May 25, 2011
In July of last year, Google announced the global roll out of modified broad match keywords that had a greater reach than phrase match and more control than broad match. PPC advertisers welcomed this new match type as a way to better target qualified traffic. I slowly began integrating modified broad match keywords into my accounts, but hadn’t taken the time to fully evaluate the effects of this new keyword match type. When I took on a new client in March I decided it would be the perfect time to test how modified broad match keywords can affect account performance.
The client had just begun building their PPC account in February. They have a limited budget of just a couple thousand dollars a month, but were competing in an industry with big name players. I immediately thought they would benefit greatly from the use of modified broad match keywords. Before I went through pausing all the current broad keywords and adding in modified broad, I decided to run an experiment to determine if campaigns that use modified broad match perform better than campaigns that use regular broad match.
To determine the performance difference between broad and modified broad keywords, I decided to run an experiment in one of the higher traffic campaigns. The control campaign used only broad, phrase, and exact keywords and the experiment campaign used modified broad, phrase, and exact keywords. All keyword bids were left at the default ad group bid level. I made very little changes to the campaign while the experiment was running to get the most accurate data possible.
After the experiment was complete I was hoping to answer the following questions:
- Would modified broad match keywords reduce impressions and clicks?
- Would it lower the average CPC?
- Would it improve my conversion rate and lower cost per conversion?
The experiment ran from April 1st to April 30th. Below are the results for the different metrics. The cells highlighted in yellow are the experiment results Google deemed to be significantly different than the control results.
Some of the results were more surprising than others. I’ll go through each of the metrics separately.
Clicks and Impressions: As expected, modified broad match keywords significantly reduced the impressions my ads received. What I didn’t expect was that this 4,000 drop in impressions didn’t have a significant impact on clicks. This shows that modified broad keywords were reducing the number of times my ad was shown with irrelevant search queries.
Click -Through Rate: With the drop in impressions and relatively same performance in clicks, the CTR increased significantly with modified broad keywords, from 1.88% to 2.25%. A better CTR is always welcome, but what my client cared most about was cost.
Spend and Cost-Per Click: CPC was exactly the same for both the control and experiment group, at $0.46. Since we maintained the same level of clicks, total spend also remained the same. Although modified broad match didn’t significantly reduce spend, our ads were being shown to more relevant traffic and we were paying the same price for our ads to be shown in higher positions. With modified broad keywords our ads went from an average position of 4.4 to 2.7.
Conversions: Modified broad keywords didn’t have much of an effect on improving any conversion metrics. After looking deeper into the accounts, this was because our conversions were coming from exact or phrase matches of our keywords.
Below are the major take aways I got from running this experiment:
- Modified broad match will improve click-through rate.
- Modified broad match keywords won’t significantly reduce spend.
- Modified broad match won’t significantly improve conversion rates.
I starred the last one for two reasons. First, since this account does have a limited budget, it also limits the amount of data we can accrue. Running this experiment on a larger spend, higher traffic account may give different results.
Second, most of my conversions came from phrase and exact match keywords. For this account, the ad groups are tightly themed and contain about 5-6 different keywords each (not including match type variations). Because each ad group is so targeted, it’s the phrase and exact match keywords that were triggering our ads. What this showed me is that although modified broad match keywords do offer a better way to target your ads, nothing is more beneficial to account performance than an organized, tightly themed account structure.
If you are interested in running your own experiment, Tom Demers wrote a great article on Wordstream that walks you through the process of creating a modified broad match experiment.
Have any of you tried running similar experiments? I’m curious if anyone has seen an improvement in conversions from modified broad keywords. Please comment below!