State of Google Search 2026: a Statistical Analysis

We analyzed 10,000 keywords across 10 industries, examining the top search results for each. Here's what we learned about how Google's search results actually work in 2026.

A note on methodology

Classifying keywords into industries and intents naturally involves some overlap. For example:

• Is "best dog food brands" a Pets keyword or Product Research?
• Is "anti-inflammatory diet for arthritis" a Health keyword or Food?
• Is "how to choose a mattress" Informational or Commercial?

Industry classifications aren't always clear-cut, as some keywords span multiple categories. The overall patterns remain reliable. For a full breakdown of our methodology, see below.

1. Industries and Intent

Before diving into the data, it's important to understand that different industries have very different intent profiles. Intent shapes everything from what types of content rank to whether AI Overviews appear.

The Industries

• Health: Medical conditions, symptoms, treatments, supplements
• Finance: Banking, investing, loans, credit cards, taxes
• Tech: Software, hardware, apps, programming, cybersecurity
• Product Research: "Best X", "X vs Y", buying guides, product comparisons
• Travel: Destinations, hotels, flights, trip planning
• Legal: Laws, regulations, legal processes, attorney services
• Real Estate: Property listings, home buying/selling, mortgages
• Education: Learning, courses, schools, academic topics
• Pets: Pet care, breeds, training, pet products
• Food: Recipes, cooking, restaurants, nutrition

Intent Categories

Keywords have been categorized by search intent:

• Informational: The searcher wants to learn something ("what is compound interest", "how to train a puppy")
• Commercial: The searcher is researching before a purchase ("best running shoes", "iPhone vs Samsung")
• Transactional: The searcher wants to buy or do something ("buy Nike Air Max", "book flight to Paris")

Intent Distribution by Industry

There is a large variance in industry intent profiles. Some are dominated by informational queries, others by transactional:

Key patterns:

• Food, Health, and Education are heavily informational (73-77%). Searchers want to learn, not buy.
• Product Research is majority commercial (52%). Searchers are comparing options before purchase.
• Real Estate is the most transactional (31%). Searchers want to find listings and take action.
• Tech is split almost evenly between informational (46%) and commercial (43%).

2. Platform Distribution

• 90.3% of rankings go to websites
• Nearly 10% of Google's top 10 organic results are occupied by non-website platforms such as forums, social media, and video sites.
• 4.9% go to forums/UGC (Reddit, Quora)
• 3.1% go to social media (Facebook, Instagram, Pinterest, TikTok, Twitter)
• 1.7% go to video (YouTube, Vimeo)
• Food has 4.7x more non-website results than Legal (19% vs 4%)
• Food searches are 12x more likely to show YouTube than Finance (42.4% vs 3.4%)

3. SERP Features

Although we are particularly interested in the traditional "10 blue links" i.e. the top ten ranking websites for each search term, we have to acknowledge that the SERP is much more diverse than this with many additional features.

• 78% of SERPs include People Also Ask
• 50% of SERPs include AI Overview
• 34% of SERPs include a featured snippet
• 23% of SERPs include video carousel
• When AI Overview is present, featured snippets appear 28% less often.

4. AI Overviews

48% of keywords trigger an AI Overview, but rates vary dramatically by industry and intent.

AI Overview Rate by Industry

• Health has the highest AI Overview rate at 73.6%, while Real Estate has the lowest at 22%

We know already that Health's intent distribution is largely skewed towards informational, so it is important to consider the AI Overview appearances also by intent.

AI Overviews by Intent

• 60% of informational queries get AI Overviews
• Only 24% of transactional queries get AI Overviews
• Informational queries are 2.5x more likely to trigger AI Overviews than transactional ones

AI Overviews by Industry AND Intent

When we control for both variables, we see that industry matters more than intent. Health transactional (67.8%) outperforms Real Estate informational (51.5%).

Where AI Overview Citations Come From

• 52% of AI Overview citations come from domains outside the organic top 10
• Positions 1-3 contribute 24% of all citations
• When an AI Overview appears, it cites 3-8 sources on average. Health AI Overviews cite the most (7.6 avg) and Real Estate the least (3.3 avg). The full range is 1-21 sources.

