Of 57 technical and on-page factors measured across 6,824 pages in the top ten positions of Google Italy, only 10 cross the threshold of statistical significance. Core Web Vitals are not among them. Neither is page speed, word count, or schema markup. Keyword in title, however, is the factor with the strongest correlation across 841 Italian keywords and five market niches.
This is the second edition of our ranking factors study specific to Google Italy. The first version, published June 17, 2026, was based on 601 URLs and 135 keywords: a pilot study useful for establishing direction, but with a dataset too limited for robust results. In this round, we expanded everything by an order of magnitude.
What changed from the first edition
The comparison between v1 (601 URLs, 135 keywords, 11 factors) and v3 (6,824 URLs, 841 keywords, 57 factors) is not just quantitative.
In v1, the three significant factors were keyword in H1 (r=-0.144), keyword in title (r=-0.139), and keyword in URL (r=-0.112). The direction was clear, but the correlations were likely amplified by the small dataset concentrated on just a few niches.
With 6,824 URLs, the same correlations attenuate: keyword in title drops to r=-0.087, URL to r=-0.071, H1 to r=-0.056. The absolute values are lower, but statistical significance is more solid (p<0.001 on n=7,404). In a large dataset, real correlations persist; spurious ones disappear.
v3 also adds 7 new significant factors that in v1 had not been tested or did not emerge: number of H2s, external links, keyword in H2, number of H1s, .it TLD, and the text/HTML ratio (the only factor that hurts ranking).
On one point, v1 was wrong: the .it TLD. In the first version it appeared non-significant with a positive correlation (r=+0.069, p=0.09). With 6,824 pages it instead emerges as a significant negative factor: r=-0.031, p=0.007. The Italian domain advantage exists; it is small but present.
Methodology
The 841 keywords come from five sites with real Google Search Console data: wikiherbalist.com (herbalism), stampa3f.it (3D printing), f-hack.com (cybersecurity), instapetshop.com (pet e-commerce), and perseodesign.com (web agency). We supplemented opportunity keywords with Google Autocomplete variants, bringing the initial pool to over 17,000 queries. From these we extracted the 841 with verified GSC signal.
For each keyword: top 10 organic results from google.it (locale it-IT, headless Chrome with 8-15 second delay), totaling 7,404 URLs, of which 6,824 unique. Each URL was analyzed by PerSeo Insights: real page load via Playwright, Core Web Vitals via Lighthouse, 57 technical and on-page signals extracted. Spearman correlation measures the association between each factor and SERP position (1-10).
Correlation does not imply causation. Datasets from five specific niches may not be representative of all segments of the Italian web.
The results
Of 57 factors, 10 cross the significance threshold. Nine improve ranking (negative correlation with position), one hurts it.
| Factor | Spearman r | Sig. | Effect |
|---|---|---|---|
| Keyword in title | -0.087 | *** | Improves |
| Keyword in URL | -0.071 | *** | Improves |
| Keyword in H1 | -0.056 | *** | Improves |
| Keyword in meta description | -0.055 | *** | Improves |
| Number of H2s | -0.039 | *** | Improves |
| External links | -0.039 | ** | Improves |
| Keyword in H2 | -0.036 | ** | Improves |
| Number of H1s | -0.033 | ** | Improves |
| .it TLD | -0.031 | ** | Improves |
| Text/HTML ratio | +0.045 | *** | Hurts |
**** p<0.001, ** p<0.01*
The keyword: the most robust factor
The factor with the strongest correlation is the presence of the keyword (or a variant) in the title tag (r=-0.087, p<0.001). URL follows (r=-0.071), then H1, meta description, and H2.
The data by position group is clear. For keyword in title: pages in position 1-3 have it in the title 11.4% of the time, those in position 7-10 just 5.7%. For URL: 8.7% vs 4.8%. The progression is consistent across all positions and all elements.
Page structure matters more than length
Three structural factors are significant: number of H2s (r=-0.039), number of H1s (r=-0.033), and external links (r=-0.039).
The number of H2s is interesting because it does not measure text length but rather content organization into sections. A page with more H2s is likely more complete and navigable. Word count, however, is not significant (r=-0.007, p=0.60): writing more does not help, but organizing better does.
External links (r=-0.039) may reflect a quality pattern: in-depth articles cite sources; superficial ones do not. This is not about outbound links as a direct factor, but as a proxy for editorial care.
The .it TLD: local advantage confirmed
With 6,824 pages, the .it TLD emerges as a significant factor (r=-0.031, p=0.007). Pages in position 1-3 are on .it domains 65% of the time; those in position 7-10, 61.1%. The difference, while modest, appears statistically solid.
In v1 this factor seemed non-significant, with a positive correlation (r=+0.069). The sign reversal and achievement of significance with a larger dataset suggests it was an artifact of the small sample. The .it advantage for Italian SERPs exists.
The only factor that hurts ranking
The text/HTML ratio is the sole factor with a significant positive correlation (r=+0.045, p=0.0006). A high value means a page that is almost entirely text, with little structural markup.
This does not contradict the H2 point. A modern page has a lot of HTML surrounding text: articulated header, rich footer, breadcrumbs, sidebar, interactive components. Plain text represents a smaller fraction of total code. A wall-of-text page with minimal layout has a high text/HTML ratio and, according to these data, tends to rank lower.
Core Web Vitals do not move the needle
This is the most surprising result. No CWV crosses the significance threshold:
- LCP: r=-0.011, p=0.36
- FCP: r=-0.007, p=0.57
- CLS: r=-0.002, p=0.84
- Lighthouse performance score: r=+0.009, p=0.45
- TTFB: r=+0.014, p=0.26
The most plausible explanation: sites in the top ten for competitive queries are already fast enough not to be penalized. Performance variation within the top 10 does not produce a statistically relevant signal. This does not mean speed never matters, but that in the Italian top 10 it is not what differentiates positions.
Other non-significant factors: schema markup (r=-0.019, p=0.10), HTTPS (r=+0.010, p=0.37), title length (r=+0.002, p=0.90), URL depth (r=+0.003, p=0.77), OpenGraph (r=-0.013, p=0.26).
Takeaways
Keyword placement first. If a page needs to rank for a specific query, that query goes in the title, URL slug, H1, and meta description. It is the signal with the strongest correlation across 841 Italian keywords and 6,824 pages, confirmed in v1 as well.
Structure your content into sections. More meaningful H2s and rich HTML structure outweigh total word count. An 800-word well-structured article can outperform a 3,000-word continuous stream.
Do not sacrifice relevance for speed. Core Web Vitals do not differentiate positions in the Italian top 10. Chasing a 200ms LCP improvement makes no sense if the keyword is not yet in the title. Performance still matters, but without a strong direct SEO impact within the top 10.
The .it TLD is worth choosing. If you are launching a site targeting the Italian market and still choosing a domain, the data supports .it over international variants.
View the infographic with the key results: View infographic · Versione italiana
Paper published on Zenodo: Download PDF
Manetti, G. (2026). Google Italy Ranking Factors 2026: A Correlational Analysis of 6,824 SERP Pages Across 57 Technical and On-Page Signals (3.0). PerseoDesign. https://doi.org/10.5281/zenodo.20797976
Dataset: June 2026. Scraping with Playwright/Chromium from google.it (it-IT), analysis with PerSeo Insights, Spearman correlation with scipy.