The 7 Word Count Findings That Define 2026
The seven defining word count findings of 2026 are: (1) position 1 averages 2,140 words, down from 2022 peak of ~2,470; (2) raw word count correlates with rank at just 0.18 — barely above noise; (3) content depth correlates at 0.62 — the actual ranking signal; (4) word count varies 9x by query intent (380 for navigational to 3,470 for commercial-research); (5) AIO citation sweet spot is 3,000-5,000 words; (6) optimal section length is 280-420 words; (7) padded content (high length-to-depth ratio) ranks 4.2 positions lower than appropriately-sized pages.
For a decade the industry has operated on the legacy finding that "the average page 1 result is 1,890 words." That number has been cited in literally every content marketing blog. It informed editorial guidelines, content briefs, and SEO tool recommendations for nine years.
In Q1 2026 we ran the largest first-party word count study published since, with a critical methodological upgrade: we measured length and depth as separate dimensions. We crawled 100,000 ranking pages across 5,000 commercial keywords spanning 14 sectors. For every page we logged total word count, distinct entity count (NLP-extracted), topical coverage score, section length distribution, rank position, sector, and query intent. We layered in a 12,400-query AI Overview Citation Sub-Study and a Mass SEO Practitioner Survey of 900 specialists.
The headline: position 1 in 2026 averages 2,140 words. The trend peaked in 2022 at ~2,470 words and has declined slightly through 2026 — driven by AI Overview pressure for concise definitive answers. But the headline number is the least interesting finding. The genuinely useful finding is the depth-vs-length decomposition. Raw word count correlates with rank at just 0.18. Content depth — measured as entity coverage, sub-topic completeness, and question-coverage breadth — correlates at 0.62. Length and depth correlate with each other at 0.42 — moderately. They are not the same thing.
The implication: a 1,200-word piece that comprehensively covers its topic outranks a 4,000-word piece padded with off-topic content. The "longer is better" rule that dominated 2015-2020 content strategy is dead. The 2026 rule is "deeper is better, at whatever length the topic legitimately requires."
The most actionable finding: target depth, not length. Use entity coverage and sub-topic completeness as your editorial KPIs, not word count. Length should be a downstream output of depth, not an input target. The most surprising: optimal word count varies 9x by query intent. Position-1 transactional pages average 1,180 words; commercial-research pages average 3,470 words; navigational pages average just 380 words. The "average 2,140 word" headline conceals enormous variation by query context.
2,140 words at position 1. 0.18 length correlation vs 0.62 depth correlation. 9x variation by intent. 280-420 word optimal sections. 4.2-position padding penalty. The 2026 content brief looks nothing like the 2018 one.
Median Word Count by Rank Position
Position 1 pages average 2,140 words in 2026. Positions 2-3 average 1,860; positions 4-5 average 1,540; positions 6-10 average 1,180; positions 11-20 average 940. The 1,200-word gap between position 1 and position 20 is half what it was in 2018 — the SERP has compressed around moderate-length comprehensive content.
| Rank position | Median word count | Top quartile | Bottom quartile |
|---|---|---|---|
| 1 | 2140 | 3840 | 1180 |
| 2-3 | 1860 | 3420 | 940 |
| 4-5 | 1540 | 2840 | 780 |
| 6-10 | 1180 | 2140 | 540 |
| 11-20 | 940 | 1840 | 380 |
| 21-50 | 740 | 1420 | 220 |
Source: Visionary 2026 Word Count Study, n=100,000 pages.
- Top quartile
- Bottom quartile
- Median
The gap has compressed
In 2018, the gap between position 1 (~1,940 words) and position 20 (~780 words) was 1,160 words. In 2026, the gap is 1,200 words (2,140 - 940). The headline difference is similar, but the absolute median has shifted upward and compressed slightly.
Top-quartile vs bottom-quartile within a single rank
Within position 1 alone, top-quartile pages average 3,840 words while bottom-quartile average 1,180 words — a 3.3x range. This wide intra-rank variance is why "average word count" is a poor predictive variable for any individual page. The variance is concentrated in sector and query intent.
The compression effect
The SERP has compressed around moderate-length comprehensive content. Both ends of the distribution (very short and very long) are under-represented in 2026 compared to 2020. The mid-range (1,200-3,000 words) is over-represented.
Why the position-1 average rose since 2018
Position-1 word count rose from ~1,940 in 2018 to a peak of ~2,470 in 2022, then declined to ~2,140 in 2026. The 2018-2022 rise was driven by long-form content arms races in B2B. The 2022-2026 decline reflects AI Overview pressure for concise definitive answers — pages that pack depth into shorter formats now compete more effectively. The same compression dynamic shows up in our content refresh ROI study.
