Once you have organized your keywords into campaigns, one strategy is to triplicate each campaign so that you have one for each match type. When each campaign contains only one match type it allows you to break out your budget more effectively so that if the budget for those keywords are being throttled you can give the most budget to exact, less to phrase and the least to broad. It simplifies budgeting so that you can give the most money to the exact match keywords that have the best chance at converting.

What this strategy will also do is allow your broad match campaign to turn into your keyword discovery tool. Using search query reports in your broad match campaign you can discover new keywords to build adgroups around for your exact and phrase match campaigns to test. When budgets shrink, you can dial up the spend on driving sales through more relevant match types like exact and phrase, and dial down the spend on discovering new keywords for account expansion.

If you are working with a very large account that is constantly shifting budgets, launching products, and you have moving goal targets, triplicating campaigns for each match type can cause the size of your account to bulge but it’s the way to go.

When making decisions on changes to keywords based on conversion rate, cost per acquisition or any other sales metric, you may be hurting important influencing keywords that don’t get the last click but drive sales nonetheless. AdWords helps attributing credit where credit is due so that you aren’t too quick to bid down a keyword because on the surface it looks like it hasn’t driven any sales.

The Search Funnels report in AdWords can show you how much you should worry about keywords that are assisting in sales but not converting. Export the Assisted Conversions report and then filter out all last click converting keywords so that you only see the keywords that have never had a conversion but have had an assist.

In this example there were 191 keywords with the last click conversion and 113 with assists but no conversions. 37% of the keywords might look bad from a last click conversion rate or CPA standpoint but were still vital to getting the sale. In this case any keyword that didn’t contribute a conversion or an assist should be bid down but all other keywords should be given their due credit and either bid up or kept around longer to see if they can drive even more.

Everyone agrees that content network ads should be separate from search ads in your AdWords account for good reason: the data of one network have no bearing on the data of the other network. Search is about an active consumer looking for information. Content is about advertising products next to content where the subject is related to your product.

I think there are other divisions just as big as content vs. search that should be separated out in an AdWords strategy. First, the device that you are targeting will cause a big difference in the results of your campaign. The motives behind a person using a mobile phone is very different from a person using a desktop computer. Targeting tablets separately will also allow for better targeting and results.

On top of network and device there are the different search engines and different methods of  advertising – brand keywords, non-brand keywords, product listing ads, display, retargeting etc. None of the data from any of these different methods should be lumped together in the same report either.

These differences are enough to cause very complicated reporting. Each Engine/Network/Device/Method requires its own analysis if you want to look at each segment in context and be able to make changes that positively affect one without damaging another. Below is a stab at visualising what this looks like when each engine/network/device/method is separated out with its own macro conversions, micro conversions and targets.

Another reason to have device-only targeted campaigns – in this case for tablet targeting. I’ve seen this happen enough in my own accounts that I’ll bet the same is true with everyone: ads that are above the fold on tablets have a click-through-rate up to 10x higher than below the fold.

Take any existing campaign that is targeted to both desktop and tablets, change it to just desktops, duplicate the campaign and target the new campaign to just tablets.  This will allow you more control to bid for higher positions on tablets that you might not want to do on desktop. It will also allow for easier reporting for tablet performance.

The below diagram breaks out the distribution of keywords that drive the most sales from Head to Mid-tail to Tail. A few “Head” keywords will drive the majority of sales, these are typically your company’s branded terms like: Burton snowboards, Volcom, etc. Everything besides the Head keywords are non-brand keywords. Non-brand can be broken up into two sections: Mid-tail and Tail. “Mid-tail” is made up keywords that have a lot of queries but drive less revenue, keywords like: snowboard bindings and helmet. The “tail” is made up of the highly specific keywords that happen less often, keywords like: Red HiFi snowboard helmet and mens Capita Mid Life Zero Snowboard.

The columns in my diagram show the amount of revenue you can expect from each segment. Head terms are the most profitable because these people are already educated about your brand, know what they are looking for and where they are going to buy it from. The Mid-Tail keywords have the least amount of sales because these people are not sure what they want or where to buy from. They are in browsing mode. The Tail has less over all traffic from each keyword but in aggregate it is a very profitable.

Before trying to tackle Mid-Tail keywords, first focus on maximizing impression share of your brand keywords in the head segment. With all other non-brand keywords, build Tail campaigns with tightly themed ad groups and relevant text ads first. If you try to go after the Mid-Tail section of non-brand keywords before working on Head and Tail, you will end up with high CPAs and low conversion rate. You’ll be doing a good job of educating potential new customers down the road with Mid-tail, but based on their search query you can tell they just aren’t in buying mode.

With nearly 7 percent of all digital traffic in the U.S. consumed away from computers, your paid search ads need to be targeting both smartphone and tablet devices on top of desktops if you want to maximize your reach. Tablet users who visited e-commerce websites in 2011 spent 54% more per purchase than smartphone visitors and 21% more than desktop or laptop visitors, according to a report by Adobe Systems. It makes sense then to make mobile-only and tablet-only campaigns with separate budgets, bids and adtext rather than having a single campaign targeting all devices.

Create a report that segments devices and Top vs Other to see what the difference in click-through-rate is for tablets and smartphones when your ad isn’t in the top spot.

You’ll probably find that click through rate drops below 1% when your ad isn’t in the top position for tablets and smartphones, so by having mobile-only and tablet-only campaigns, you can better optimize bids for those device specific positions.

Another reason to make mobile-only and tablet-only campaigns different from desktop targeted campaigns is to write adtext tailored to the devices, something like “…Shop Now From Your Tablet!” or “…Find A Store Near You On Your Phone.”

Google has ad extensions for mobile phones like location extensions and call extensions that boost click through rates and make mobile search ads very effective at driving in store traffic. Optimizing and reporting is easier when you have a mobile-only campaign.

Ad text should guide the organization of ad groups and account structure, not keywords. Traditionally, people start their account structure by organizing their keywords into themes and then the keyword themes get placed into ad groups. At the very end they try to write ad copy for each ad group that will try to be relevant to all of the search queries that get matched to those keywords in the ad groups. But too often the keywords vary too much and you’re forced to either write generic ad copy that appeals to all of the keywords, or write targeted copy that is well-suited to some queries and poorly-suited to others.

They call them ad groups and not keywords groups because its the ads that should guide the structure, not the keywords. This means that generally the amount of keywords in an ad group is small because any single text ad needs to be relevant for every keyword in that ad group.

The filter function in AdWords is pretty powerful. Definitely should be one of the first steps in discovering opportunities and diagnosing problems. Here are a few of my favorites:

This one helps identify low hanging fruit. These keywords have great quality score, low CPA but just need a little increase in bid to get to a higher position so they can make more of an impact.

CPA is high, conversion rate is low and average position is also high. The bid unnecessarily high for these keywords. They’re not doing much for you so you might as well pay less for them.

These keywords are the bottom feeders. Really expensive and nothing to show for it.

Lots of clicks, high CTR and low conversion rate generally means that the ad does a good job of being relevant with the keyword but there is a disconnect when it comes to the landing page. Test different landing pages with these ones.

Lots of impressions and no clicks means these keywords are going to kill your quality score over time. Try different adtext.

Lots of clicks with a low quality score means these keywords are really expensive. Is it worth it?

There are endless ways to organize the structure of your AdWords account. I put together this flowchart to help get the process started. Its easy to see how complex an account can be if you sell multiple products, in multiple locations, for products that all have different uses and seasons.

Click For Larger Version