Andreas Voniatis dug deep on stats to sculpt PageRank, and reaped 208% more SEO traffic for their enterprise client with a 68-position average ranking increase across 588 keywords. Besides for the remarkable case study, Andreas and I discussed the exact statistical approaches his SEO agency – Artios.io – uses to differentiate itself, and eliminate ego from SEO-implementation discussions. As a result, their clients are much likelier to implement Artios’ recommendations, rather than responding with something like “but expert A says blah blah blah” because it’s not this opinion or that one – these are the facts.
Learn how they’re achieving all these awesome results in this eye-opening interview with Andreas Voniatis of Artios.io
Elaborating on the research Andrea mentioned in the video with image sizes, here’s some material from a presentation they did highlighting this:
“We’ve used Artios technology to analyse the two drinks companies in attendance here today:
- T Plus
We’ve scanned all the pages of both websites and retrieved data on over 35 features of their website. Here’s an example:
The graph shows the height of the web page images for both sites and how they vary across the pages. The peaks represent the most frequent image heights for the two websites.
One of the things we found quite interesting is that on average, the images on the T Plus are around 100 pixels compared to Goodshot which has image heights of ~270 pixels.
The image widths on the other hand are quite similar:
Here we can see, both of these sites are have average image widths of just under 250 pixels although, Goodshot is much more varied with some images being as wide as 700 pixels.
Which site is better?
It depends on what it is you’re trying to achieve. Is it:
- getting more shares on social media?
- getting more only sales or enquiries or
- ranking higher in Google i.e. SEO?
Say SEO was important to you, one of the things you can do is plot the image sizes to see whether the image width shows a predictable relationship between you and your competitors in Google.
As the graph above shows, the higher ranked leading competitors have larger images on average and should aim an image width of 275 pixels.
The R-Squared statistic tells us what percentage the feature can explain the differences in Google ranking, making your SEO campaigns less of a guess and more predictable. It’s a step beyond correlation, with these maths looking at causation, aka the ability of a metric to explain a site’s ranking ability.
Here are a few blog posts where you can read more about Artios’ work:
https://artios.io/unlocking-the-real-potential-of-pagerank-with-data/ – their own description of the case study