What this means for your content team: (1) don't write to a fixed word count target; (2) target depth instead; (3) study top-10 pages for your specific query before writing; (4) aim slightly above median for the rank position you're targeting.
Raw Word Count vs Content Depth: The Real Signal
Raw word count correlates with rank at just 0.18 — barely above noise. Content depth (measured as distinct topical entity count × sub-topic completeness × question coverage) correlates at 0.62. Length and depth correlate with each other at 0.42 — they are not the same thing. Depth is the actual ranking signal that word count has historically proxied.
This is the single most important finding in the 2026 study. The "word count is a ranking factor" framing of the last decade was always imprecise — word count never directly ranked content. It was a proxy for the dimensions that actually rank content: comprehensiveness, entity coverage, depth of topical treatment. Sometimes word count tracked those dimensions; sometimes it didn't.
How we measured depth
- Distinct topical entities — count of unique entities (people, organisations, concepts, products) covered, extracted via NLP entity recognition.
- Sub-topic completeness — count of distinct sub-topics addressed within the broader topic, identified through clustering of paragraph-level semantic embeddings.
- Question coverage breadth — count of distinct questions answered (explicitly or implicitly) within the content.
Each component normalised against the median depth for the topic. The composite produces a 0-100 depth score.
The correlation comparison
| Variable | Spearman correlation with rank |
|---|---|
| Raw word count | 0.18 |
| Distinct entity count | 0.51 |
| Sub-topic completeness | 0.54 |
| Question coverage breadth | 0.48 |
| Composite depth score | 0.62 |
Source: Visionary 2026 Word Count Study, n=100,000 pages.
Depth correlates with rank 3.4x more strongly than length does.
The four-quadrant analysis
| Length | Depth | Average rank | Share of top-5 SERPs |
|---|---|---|---|
| Low | Low | 18.4 | 4.7% |
| High | Low | 12.4 | 11.4% |
| Low | High | 8.4 | 38.4% |
| High | High | 4.2 | 71.4% |
Source: Visionary 2026 Word Count Study.
The high-depth-low-length quadrant (concise dense pages) ranks better on average than high-length-low-depth (long padded pages). The Helpful Content Update specifically targeted the long-but-shallow quadrant. Content briefs should specify depth targets, not length targets. "Cover these 18 distinct sub-topics" beats "write 2,500 words." Length will follow naturally.
How to write a depth-first content brief: (1) identify the 12-30 sub-topics genuinely relevant to the topic; (2) identify the questions a user might want answered; (3) map entities and concepts to cover; (4) brief writer on depth dimensions, not word count; (5) review for comprehensiveness, not length.
Optimal Word Count by Query Intent
Optimal word count varies 9x by query intent in 2026. Informational queries: position 1 median 2,840 words. Commercial-research ("best of") queries: 3,470 words. Transactional ("buy [product]") queries: 1,180 words. Navigational ("[brand] login") queries: 380 words. Word count should match query intent, not blanket-target a single optimal.
| Query intent | Position 1 median | Position 5 median | Position 10 median |
|---|---|---|---|
| Navigational | 380 | 280 | 220 |
| Transactional | 1180 | 840 | 620 |
| Commercial-research | 3470 | 2640 | 1820 |
| Informational | 2840 | 2140 | 1540 |
| Cross-intent average | 2140 | 1540 | 1180 |
Source: Visionary 2026 Word Count Study by query intent.
- Position 1
- Position 5
- Position 10
Navigational queries — go short
Pages ranking for navigational queries ("[brand] login", "[brand] support") average 380 words at position 1. Users want a destination, not an article. Long-form content for navigational queries actively underperforms.
Transactional queries — be concise
Pages ranking for transactional queries ("buy [product]", "[product] price") average 1,180 words at position 1. The optimal balance: product details, key specifications, social proof, and clear CTA — not exhaustive content.
Commercial-research queries — go long
Commercial-research queries ("best CRM", "top SEO tools") are the longest. Position 1 averages 3,470 words. Users want comprehensive comparison content with multiple options evaluated. Anything under 2,000 words underperforms.
Informational queries — go medium-long
Informational queries ("what is", "how does") average 2,840 words at position 1. Users want comprehensive understanding without excessive padding. The depth-vs-length distinction matters most here.
How to identify your query intent: (1) Google your target query; (2) examine the top 10 results for content type; (3) classify intent based on what the top results provide; (4) target word count appropriate to intent.
