Ranking factors 2019–2026: what changed

93

months of data

January 2019 – September 2026

619

distinct factors

200 ranked each month

18 600

factor-months

one row per factor per month

10

pillars

the topical-authority framework

Four findings, each computed from the dataset:

1. Authority still heads the latest month. In September 2026 the strongest factor is Number of Referring Domains (rank 1, r −0.28), and the top of the list is off-page: links and referring domains.
2. The most durable strong factor is about entities. Number of Distinct Entities Used has been in the top 200 for 81 of 93 months with an average r of −0.27. The rest of the durable list is variation: different words for the same thing in headings, paragraphs, links and lists.
3. Entity & topical coverage grew; query-term placement shrank. Entity & topical coverage went from 2% of the top 200 in 2019 to 14% over the last twelve months (+11.8 pts); Query-term placement went from 44% to 29% (−14.4 pts). Pillar membership is a keyword-rule mapping, which the caveats below say plainly.
4. Correlations are modest and got slightly weaker. The average r of the top 10 factors was −0.27 in 2019 and −0.23 in 2026 (9 months, month-to-date); across the whole top 200, −0.16 and −0.16. No single factor explains a ranking.

The source is TopSEOFactors.com monthly 'Top 200 SEO Factors' pages, compiled into an analyst workbook. Each month lists the 200 factors most correlated with position in Google, with the correlation, the value of the number-one result, the average across page-1 results and how often the factor is used. The factor library in AwesomeSEO imports those pages, keeps every row, and maps each factor to one of ten topical-authority pillars by keyword rule.

Caveats, as recorded in the dataset

1. Correlation is negative because rank 1 is the top result: more negative means a higher factor value goes with a better position.
2. Correlation is not causation. Several factors are proxies (page weight, include counts) that rise with richer pages.
3. The data is global and mostly US/English SERPs. Treat page-1 averages as benchmarks for Nordic markets, not quotas.
4. The latest month may be month-to-date.
5. Pillar mapping is keyword-rule judgement (Koray framework), not measured data.

Every table below should be read with these five sentences in mind. Where a finding depends on the pillar mapping, the paper says so.

Correlation is negative because rank 1 is the top result: more negative means a higher factor value goes with a better position. In September 2026 the range runs from Number of Referring Domains at r −0.28 (rank 1) to Uses Schema description at r −0.12 (rank 200). A correlation of −0.28 is modest: rankings are decided by many factors at once, and the strongest of them explains only a slice of the order.

Two values travel with each factor. The number-one result’s value is one page, often an outlier. The page-1 average is the benchmark the product uses. For Number of Unique LSI Words Used , for example, the number-one result shows 908 while page 1 averages 250: reach the average, do not chase the outlier.

The right question for a brand is therefore not “which factor matters most?” but “on which factors does my page sit outside the top three of its own SERP?”. That is what SERP Lab measures.

Factors in the top 200 for at least 60 of the 93 months, strongest average correlation first. Durable factors are the ones to build a template on; a factor that spikes for a year is a curiosity.

Read the pattern, not the names: distinct entities, and variation of the same terms across HTML, headings, paragraphs, links and lists. Pages that name real things and describe them in more than one way have sat near the top of this list for the whole window.

Rank in the latest month against the same month a year earlier. A climb means the factor correlates more strongly than it did; a fall means less. Both lists are read with the month-to-date caveat.

The fallers are mostly density and exact-match measures. The climbers are structure and presence: headings, forms, a charset, social accounts. The direction matches what the playbook retired in 2024 (keyword density) and what it keeps (structure like a page-1 result).

Each factor is mapped to one pillar of the topical-authority framework by keyword rule. The mapping is judgement (pillar mapping is keyword-rule judgement (koray framework), not measured data), so the pillar table says where the dataset’s attention went, not what Google weighs.

Entity & topical coverage (P1) and structured data (P4) took the share that query-term placement (P2) gave up: the list counts more entity and markup factors and fewer query-placement factors than it did in 2019. For a new site this matters because entity coverage, structure and structured data are the pillars it controls from day one; authority (P7) it does not.

