How SEO Testing Saved This Retailer Millions in Future Revenue

traffic modelling
0 %
increase in revenue YoY
0 %
increase in traffic YoY
0 M+
in saved in lost revenue
0 K+
sessions protected*

Brief

In mid-2019, Dejan was engaged by one of Australia’s largest home and furniture retailers.

Prior to engaging Dejan, the client tested website changes that were deemed a success based on an increase in overall organic performance compared to the previous year.

However, an initial assessment by Dejan identified the changes did not contribute to the increase and, in fact, categories without the change were performing much better.

The project had two key goals that would define its success:

  • Assess current performance and effectiveness of SEO changes
  • Identify new opportunities for traffic growth

Activities

  • SEO Performance Review
  • Website Risk Assessment
  • Technical SEO
  • eCommerce SEO
  • SEO Testing Framework
  • Financial Forecasting

Achievements

Following our SEO audit and findings, the client agreed to cease the rollout of proposed changes to the site, and work with Dejan to deploy new tests using the suggested SEO testing framework.

Since campaign commencement, Dejan’s work across the following activities have contributed to a 100% increase in traffic YoY and a 139% increase in revenue over the same period.

  • Snippet optimisation
  • On-page targeting
  • Facet indexing
  • On-page copy testing
  • Mobile first index
  • Review
  • Buying guide process
  • Development
  • Adhoc tech support

Client Hypothesis

Updating category URLs to follow the structure of the breadcrumb menu will lead to increased organic traffic as Google will better understand the relationship between pages.

Before: /wet-food-c73829
After: /dog/food/wet-food

Traffic Modelling in the Pre-Period

Using the data analysis package CausalImpact, for the R programming language, Dejan completed a review to identify what impact URL changes had on organic traffic.

First, we established a relationship between the URLs that were updated and those that were not – both during the pre-period, before changes were made. For example, when an unmodified URL category saw a traffic increase, the same traffic trend was observed in a modified URL category.

traffic model
Updated pages (solid black line), Non-updated pages (dotted lines)

Traffic Modelling in the Post-Period

Using the relationships identified during the modelling phase, we forecasted what would have happened if the updates were never made.

Although traffic to the category continued to grow, our forecasts identified that, in the absence of URL changes, the growth rate would have been much higher.

During the 130-day post-period, the client would have seen an additional 30,000 sessions to this area of the site. Using the site’s average session value of $6.00, this equated to $120,000 of lost revenue to this section of the site alone.

If this update was rolled out across the entire website, the lost revenue would have translated into millions of dollars in a matter of months.

Key Takeaways

  • The fact that URL changes were found not to contribute to traffic growth tells us Google already had a good understanding of the site’s structure and relationships between pages.
  • This is another good example of how Google’s understanding of each site is different, and therefore precaution should be taken when rolling out site changes – especially at scale.
  • In addition to tracking accuracy, it is important for brands to also consider the methods used to measure the success of website enhancements.
  • Dejan specialises in SEO test design and deployment, working closely with brands to safely execute new tests and website changes (however large or small) for optimal return on investment.

For further information on our SEO testing methodologies and eCommerce campaigns, feel free to get in touch.