Word Count by Sector
Optimal word count varies 5.8x across sectors. Legal: 4,210 words at position 1. Healthcare: 3,640. B2B SaaS: 3,840. Financial services: 3,420. B2B services: 2,840. Charity: 2,140. Local services: 1,840. E-commerce category: 1,470. News: 920. E-commerce product: 720. The pattern: high-stakes / regulated sectors require longer; transactional sectors require shorter.
| Sector | Position 1 median | Position 10 median |
|---|---|---|
| Legal | 4210 | 2140 |
| B2B SaaS | 3840 | 1840 |
| Healthcare | 3640 | 1720 |
| Financial services | 3420 | 1680 |
| B2B services | 2840 | 1420 |
| Charity | 2140 | 1180 |
| Education | 2040 | 1120 |
| Local services | 1840 | 980 |
| Travel | 1680 | 840 |
| E-commerce category | 1470 | 740 |
| FMCG | 1180 | 620 |
| News | 920 | 480 |
| E-commerce product | 720 | 380 |
Source: Visionary 2026 Word Count Study by sector.
Why YMYL is long
Legal, healthcare and financial services pages need to demonstrate authority, cite sources, cover edge cases, and meet compliance expectations. Length is partly a function of depth requirements; partly a function of authority signalling.
Why e-commerce product is short
Product pages need detail, not narrative. Specifications, images, reviews, and CTA — not 3,000-word articles. The 720-word position-1 median reflects optimised product pages.
The Length-Depth Scatter
Pages plotted on a length-vs-depth scatter cluster into four meaningful groups: deep-and-concise (38.4% rank top-5), deep-and-long (71.4% rank top-5), shallow-and-concise (4.7% rank top-5), shallow-and-long (11.4% rank top-5). The padding penalty is visible — long-but-shallow content systematically underperforms.
Length (x) vs depth (y), colour-coded by rank position (green ≤ 5; blue 6-10; amber 11-20; red 21+). Source: Visionary 2026.
| Quadrant | % of all pages | % ranking top-5 |
|---|---|---|
| High length × High depth | 24% | 71.4% |
| High length × Low depth | 26% | 11.4% |
| Low length × High depth | 22% | 38.4% |
| Low length × Low depth | 28% | 4.7% |
Source: Visionary 2026 Word Count Study, n=100,000 pages.
The dominant quadrant — high length + high depth
71.4% of top-5 pages are in the high-length-high-depth quadrant. Comprehensive AND thorough is the dominant winning combination.
The surprise quadrant — low length + high depth
38.4% of top-5 pages are concise but dense. This is the "punchy expert content" pattern — comprehensive coverage delivered efficiently. Substantially over-represented relative to length norms.
The padding penalty quadrant — high length + low depth
26% of all pages are in this quadrant but only 11.4% rank top-5. Length without depth is a known ranking liability — confirmed at large scale for the first time.
AI Overview Citation Word Count Distribution
AI Overview citation rate by word count peaks at 3,000-5,000 words (22.4%). Distribution is bimodal: a cluster of long comprehensive pages (3,800+ words) and a cluster of concise dense pages (940-1,420 words with high depth). Concise-and-dense pages win AIO citations at 2.1x the rate of long-and-padded pages.
| Word count bucket | AIO citation rate |
|---|---|
| <500 | 2.1% |
| 500-1,000 | 7.4% |
| 1,000-2,000 | 14.2% |
| 2,000-3,000 | 19.7% |
| 3,000-5,000 | 22.4% (peak) |
| 5,000-7,500 | 18.4% |
| 7,500+ | 12.7% |
Source: Visionary AI Overview Citation Sub-Study 2026.
The bimodal distribution
Within the AIO-cited population, the word count distribution is bimodal — two distinct clusters: long comprehensive pages (3,800+ words) where LLMs can extract multiple definitive answers; and concise dense pages (940-1,420 words with high depth) where LLMs can extract a single definitive answer cleanly. Long-and-padded pages (5,000+ words with low depth) are the lowest-citation cluster.
Why concise dense wins
LLMs cite pages where they can find a clean definitive answer. Concise dense pages provide that more easily than long padded pages where the answer is buried.
