Case Study Showcases Conventional Wisdom Failing when Search and Content are Not Treated Separately
Search and content are different. I think we can all finally agree on that sentiment.
With search, someone is actively looking for information. Their goal is to click on a link.
With content, someone is looking at related information. They may click when they are finished reading the page.
The mentality of active search versus random discovery lends itself to not just using different keywords for your search and content campaigns. It also shows that you should test different landing pages for the two mediums as well.
Here are the results of a test that was conducted using different landing pages across search and content.
The Company’s Products
This company offers two products (Since they are so niche, I’m using the proverbial plumber services instead of their actual product – everything else is the same):
- A database where subscribers can find local services
- It’s a subscription based model – this is where they make money
- A free listing where local services can input their information and hopefully be hired by the subscribers.
- The listings are verified. This leads to creating a high trust factor for the consumer
- The company doesn’t make money on these listings, this is increased value to the subscribers
The Website’s Landing Pages
The website has two different types of pages.
Segmentation page. The home page has two navigation paths: One is for listing services (free) and the other is for finding services (subscription). The page has many trust elements (BBB, SSL, Associations, etc).
City Landing pages. For each major metro, they have a page that lists the city, has a picture of the city, lists the services, how many companies are in their database for that city, etc.
Two very different pages.
Conventional wisdom says to put the Chicago searchers on the Chicago city page, and the Miami searchers on the Miami city page.
Was conventional wisdom correct?
The Actual PPC Test
There were 21 campaigns set up for this test.
- Campaign 1: Search only using Geo-modified keywords (i.e. Chicago Plumber)
- Campaign 2-11 (10 campaigns): Search only using location targeted campaigns (i.e. each campaign only targeted a single city such as Chicago, NYC, Miami, etc). Keywords were not geo specific (i.e. Plumber)
- Campaign 12-21 (10 campaigns) Content only. Location targeted campaigns using the same cities, keywords and ad copy as the Search only campaigns above (essentially, the exact same campaign, just content instead of search targeted)
The visitors for all the campaigns were sent to two different landing pages:
- Home page (which is a segmentation page)
- City Landing page which corresponded to the appropriate city for the IP targeting campaign or geographic keywords
For this test, only a subscription was considered a conversion. Listing free information was not considered a conversion as it’s not the way the company makes money.
A very simple split test was used to execute all of the testing.
The Resulting PPC Data
The Search Campaigns (campaigns 1-11) all had very similar data. There were a few outliers at the keyword / ad copy intersection. However, in general every campaign behaved very similarly. This includes both geo-modified keywords and location targeted campaigns (which doesn’t always happen).
The Content Campaigns (12-21) also had similar data. There were a few more outliers in this campaign, but those outliers made up less than 10% of the keyword / ad copy intersections, and less than 8% of the total clicks and impressions.
Was conventional wisdom correct?
Conventional wisdom says the city pages should outperform the home page. After all, those pages are more specific to the user – correct?
Upon first examination, it appears that conventional wisdom was correct. The conversion rate and cost per conversion are lower for the city landing page than the home page.
However, never combine search and content when examining data – NEVER.
On the content network, conventional wisdom was correct. The City Landing Pages had a cost per conversion of $8.99 vs. the $13.34 for the Home Page. Not surprising.
On the search network, conventional wisdom failed. The Home Page had a lower cost per conversion and higher conversion rate than the City Landing Pages.
Here’s the data to examine:
Another reason to test search vs. content is conversion rates. In this instance, the content network outperformed the search network in terms of cost per conversion and conversion rates. This is not always true – it’s why you test.
What’s the lesson?
Search and content are very different.
The user engagement process is different.
The offers can be different.
The keywords can be different.
The ad copy can be different.
In fact, every campaign setting and ad group feature can be optimized differently for search vs. content.
Your conversion rate and cost per conversion are often very different.
When running reports, make sure you’re looking at just search or just content data. If you combine them, you won’t see an accurate picture of what’s really transpiring in your marketing efforts.
If you’re not testing, then you’re losing money. It really is that simple. Test, but test and measure appropriately.
superb article and insights, thanks Brad.
I admit we don’t do enough landing page testing; it’s confusing as we are testing ad copy, so when do you stop a/b ad copy testing, and switch to landing page testing, as I ususally just run two ads per group.
You can test ad copy and landing pages at the same time. That can help get to my absolute favorite metric – profit by impression (or profit per click for the content network):
Take your two ads and send them to page 1.
Take the same ads, duplicate them, and send them to page 2.
That will help you see the full cycle of consumer thought (keyword) to ad copy to landing page interaction. Different ad copies will have different conversion rates on the same landing page.