The source’s own categories, as a share of the top 200 each year, and the average correlation of the whole list and of its top 10. Relevancy has held between 39% and 47% of the list in every year.

The data is global and mostly US/English SERPs. Treat page-1 averages as benchmarks for Nordic markets, not quotas. With that said, the practices below are the ones the playbook ties to this data. Each is tier measured: it correlates with position in the dataset, and the product tests it on your own pages before it is presented as a finding.

• Write about entities; keywords are their attributes

  Name the real things (products, places, people, models, standards) and describe them through their attributes.

  • Number of Distinct Entities Used #45 · r −0.18
  • Entities in Title Tag #73 · r −0.17
  • Entities in H2 Tags #44 · r −0.18

• Answer the heading in its first sentence

  Open each section with a direct answer of about 40 words, exact where consensus allows, with the answer in bold. Use real questions as H3s.

  • Number of Questions In Headings #76 · r −0.16

• Give every money page a component that shows what it does

  Add a calculator, finder, comparison table, size guide or quote form above the fold.

  • Has Form #65 · r −0.17
  • Number of Images #12 · r −0.21

• Structure like a page-1 result

  One H1, about 7 H2s and 9 H3s on a money page, one H2 per attribute with data.

  • Number of H1 Tags #22 · r −0.20
  • Number of H2 Tags #23 · r −0.20

• Use the niche’s full vocabulary, not repeated keywords

  Cover related terms, jargon and attribute words naturally. Do not stuff them into the title.

  • Number of Unique LSI Words Used #5 · r −0.24
  • LSI Words in Title Tag #150 · r −0.14
  • Clean Keyword Density in the HTML Tag #134 · r −0.14

• Cover the variants Google bolds

  Collect the words Google highlights in top-20 snippets (singular, plural, definite forms, synonyms) and use them across headings and body.

  • Variations in H1-H6 Tags #28 · r −0.19
  • Variations in H1, H2, and H3 Tags #26 · r −0.19

• Keep cost of retrieval low: server response under 200 ms

  Aim for a server response under 200 ms (100 ms on money pages), clean HTML and few includes.

  • Load Time Milliseconds #6 · r −0.23
  • Number of Includes #9 · r −0.21

• Ship JSON-LD on every template

  Organization and WebSite sitewide; Product, Article, BreadcrumbList, ItemList and LocalBusiness where they apply; SoftwareApplication for apps.

  • Has JSON-LD Schema Markup #11 · r −0.21
  • Uses JSON-LD name #71 · r −0.17
  • JLDapplicationCategory

• Earn referring domains with data, not deals

  Publish original data and indexes, supplier and partner pages, local press and trade-body memberships.

  • Number of Referring Domains #1 · r −0.28
  • Number of Do Follow Backlinks #2 · r −0.27
  • Number of Backlinks #3 · r −0.27

• Keep a natural link mix

  Branded and descriptive anchors, followed and nofollow links, image links, in proportions normal for the niche.

  • Number of No Follow Backlinks #4 · r −0.25

• Claim every profile that describes the brand

  Social profiles, Google Business Profile, Bing Places, trade directories and Wikidata, with identical name, address, phone and description.

  • Number of Social Accounts #79 · r −0.16
  • Number of Facebook Pages #15 · r −0.20
  • Number of Instagram Pages #41 · r −0.18

The order of work for a Nordic brand follows the pillars it controls: entities and structure first (P1, P3), structured data that matches the page (P4), a fast template (P5), then the authority that takes years (P7). The academy track Evidence-based SEO walks through each practice with its tier.

The analyst workbook of monthly Top 200 pages is imported by a script in the AwesomeSEO repository (packages/factors/scripts/import-topseofactors.py) into plain JSON: one row per factor per month, a summary per factor (months in the list, average and yearly correlations, rank change year on year), the pillar mapping and the category shares. The package @awesomeseo/factors exposes those tables, and this page is rendered from them at build time; a re-import changes every number here at once.

Each factor’s source page follows the pattern https://topseofactors.com/factor.php?id={id}, which is where every factor link on this page goes. Imported 2026-10-20. The interactive library with all 619 factors is on the evidence page.