How We Increased Traffic 110% By Using Keyword Clusters to Reduce Cannibalization

The Story of Zerodown

Zerodown, a distinguished US real estate company, has revolutionized the way users find their dream homes through their cutting-edge website technology. Unlike its competitors, Zerodown goes beyond conventional filters by leveraging data and AI. Their ingenious approach enables users to personalize their home search with unique criteria, such as desired levels of darkness in the sky, greenery, local voter statistics, and a host of other cool filters. With Zerodown, finding a home becomes an immersive and tailored experience like never before.

As they eagerly pursued further growth in a fiercely competitive market, Zerodown approached us for assistance.

Their primary request was to enhance the organic visibility of their colossal website, which was clocking in at over 50 million pages. Specifically, they sought our expertise to optimize the technical aspects, enhancing crawlability and indexation for key folders that were not receiving organic traffic.

The Challenge of Improving Indexation of a Several-Million-Page-Site

The types of queries Zerodown wanted to rank for included:

• Homes for sale in California
• Homes for sale in Las Vegas
• Homes for sale in Nevada
• 2 Bedroom homes for sale in Florida
• Log cabins for sale in Boulder
• Mansions for sale in Los Angles
• Colonial homes for sale in Orange country

And so on.

There were a number of issues including a constant struggle to rank many of their town-based queries beyond the second page, having hundreds of thousands of URLs “crawled but not indexed” in search console and often having the wrong page rank altogether.

We started trying to diagnose the issue by crawling the site, but two weeks later our crawler crashed and had only completed 50% of the crawl (it tapped out around 25 million URLs).

The size of the site appeared to be a direct result of the fact that a “property type” landing page had been created for every single keyword in their keyword research. They had around 400+ property types including:

• Homes for sale in …
• Houses for sale in…
• Properties for sale in…
• Log cabins for sale in…
• Wood cabins for sale in…
• Lakeside houses for sale in…
• Waterfront houses for sale in….
• 9 bedroom houses for sale in…
• Mansions for sale in…

And so on.

These keyword, property type pages would be created for every state, town, neighbourhood and postcode in the USA leading to MILLIONS of pages.

Our Hypothesis

Our theory suggested that we were inadvertently causing self-cannibalization and creating an excessive number of URLs for search engines to crawl.

We found it hard to believe that phrases like “properties for sale,” “houses for sale,” and “homes for sale” were significantly different to justify separate categories. Similarly, categories like “colonial homes for sale” and “Spanish colonial homes for sale” seemed redundant. There were also potential overlaps in categories with vague distinctions such as “9-bedroom houses” versus “mansions”.

The Problem

We were faced with more than 400 “property categories” to examine. Reviewing each one individually in the search engine results pages was not feasible due to the scale of the task and the monotony it would entail.

How We Used Keyword Clusters to Reduce Cannibalisation and Improve Indexation

We acquired a list of all property types from our client and appended the phrase “for sale in California” to each. We included a state in our searches because just entering the property type often resulted in blog-related content, not property listing pages. For instance, searching “farmhouses” alone often yielded articles describing them, not a page listing them for sale, which was the information we required.

So our list looked something like this:

• Homes for sale in California
• Houses for sale in California
• Properties for sale in California
• 2 bedroom homes for sale in California
• 3 bedroom homes for sale in California
• 4 bedroom homes for sale in California
• Castles for sale in California
• Mansions for sale in California
• 9 bedroom houses for sale in California
• Log cabins for sale in California
• Colonial homes for sale in California

And so on. 

We then added some of our own keyword research to the list, to make sure we weren’t missing out on any additional opportunities.

We subsequently fed these data into our own keyword clustering tool, Keyword Insights, which provided a detailed analysis of the clusters and their rankings.

Our tool scrutinizes the search results for every query entered and clusters them based on whether they could, and should, be targeted on the same page. Essentially, if the queries “properties for sale in California,” “houses for sale in California,” and “homes for sale in California” were clustered together in the report, it would indicate that one page, not three, should target these queries.

Additionally, we extracted ranking data and ranking URLs. This allowed us to swiftly detect if multiple URLs were competing for the same cluster, a clear indicator of self-cannibalization.

What Did It Show?

As anticipated, we discovered an excess of categories and significant overlap among them, leading to cannibalisation.

Some overlaps were predictable, like those between ‘homes’, ‘houses’, and ‘property’ pages. However, we also stumbled upon less apparent ones that might have gone unnoticed, such as overlaps among ‘cabins’, ‘chalets’, and ‘log home’ property pages.

Here’s a screenshot demonstrating the first 30 or so property-type pages. The right column indicates property types that could be merged with the corresponding property type in the left column:

In total, we’ve narrowed their property type pages from 413 to 85.

Across all their different states, postcodes, towns and cities this has resulted in a reduction of around 15 million URLs.

The Result

An incredible 110% rise in organic traffic: