Let me preface this post by stating clearly: this is not the best way to work all accounts. With that said, I can think of a few scenarios where it is worth a shot as there are many pros to working an account from more targeted to less. There are also some cons so I’ll start by addressing both.


  • You’ll have less initial financial waste while prospecting keywords
  • It is lower risk
  • You should see higher starting conversion rates and lower conversion costs
  • The build out will be more progressive and less intense to start
  • Lends to good account structure best practices
  • You will find expansion to be more predictable


  • You need to put more thought and research into your starting keywords
  • You will be missing some opportunities initially
  • Optimization takes longer
  • You’re conversion volume won’t be as high initially
  • You need to have patience and work through the process

You can see the pros and cons are fairly balanced. I would consider this strategy in a few cases. If I had extremely tight goal metrics and budget but had an extremely high spend potential, I think this would be a good option. The other case would be if the product or service was extremely niche and focused. Essentially, the smaller number of keywords you envision being successful the more likely this strategy will be successful for you. Now, I’ll layout a step-by-step process of how to best use this strategy to your advantage.

Step 1: Keyword Research

Keyword research, like in any other situation, is going to be the foundation of your campaign breakout. What is a little different in this scenario is that you need to try to identify as many long tail keywords as possible. The whole purpose of this strategy is to eliminate waste so you want to get right to the targeted exact match keywords you feel most represent what you’re advertising. If you are advertising a niche product, the keyword list might not be long. When I used this strategy, my initial keyword list was roughly 50 terms.

Step 2: Exact Match Campaign(s)

Now that you have your keyword list, you’ll want to break out your initial campaigns. When doing this, consider best practices and use a silo technique of tightly focused ad groups. I would break your campaigns out by match type and keywords. For example, if you sell red and blue pens, make an ‘exact red pens’ and ‘exact blue pens’ campaign. From there break out ad groups accordingly, for example, create ballpoint and felt ad groups. All of your keywords should be exact match and you should run the full set of keywords discovered in your research.

Once you have your campaigns, go ahead and turn them on to start collecting data. I generally write two ads per campaign and first focus on finding the “sweet spot” for my bid price, while trying to identify the first ad I want to break down to test further. You should be able to determine the better performing ad while testing positions but take your time with this. Since it’s a new account, you need to let your bids run for an adequate amount of time to understand how they perform. Depending on the pace this could be a week or longer but initially I would run it for a week (you’ll still have something to work on in creating your phrase and broad campaigns).  What you want to do is test your keywords in three different positions. You’ll want to test them somewhere near position 1, somewhere near position 3, and somewhere near position 5. I would test these positions over 3 weeks, find the lowest CPLs, and then start testing up and down in smaller increments until you find the sweet spot. You can start reducing the amount of time between tests at this point but I would still look at half a week on a minimum.

It’s also important to take benchmarks for your ads at every bid change. You might find that ad 1 performs better in higher positions but ad two performs better in lower. This is all important data to collect because you are after the right mix for low CPLs.

Step 3: Phrase Match Campaign(s)

While you’ve been running your tests in exact match, you shouldn’t have been sleeping on breaking our additional campaigns. Phrase is the next step in the process and you should try to get your phrase campaigns going quickly after your exact campaigns. What I do is create identical campaigns, but I change the match type to phrase. Easy enough right? Well, there is a little more to it.

The first change is that I don’t want my keywords to compete for quality score or position with the exact match campaigns. While it’s arguable that this step is truly necessary, you could run into a situation where maybe your exact match quality score is 3 but your phrase match is 10 and phrase could start winning auctions because of it. I want to remove this possibility. What I’ll do is add negative keywords that are identical to all of the phrase match keywords in the campaign. Essentially, that ensures that each click and query that comes from the phrase campaigns will be different than the exact campaigns.

The other difference is that I’ll test two unique ads per ad group – not the same ads that are in the Exact campaign. What I’m trying to determine here is two independent winners, one from the exact ad groups and one from the phrase ad groups. I’ll go through the same process I did in the exact campaign ad test but once I find my bid sweet spot and ad winner, I’ll start testing them against each other while introducing multivariate tests like landing pages and ad copy mix. The idea is to first find the style of ad copy that seems to work and then to hone in on the specifics of improving the actual ads.

Here is where it starts to get interesting. Because you’re only serving phrase ads on queries that are different than the search term itself, you should be collecting conversion data on keywords that you haven’t thought about. These keywords should be more long-tail and should produce less volume but higher conversion rates. When you recognize, through search query reports, keywords that have converted, add them to your exact match campaigns AND your phrase match campaigns. Just remember to add the negative if the same keyword to your phrase campaign. You can hopefully see the cycle of your campaign breakout now. It’s controlled, safe, and should allow you to grow on keywords that are known converters rather than taking the shotgun approach and having to trim the fat of poor performing keywords later.

Step 4: Broad Match Campaign(s)

The same philosophy of campaign creation we went through with phrase should be used with broad. There are some small differences however. First, be cautious with your budget. You might be at a comfortable CPL with phrase and exact but broad opens up a whole new can of worms. You should start thinking about how much room you have to test. If there is plenty, I would import the full lists, along with the negatives, you created in phrase. If you don’t have much wiggle room, you could think about only bringing in a few of the more broad keywords but then add the full negative list. The goal here is to identify keywords that aren’t being utilized in the existing campaigns. You want to continue to spread your reach slowly and using this strategy should allow you to do so.

Remember, this isn’t the best strategy for every account but if you are in an ultra competitive industry, it’s certainly a safe and reliable way to grow your account.