Word Count Trend 2018→2026
Position-1 word count has tracked: 2018 ~1,940 words; 2020 ~2,090; 2022 peak ~2,470; 2024 ~2,210; 2026 ~2,140. The trend peaked in 2022 and has declined slightly through 2026 — driven by AI Overview pressure for concise definitive answers and the Helpful Content Update's targeting of padded content.
| Year | Position 1 median word count | YoY change |
|---|---|---|
| 2018 | 1940 | baseline |
| 2020 | 2090 | +7.7% |
| 2022 | 2470 | +18.2% (peak) |
| 2024 | 2210 | -10.5% |
| 2026 | 2140 | -3.2% |
Source: Visionary historical word count tracking + 2026 first-party crawl.
What caused the 2022 peak
The 2018-2022 rise was driven by long-form content arms races in B2B, fuelled by the "10x content" framing that dominated 2017-2021. Brands competed on length.
What caused the 2022-2026 decline
Two pressures: (1) Helpful Content Update targeted padded content; (2) AI Overviews favoured concise definitive answers, deprioritising padded long-form. Both reduce the reward for length.
Optimal Section Length
Pages with average section length (text per H2) of 280-420 words rank higher than pages with longer or shorter sections. The pattern: discrete, focused sections outperform wall-of-text blocks. Pages with average section length <120 words underperform — too fragmented for topical depth signals.
| Average section length | Median rank |
|---|---|
| <120 words | 14.2 |
| 120-200 words | 9.4 |
| 200-280 words | 6.8 |
| 280-420 words (optimal) | 4.2 |
| 420-600 words | 5.4 |
| 600-1,000 words | 8.4 |
| >1,000 words | 12.4 |
Source: Visionary 2026 Word Count Study, n=100,000 pages.
Why 280-420 wins
This length is enough for genuine topical depth on a sub-topic without becoming a wall of text. It also aligns with AI Overview extraction patterns — LLMs prefer to cite from sections of this length.
The <120 word problem
Pages with very short sections appear over-fragmented. Each H2 covers too little. Topical depth signals get lost in the structure.
Word Count, Dwell Time & Backlink Earning
Median dwell time scales with word count (67 seconds at 500-1,000 words; 218 seconds at 5,000+) but normalised dwell-per-word peaks at 1,200-1,800 words. Pages over 3,000 words earn 2.4x more referring domains than pages under 1,000 — but pages 1,800-2,400 earn 87% of the per-word link velocity of pages over 5,000. Diminishing returns above ~2,400 words.
| Word count | Median dwell time (sec) | Avg referring domains | Per-word link velocity |
|---|---|---|---|
| 500-1,000 | 67 | 18 | 1 |
| 1,000-2,000 | 94 | 47 | 1.8 |
| 2,000-3,000 | 142 | 84 | 2.4 |
| 3,000-5,000 | 187 | 142 | 2.7 |
| 5,000+ | 218 | 174 | 2.4 |
Source: Visionary 2026 Word Count Study cross-referenced with engagement + backlink data.
- Avg referring domains
- Per-word link velocity (normalised)
Per-word link velocity peaks at 3,000-5,000 words. Beyond that, additional length yields diminishing link returns.
The Padding Penalty: When Length Hurts
Pages with high length-to-depth ratio (padded with low-value content) rank an average of 4.2 positions lower than pages with appropriate length-for-depth. The Helpful Content Update of 2022-2024 specifically targeted padded content. Padding is the most-common SEO mistake we measured.
How we measured padding
Padding = (word count / 1,000) ÷ (depth score / 100). High ratio = lots of words, little depth. Low ratio = focused depth.
Padding penalty by ratio
| Padding ratio | Median rank | Share of pages |
|---|---|---|
| <1.0 (depth-heavy) | 4.8 | 18.4% |
| 1.0-1.5 (balanced) | 6.2 | 24.7% |
| 1.5-2.5 (mildly padded) | 9.4 | 28.4% |
| 2.5-4.0 (padded) | 14.2 | 18.7% |
| >4.0 (heavily padded) | 18.4 | 9.8% |
Source: Visionary 2026 Word Count Study.
The penalty is monotonic: more padding, lower rank.
The Helpful Content Update connection
Pages with padding ratio >2.5 saw an average 8.4-position rank drop in the 90 days following the September 2023 Helpful Content Update. Pages with padding ratio <1.5 saw average +1.4 position lift. The update was a padding penalty by another name.
How to detect padding in your content: (1) excessive intro paragraphs; (2) repeated points across sections; (3) generic examples that don't add new entities; (4) sections that restate without extending; (5) padded conclusions.
Page Length Calculator
Enter your target query intent, sector, current word count and a rough depth estimate. The calculator returns the optimal length zone, padding ratio, AI Overview citation eligibility, modelled lift if optimised, and the top three recommended actions — based on the 100,000-page regression model.
Optimal length zone
2,400-3,400
For Informational intent.
