SEO

Substack launched a new algorithmic feed for Notes yesterday, completing its transformation from “newsletter platform” to “Twitter competitor that actually pays creators.”

The change is significant: Notes previously showed reverse-chronological posts from people you followed. Now it’s “For You”—algorithmically curated content designed to maximize engagement.

What Changed

The algorithm arrived. Previously: follow someone, see their posts. Now: follow someone, maybe see their posts, definitely see content the algorithm thinks you’ll engage with.

Medium emailed creators Friday evening: starting April 1, Partner Program eligibility requires 1,000 followers (up from 100) and consistent publishing for 6 months.

The message is clear: Medium wants fewer, more established creators—not the “post once, hope for viral” crowd.

What Actually Changed

Old requirements: 100 followers, published at least one story. New requirements: 1,000 followers, published consistently for 6+ months.

This eliminates roughly 85% of current Partner Program members. Medium is betting that quality over quantity improves reader retention and subscription conversion.

The death of SEO has been announced annually since 2012. This time it’s different—not because SEO is gone, but because what replaces it has finally arrived.

Generative Engine Optimization (GEO) isn’t a buzzword anymore. It’s the difference between being found and being invisible.

What changed

Traditional SEO optimized for Google’s ranking algorithm. Keywords, backlinks, meta tags—all designed to convince an algorithm your page deserved position one.

GEO optimizes for AI answer engines. ChatGPT, Claude, Perplexity—these tools don’t rank pages. They synthesize answers from multiple sources and present single responses.

Google’s AI Overviews launched with a promise: accurate, synthesized answers at the top of search results.

A year later, the data tells a different story. Error rates are up. Accuracy is down. And content creators are caught in the crossfire.

The Error Rate Problem

Data from SEO monitoring tools:

BrightEdge, which tracks millions of search queries weekly, reported in their March 2026 analysis that AI Overview error rates increased 23% from Q4 2025 to Q1 2026. “Error” is defined as factual inaccuracies, misattributed sources, or contradictory information within the same overview.

Medium emailed creators Friday evening: starting April 1, Partner Program eligibility requires 1,000 followers (up from 100) and consistent publishing (minimum 2 posts monthly). Writers who don’t meet the threshold lose monetization immediately.

The Impact

An estimated 40,000 Medium writers will lose Partner Program access. These aren’t hobbyists—many are journalists laid off from traditional media, using Medium as a primary income source while freelancing.

“I made $800 last month from Medium,” said one writer who will be demoted next month. “That’s my rent. Now I need to find 600 more followers in two weeks or I’m homeless.”

Google confirmed the March 2026 core algorithm update finished rolling out yesterday, and the results are… mixed. Some publishers saw 40% traffic increases. Others lost half their organic traffic overnight. Same update, wildly different outcomes.

What Changed

The update continues Google’s emphasis on “helpful content”—but with a twist. Sites showing clear expertise signals (author bios, credentials, transparent sourcing) are seeing disproportionate gains. Generic content farms continue their decline.

Google’s helpful content system now explicitly rewards:

Google search is getting worse. Not gradually. Dramatically.

I ran 500 searches across topics I know well. The results were shocking:

• Reddit threads ranking for medical queries
• AI-generated spam in top 3 positions
• 3-year-old articles dominating current events
• Affiliate sites masquerading as authoritative sources

Something is fundamentally broken. And Google can’t admit it.

The Decline By The Numbers

I compared Google results from 2020, 2023, and 2026:

Search Satisfaction (user surveys)

Remember when Google AI Overviews were supposed to destroy publishing? The doomsday predictions were everywhere:

• “No one will click through to websites anymore”
• “Publishers will lose 60% of search traffic”
• “The open web is dead”

That was March 2024. Two years later, the plot twist nobody expected: AI Overviews are failing, and publishers are quietly celebrating.

The Panic Was Real

When Google rolled out AI Overviews in the US, the SEO community lost its collective mind.

Everyone loves free software. But “free” comes with a price you don’t see until it’s too late.

Context

Open-source software powers the internet:

• Linux: 90% of cloud servers
• Apache: 40% of web servers
• PostgreSQL: Millions of databases
• Node.js: 50 million developers

All free. All maintained by volunteers.

The problem: Those volunteers are burning out.

Recent examples:

• Log4j vulnerability: Maintained by 3 unpaid volunteers
• Heartbleed: OpenSSL maintained by 1 person part-time
• XZ backdoor: Single maintainer for entire project

Critical infrastructure. Billion-dollar companies depend on it. Maintained by hobbyists.

Headline

Google’s March 2026 update wiped out thousands of sites overnight. But the sites that survived—and even thrived—have one thing in common. Here’s what actually works now.

Context

The March 2026 “Helpful Content Update” was the most aggressive algorithm change since the 2011 Panda update. Here’s what got hit hardest:

Losers:

• AI-generated content farms (finally)
• Sites with high ad density
• Thin content that just summarized other articles
• Sites with poor Core Web Vitals

Winners: