A analysis of over 400 websites by Cyrus Shepard, founder of Zyppy, identifies five characteristics associated with whether a site has gained or lost estimated organic traffic over the past 12 months.
Shepard has categorized the sites by revisiting many of those covered in Analysis of Lily Ray’s December core updateclassifying them by business model, content type, and other features, then measuring the correlation with traffic changes. Traffic estimates come from third-party tools and not verified Search Console data.
Five features showed the strongest association with traffic gains, measured by Spearman’s correlation:
- Offers a product or service: 70% of the winning sites offered their own product or service, compared to the 34% of losing sites. Service-based offerings such as subscriptions and digital goods have performed well alongside physical products.
- Allows task completion: 83% of winners let users accomplish the task they were looking for, versus 50% losers. Sites don’t need to sell anything to score here.
- Exclusive assets: 92% of winners had something difficult to replicate, such as unique datasets, user-generated content, or specialized software. Among the losers, this figure was 57%.
- Tight thematic focus: Winners tended to cover a single narrow topic in depth. Shepard noted that a general “thematic” classification showed no difference between winners and losers, but narrowing the definition to the depth of a single topic revealed the trend.
- Strong brand: 32% of winners had high branded search volume relative to their overall traffic, compared to 16% losers. Shepard measured brand strength by examining each site’s top 20 keywords for navigational branded terms using Ahrefs data.
The effects were additive. Sites without features had a success rate of 13.5%. Sites including all five achieved 69.7%.
What was not correlated
The study also tested features Shepard hoped would matter, but found no correlation with traffic changes. These included direct experience, personal perspectives, user-generated content, community platforms and uniqueness of information.
Shepard cautioned against misreading these results.
He suggested that these features might already be baked into Google’s algorithm from earlier updates, meaning they might still matter even if they don’t show differential results between winners and losers in this dataset.
Why it matters
Shepard’s results suggest that sites that offer a product, complete a task, or have assets that are harder to replicate were more likely to show estimated organic traffic gains in this dataset. The study gives precise figures behind this pattern, although it does not establish a causal link.
The additive model is the most useful finding for those evaluating their position. A site with a winning feature had a success rate (15%) about the same as a site without a winning feature (13%). The gap only widened for three or more features.
Roger Montti’s analysis for Search Engine Journal in December identified models related in the other directionnoting that Google’s topic classifications have become more accurate and that core updates sometimes fix overranking rather than penalizing sites.
Looking to the future
The correlation values in this study are moderate (0.206 to 0.391) and the methodology relies on third-party traffic estimates rather than verified analytics. Correlation does not establish causation.
Sites that offer products may perform better for reasons beyond Google ranking preferences, including higher return visitor rates and more natural backlink profiles.
THE the full dataset is publicmeaning others can test these classifications against their own data.
Featured image: Master1305/Shutterstock
