learn even more about testing and tracking, here is a really great set of four (at the time of writing) articles I just found: WebProfits.com.au Testing and Tracking Archive