How AI-generated content performs in Google Search: A 16-month experiment

With AI, you may generate dozens (if not a whole lot) of articles in hours and publish at scale. However publishing is the simple half. What occurs after they go dwell is what issues.

Along with the analysis staff at SE Rating, we ran a 16-month experiment to trace how nicely AI-generated content material carried out on brand-new domains with zero authority.

As you will note, the outcomes are exhausting to name a hit.

Right here’s the total story behind our experiment.

Methodology

The objective was easy: check how far AI content material — with no human modifying, rewriting, or enhancement — may go in search.

How rapidly would it not get listed? May it rank for related queries? Most significantly, may it drive site visitors?

We began by buying 20 new domains with no backlinks, area authority, model recognition, or search historical past.

Every area centered on a unique area of interest, protecting matters comparable to:

• Arts & Leisure
• Enterprise & Companies
• Neighborhood & Society
• Computer systems & Expertise
• Ecommerce & Buying
• Finance & Accounting
• Meals & Drink
• Video games & Equipment
• Well being & Medication
• Business & Engineering
• Hobbies & Pursuits
• Dwelling & Backyard
• Jobs & Profession
• Regulation & Authorities
• Way of life & Properly-being
• Pets & Animals
• Science & Training
• Sports activities & Health
• Journey & Tourism
• Autos & Boats

For every area of interest, we gathered 100 informational “how-to” key phrases—long-tail phrases with decrease competitors.

Every website obtained 100 AI-generated articles, totaling 2,000 items throughout the experiment.

After publishing, we added the websites to Google Search Console and submitted sitemaps.

From that time on, we left the websites untouched to look at efficiency over time.

Timeline & key outcomes 

Month 1: indexing and early visibility

About 71% of latest AI-generated pages had been listed inside the first 36 days. They generated over 122,000 impressions and 244 clicks. Even at this early stage, 80% of websites ranked for no less than 100 key phrases every.

Months 2–3: development continues

Cumulative impressions grew to over 526,000, with 782 clicks. Content material continued to carry out nicely with out backlinks, promotion, inner linking, or further search engine optimisation techniques.

Months 3–6: rating collapse

By about three months, solely 3% of pages remained within the prime 100. Early relevance helped pages get listed and briefly seem in search, however with out authority, uniqueness, or E-E-A-T indicators, rankings dropped sharply. Google nonetheless listed the pages, however customers hardly ever noticed them.

Month 16: long-term stagnation

After over a 12 months, visibility remained low throughout most websites. Impressions and clicks had been minimal, and no website confirmed significant restoration. After the August 2025 Google spam update, pages rating within the prime 100 rose to twenty% — up from 3% at six months.

Month 1: indexing and early visibility

Simply over a month after publication (36 days), the primary outcomes got here in — and so they had been stronger than anticipated for brand-new websites.

Of two,000 articles, 70.95% had been listed (1,419 pages). For zero-authority domains, that’s notable, as getting new websites totally listed is commonly a problem. This reveals Google remains to be keen to crawl and index AI-generated content material usually.

Some websites carried out significantly nicely. Eleven of the 20 domains had all 100 pages listed.

• Most had been in broad, evergreen niches like Meals & Drink, Dwelling & Backyard, Jobs & Profession, and Way of life & Properly-being.
• Extra aggressive or specialised areas, like Ecommerce & Buying, noticed slower indexation, possible because of stricter analysis.

Together with indexation got here early visibility. Throughout this primary month, the websites collectively generated:

• 122,102 impressions
• 244 clicks

A number of niches stood out producing greater than 10,000 impressions within the first month alone.

• Hobbies & Pursuits: 17,425 impressions

• Enterprise & Companies: 17,311 impressions

• Journey & Tourism: 13,598 impressions

• Way of life & Properly-being: 13,072 impressions
• Regulation & Authorities: 11,794 impressions
• Video games & Equipment: 11,083 impressions
• Autos & Boats: 10,677 impressions

When it comes to key phrase protection, many websites carried out surprisingly nicely inside the first month. Eight websites ranked for greater than 1,000 key phrases, whereas one other eight ranked for 100 to 1,000.

Even at this early stage, 80% of websites with totally AI-generated content material appeared in seek for a whole lot or 1000’s of queries.

Notably, over 28% of rating URLs had been already within the prime 100. Throughout the first month, many pages reached positions the place searchers may see them.

General, these outcomes present AI-generated content material can achieve traction rapidly—even with out backlinks, editorial enter, or further search engine optimisation work. Within the brief time period, content material alone was sufficient to get listed and seem in search.

Months 2–3: development continues

This early visibility wasn’t short-lived. Over the next weeks, impressions and clicks saved rising as Google Search found and examined pages.

By about two and a half months after publication, cumulative outcomes throughout all websites had grown:

• Impressions: 122,102 to 526,624
• Clicks: 244 to 782

Key phrase protection additionally expanded:

• 12 websites ranked for 1,000+ key phrases (up from 8 within the first month).
• The remaining 8 websites ranked for 100–1,000 key phrases.