Top Citation Sources

Of the 10,000 keywords analysed, 4,780 triggered an AI Overview. Of those, we were able to retrieve sources from 3,237 (68%).

Across those 3,237 AI Overviews with citations, we found 28,768 total citations.

Top 10 most-cited domains in AI Overviews:

• YouTube is cited in 44.2% of all AI Overviews that include citations
• Reddit is cited in 22.9%
• User-generated content platforms (YouTube, Reddit, Quora, Facebook) make up 4 of the top 10
• The top 4 are all UGC platforms (YouTube, Reddit, Facebook, Amazon reviews)

Top Citation Sources by Industry

YouTube is the most cited source across every industry.

In YMYL industries (Health, Finance), Reddit gets replaced by authoritative sources like Mayo Clinic, Bankrate, and IRS.gov.

Top Citations from Domains Ranking in the Top 10

When a domain both ranks in the organic top 10 and gets cited in the AI Overview for the same query:

Top Citations from Domains NOT Ranking in the Top 10

These domains get cited in AI Overviews even when they don't appear in the organic top 10 for that query:

These authoritative sites often get cited even when they don't directly answer the query:

• "natural remedies for kidney infection": The top 10 includes Healthline, HealthGrades, Kidney.org, and MedicalNewsToday. But the AI Overview cites WebMD, which doesn't rank for this query.
• "what are personal loan rates": The top 10 is dominated by banks (Wells Fargo, US Bank, Navy Federal). But the AI Overview cites NerdWallet, an explainer site that doesn't rank.

5. Domain Concentration

Domain concentration measures how much of an industry's search visibility is controlled by a small number of websites. High concentration means a few dominant players capture most rankings, making it harder for new sites to break in. Low concentration means rankings are spread across many different domains, suggesting more opportunity for newcomers.

• 500 domains hold 63% of all page one positions
• 72.9% of domains appear only once in the top 10 results
• The top 10 domains per industry hold 11-37% of all positions in that industry
• The top 10 domains in the Real Estate industry hold 36.9% of positions
• The top 10 domains in the Legal industry are the least concentrated, holding only 11.7% of positions
• Reddit appears in 10/10 industries, averaging position 3.5
• YouTube holds 1.7% of all top 10 positions

The Reddit Anomaly

Reddit is the most frequently appearing domain in 6 industries, but it never ranks #1 in AI Overview citation frequency. Additionally, it's completely excluded from AI Overview citations in YMYL industries (Finance and Health), which tend to favour larger publications and banks:

Multiple Appearances in a Single SERP

14.5% of SERPs have at least one domain appearing 2+ times in the top 10.

Domains that most frequently appear multiple times:

By industry:

Real Estate: The Zillow Monopoly

Zillow appears in 60.8% of all real estate SERPs. When it appears, it ranks #1 nearly 6 out of 10 times (59.3% #1 rate) with an average position of 2.5.

Head-to-head: Zillow vs competitors

The "stuck at the bottom" effect

Legacy brokerages appear frequently but can never crack the top positions:

The top 5 real estate portals (Zillow, Realtor.com, Trulia, Redfin, Homes.com) control 26.8% of all real estate search positions, the highest concentration of any industry.

6. Small Sites vs Large Sites

We classified sites by how often they appear in search results as a proxy for authority. Large sites are those that appear 100+ times across our dataset, while small sites appear only 1-4 times.

• Small sites hold only 21% of top-3 positions when the Keyword Difficulty is above 80
• Small sites hold 44.5% of top-3 positions when the Keyword Difficulty is below 20
• When a small site beats a large site on the same SERP, the small site has the keyword in the title 14.5% of the time vs only 5.1% for the large site.

Real Examples: Topical Relevance Beats Authority

Query: "how to remove dark circles permanently"

• Small site (#2): "How to Remove Dark Circles Permanently: Expert Tips"
• Large site (#8): "The Best Eye Creams, Tested and Reviewed"

Query: "best running shoes for flat feet"

• Small site (#1): "The 11 Best Running Shoes for Flat Feet"
• Large site (#9): "Our Favorite Running Shoes Right Now"

Query: "how long do dental implants last"

• Small site (#2): "How Long Do Dental Implants Last? Longevity Guide"
• Large site (#10): "Everything You Need to Know About Dental Implants"

7. Content Structure

We classified all ranking URLs by page type to understand what types of pages actually rank:

Based on 76,014 classified URLs (86.7% of total). 11,702 URLs (13.3%) could not be classified into a specific page type.

