In my current role, I began to optimize the a few pages on the UK version of the website for search engines. I started with a data-driven approach, implementing a series of A/B tests to gauge the impact of various changes.
Here’s what I did:
- Updating spellings to reflect British English since the pages had Americanized versions of words. This mostly involved changing z’s to s’s and o’s to ou’s
- Reoptimized title tags and meta tags to better reflect the pages’ content and better target intent
- Added UK to title tag. Through discussions with UK colleagues, I identified an opportunity to strengthen our global brand perception, particularly among UK searchers. I strategically added “UK” to title tags, aiming to increase clicks and address the perception of being solely a US company.
I grouped pages into two different groups for this test, including a control group.
- Pages with major changes: this includes updates to title tags, meta descriptions, h1 tags and Americanized spelling
- Pages with minor changes: this includes pages with minor updates to title tags only
- Pages that were untouched: These acted as my control group.
Changes in organic traffic were analyzed for pages with major and minor changes from before and after these updates were implemented, using a a simple python code:
#load the necessary libraries
library(CausalImpact)
#updated the highlighted portion to be your data
mydata <- read.csv(file="file_name.csv", header=TRUE, sep=",")
#optional, but recommended to verify your data; can paste all at once
head(mydata)
matplot(mydata, type="l")
dim(mydata)
#change range depending on rows; can paste all at once
pre.period <- c(1,200)
post.period <- c(201,400)
#To perform inference, we run the analysis using:
impact <- CausalImpact(mydata, pre.period, post.period)
#plot and summarize the results; can paste all at once
plot(impact)
summary(impact)
summary(impact, "report")
For pages with major SEO updates, the implemented changes led to a positive and statistically significant impact. We observed an average increase of 2.80 compared to what we expected without the intervention. This translates to a 33% overall increase. I saw similar results for pages that had minor changes; my implemented changes led to a statistically significant positive impact. We saw an average increase of nearly 35% compared to what we expected without the intervention.
While I plan to do additional analysis on these pages, my results were statistically significant, highlighting that even small changes can have a big impact on organic performance.