This sample is typical for brand spanking new websites. When Google finds recent content material that matches actual queries, it assessments that content material throughout outcomes. Pages seem for associated queries as Google evaluates their helpfulness.

That’s what occurred right here. Even with out backlinks, inner linking, or search engine optimisation enhancements, the content material gained publicity as a result of it focused low-competition queries and adopted fundamental search engine optimisation construction.

At this stage, it may appear to be a robust case for large-scale AI content material. The websites had been new, the content material totally AI-generated, and impressions saved rising.

However the development didn’t final.

Month 3-6: the rating collapse

Round Feb. 3, 2025, roughly three months after publication, the experiment hit a turning level.

• Solely 3% of pages remained within the prime 100, down from 28% within the first month. 

In sensible phrases, the content material remained listed however hardly ever appeared the place customers may see it.

Early relevance may also help pages get listed and seem in search outcomes for a time. With out stronger indicators — authority, E-E-A-T, distinctive insights — these rankings are exhausting to maintain.

By the six-month mark, Google Search Console confirmed the next cumulative totals throughout all websites:

• Impressions: 526,624 to 706,328 
• Clicks: 782 to 1,062

At first look, these numbers counsel continued development. However that’s not what occurred.

Most exercise occurred early. Within the first 2.5 months, the websites generated roughly 70% to 75% of whole impressions and clicks. Over the subsequent 3.5 months, development slowed sharply, including solely 25% to 30%.

Month 16: the long-term image

The experiment ran for over a 12 months to see if rankings would get better.

For probably the most half, they didn’t.

After the drop across the three-month mark, visibility remained extraordinarily low for the remainder of the experiment.

There have been just a few transient fluctuations. Probably the most notable got here in late August 2025.

Beginning in August, 50% of websites (10 out of 20) noticed a two-week spike in impressions. This intently aligned with the rollout of the Google August 2025 spam replace, which began Aug. 26.

Nonetheless, the increase didn’t result in a sustained restoration.

Among the many websites that noticed a short-term carry:

• Six rapidly misplaced visibility and returned to prior lows
• 4 maintained barely improved efficiency, just like early post-publication ranges

Following the replace, pages rating within the prime 100 rose to twenty% — up from 3% at six months. This remained beneath the 28% seen within the first month, however the August 2025 spam replace appeared to have improved some rankings.

In whole, 66.9% of pages had been nonetheless listed, up barely from 61.45% at six months.

The next websites had a few of the lowest numbers of listed pages:

• Finance area (9 of 100)
• Well being area (14 of 100)

That is possible because of their YMYL nature, the place Google applies stricter high quality and belief requirements.

By month 16, cumulative outcomes throughout all websites had been:

• Impressions: 706,328 to 1,092,079
• Clicks: 1,062 to 1,381

Most impressions nonetheless got here from the early development section, earlier than rankings dropped.

Why search engine optimisation visibility didn’t final

The obvious clarification is that the content material didn’t meet Google’s high quality requirements — and understandably so.

The two,000 articles lacked many indicators Google makes use of to evaluate high quality and belief:

• Authority. No backlinks or exterior validation. With out these, new domains battle to compete with established websites.
• Experience and credibility. No authors, credentials, or real-world experience — particularly essential in finance, well being, and regulation.
• Content material differentiation. A lot of the content material resembled what already exists. With out distinctive insights, pages battle to face out.
• Website construction. No inner linking, topical group, or clear hierarchy to assist Google perceive web page relationships.

Google can determine AI-generated patterns. With out authority, uniqueness, or supporting indicators, early visibility declines.

Bonus perception: how new AI content material helps present pages

In early March 2026, we ran a follow-up experiment, including new AI-generated content material to eight tracked websites.

As of March 13, not all new content material has been listed. Nonetheless, websites with new content material already present a noticeable improve in search impressions.

Apparently, this carry comes primarily from older posts, not the newly printed ones.

For instance:

• Enterprise-focused web site (from 458 impressions in February 2026 to 7,750 impressions in March 2026)  – 17x improve.

• Regulation-focused web site (from 19 impressions in February 2026 to 356 impressions in March 2026)  – 19x improve.

• Science-focused web site (from 34 impressions in February 2026 to 633 impressions in March 2026)  – 19x improve.

This experiment reveals that publishing new content material—even totally AI-generated—can carry site visitors to older pages that had been stagnant for months. Recent content material could sign to Google that the location is energetic and updated, giving the location a brief increase.

Nonetheless, these are early outcomes and don’t assure lasting good points in rankings or site visitors.

Key takeaway: AI can velocity up content material creation, however not substitute search engine optimisation

The outcomes of this 16-month experiment don’t imply AI content material is ineffective. They present AI alone isn’t sufficient to drive lasting affect.

Early site visitors and impressions could look promising, however with no clear search engine optimisation technique and human steerage, these good points will possible fade inside just a few months.