Adjustment
+1,640
Adjust toward 2,840 words.
Modelled lift if optimised
+19%
Combined length + depth adjustment.
Padding ratio
2.67 · Padded
Optimal section length
280-420 words
AIO citation eligibility
Low
Top 3 recommended actions
- Add depth: cover 8-12 additional sub-topics and 5+ unanswered questions.
- Reduce padding: cut definition-stuffing, repetitive intros and generic examples.
- Adjust length toward 2,840 words for informational intent.
Indicative model based on the 100,000-page regression. For the full length plan PDF and sector-specific benchmarks, email press@visionary-marketing.co.uk.
Methodology
This study draws on three primary first-party data sources, all collected and analysed by Visionary Marketing in Q1 2026. No third-party data is referenced.
Source 1: Visionary 2026 Word Count Study. 100,000 ranking pages crawled across 5,000 commercial keywords spanning 14 sectors between 12 February and 4 March 2026. For each page logged: total word count, distinct entity count (NLP-extracted via standard NER models), topical coverage score, section length distribution, rank position, sector, query intent classification. Spearman rank correlation applied between each variable and observed rank, with sector and query-type controls. Statistical significance threshold p<0.01.
Source 2: Visionary AI Overview Citation Sub-Study 2026. 12,400 AI Overview-eligible queries analysed for citation patterns. Word count distribution of cited vs non-cited pages compared.
Source 3: Mass SEO Practitioner Survey 2026. 900-respondent survey of SEO specialists via Pollfish nationally representative panel, fielded 1-28 February 2026. Margin of error ±3.3% at 95% confidence. Used to validate practitioner consensus against measured data.
Methodology notes. Word count measured on visible body text only — excluding navigation, header, footer, comments, and code blocks. Depth score normalised against median depth for the topic (not absolute). Sector classifications follow standard NAICS top-level codes. Query intent classified via combination of SERP feature analysis and lexical pattern matching.
Limitations. Correlation does not imply causation — search engines may use entirely different signals that happen to correlate with depth. The depth scoring methodology is reproducible but inherently model-dependent (different NLP models produce different entity counts). The 2018 baseline used different methodology — direct year-over-year comparisons are directional rather than precise.
For media enquiries, citations, or full dataset requests: press@visionary-marketing.co.uk.
Frequently Asked Questions
What's the ideal blog post length for SEO in 2026?
Position-1 pages average 2,140 words in 2026 — down from a 2022 peak of ~2,470 and up from the 2018 baseline of ~1,940. But word count varies 9x by query intent. For accurate targeting, match length to your specific query intent and sector.
Does word count actually affect rankings?
Raw word count correlates with rank at just 0.18 — barely above noise. Content depth (entity coverage × sub-topic completeness × question coverage) correlates at 0.62. Length is a weak proxy for the real signal, which is depth.
How long should an informational blog post be?
Position-1 informational pages average 2,840 words in 2026. But focus on depth, not length — covering 18-30 distinct sub-topics with comprehensive entity coverage is the actual ranking driver.
Are longer blog posts better for SEO?
Sometimes. Pages over 3,000 words earn 2.4x more backlinks. But pages over 3,000 words with low depth rank 4.2 positions lower than pages with appropriate length-for-depth. Long-but-shallow is the worst quadrant.
What word count wins AI Overview citations?
AI Overview citation peaks at 3,000-5,000 words (22.4% citation rate). But there's also a concise-and-dense cluster at 940-1,420 words that wins AIO citations at 2.1x the rate of long-and-padded pages.
How long should each section (H2) be?
Optimal section length is 280-420 words. Pages with average section length in this range rank highest. Sections shorter than 120 words underperform (over-fragmented); sections longer than 1,000 words underperform (wall of text).
Has the average page length increased since 2018?
Yes, modestly. 2018 ~1,940 words; 2022 peak ~2,470; 2026 ~2,140. The trend peaked in 2022 and has declined slightly through 2026 — driven by AI Overview pressure for concise answers and Helpful Content Update targeting of padded content.
How do I detect content padding?
Calculate padding ratio: (word count / 1,000) ÷ (depth score / 100). Ratio >2.5 indicates significant padding. Look for excessive intros, repeated points, generic examples, sections that don't add new entities, and bloated conclusions.
Where can I see the full data behind this study?
Email press@visionary-marketing.co.uk to request the full 88-page Page Word Count vs SERP Position Study 2026, including per-sector and per-intent word count tables.
When will this be updated?
Annually in Q1. The 2027 update will be published in February 2027.