Certain industries have much more editorial content than others. Here's how it breaks down:

For the following sections on content structure, word count, freshness, and title patterns, we've only analysed editorial content rather than homepages, landing pages, tool pages, or listing pages.

7A. Word Count

• The median word count of content in position 1 is 2,246 words
• The median word count of content in position 10 is 1,682 words
• Position 1 is 33% longer than position 10

Longer content ranks better, but the relationship is industry-dependent. YMYL topics (finance, health, legal) show the strongest length-ranking correlation. Comprehensive coverage signals expertise in high-stakes areas. Lifestyle topics (food, pets) show weaker correlation; a focused 1,500-word recipe can outrank a meandering 4,000-word one.

• Tech has longest content: 2,546 median words
• Pets has shortest content: 1,509 median words
• Finance shows biggest position gap: top 3 content is 38% longer than positions 4-10
• Food shows smallest position gap: top 3 is only 16% longer

7B. Content Structure: Word Count is What Matters

When we normalized structural elements per 100 words, the differences between positions virtually disappeared:

Higher-ranking content doesn't have proportionally more headings, images, or links. It just has more words, which naturally means more of everything.

Do TOCs, FAQs, and tables help rankings?

Longer content is more likely to have these features:

• 57% of 4,000+ word articles have a TOC vs only 15% of articles under 1,500 words
• 56% of 4,000+ word articles have an FAQ vs only 17% of shorter articles

But when we compared articles of the same length, FAQs and tables showed no ranking benefit. The only exception is TOC, which shows a small independent effect: top-ranking pages are 5-10% more likely to have a TOC even when comparing articles of similar length.

7C. Content Freshness

• Only 37% of editorial pages have machine-readable published dates
• Finance content is freshest: 12.8 months median age
• Food content is oldest: 39 months median age
• Content ranking positions 1-3 is 36 days older than positions 4-10 on average
• Food: top 3 content is 216 days older than positions 4-10
• Real Estate: top 3 content is 57 days newer than positions 4-10
• Industries where newer ranks better: real estate, education, legal
• Industries where older ranks better: food, pets, health, travel

Year in title:

• 10% of editorial pages have a year in the title
• 33.4% of those are outdated (2025 or earlier while it's 2026)

8. Content Homogeneity

Detecting AI-generated content is extremely difficult if not impossible. Therefore we simply measured how formulaic or predictable content is using a composite "homogeneity score" (0-100 scale, higher = more formulaic).

What we measured:

• Perplexity (GPT-2): How predictable the text is to a language model
• Burstiness: Variation in sentence lengths (human writing tends to have more varied "bursts")
• Type-Token Ratio: Vocabulary diversity (lower = more repetitive word usage)
• Hapax Ratio: Percentage of words appearing only once (lower = less vocabulary richness)

• Clear downward trend: higher-ranking content is more formulaic
• Position 1 scores 63.3 vs 59.1 for position 10
• The difference is fairly minimal (about 4 points on a 100-point scale)

---

Methodology

Step 1: Keyword Collection

• 918,000+ keywords collected across 10 industries (health, finance, tech, product research, travel, legal, real estate, education, pets, food)
• Keywords sourced from industry-specific seed lists and expanded via related keyword tools (Ahrefs)
• All keywords monthly search volume between 100-100,000

Step 2: Stratified Sampling

• Sampled down to 1,000 keywords per industry (~10,000 total)
• Stratified sampling to maintain representative distribution across:
  • Search volume (low, medium, high)
  • Search intent (informational, commercial, transactional)
• Final dataset: 10,000 keywords

Step 3: SERP Data Collection

• DataForSEO API used to collect top 10 organic results for each keyword
• 87,716 total rankings captured
• 76,995 unique URLs across 30,880 unique domains
• US-only, desktop SERPs
• Point-in-time snapshot (February 2026)

Sample Sizes

Limitations

• US-only search results
• Desktop SERPs only (no mobile)
• Point-in-time snapshot (rankings change daily)
• Homogeneity score measures linguistic uniformity, not definitively AI vs human authorship