Amazon Is Losing Its Grip on E-Commerce -- And Its Own Actions Prove It

400 sellers turned off their Amazon ads on the same day.

Combined, they generated over 2 billion dollars in annual revenue. Amazon knew the number. Amazon called their bluff anyway.

Then, 48 hours later, Amazon reversed course.

If you've been selling on Amazon for any length of time, you understand what makes that remarkable. Amazon doesn't blink. The platform that absorbed seller complaints for two decades without meaningful response reversed a significant policy decision in under two days.

That didn't happen because Amazon suddenly developed concern for seller cash flow. It happened because the coalition was large enough to threaten a real, measurable dent in their advertising revenue. That tells you something the platform has been working hard to conceal: Amazon needs sellers more than it lets on.

The Discovery Moat Amazon Is Losing

For 20 years, Amazon had the most powerful product discovery engine in retail. If someone wanted to buy something, they started on Amazon.

That was the moat. That was why sellers absorbed fee increases year after year.

Competed against Chinese sellers with structural cost advantages. Accepted policies that would have been unreasonable on any other platform.

The customers were there. You had to be there too.

That moat is cracking.

In 2022, 61% of all online product searches started on Amazon. By 2025, that figure had dropped to approximately 50%. Based on current trend data from Marketplace Pulse and eMarketer, it's declining further in 2026.

A drop from 61% to 50% over three years might look like normal market competition. The breakdown by age group makes clear this is structural, not cyclical.

56% of Gen Z now begin product searches on Google, YouTube, or AI tools. 12% go to TikTok first. This is the cohort that will be the dominant consumer group for the next 30 years.

They grew up watching creators recommend products on YouTube before they ever thought to search Amazon. They found things on TikTok before they knew they were in the market.

Amazon's search-and-filter model doesn't map to how this generation discovers what to buy.

And it gets worse: Amazon isn't just losing discovery.

Amazon Is Getting Cut Out of Both Ends of the Funnel

Product discovery is the top of the funnel -- someone realises they want a product and goes looking. Amazon is losing that. But they're simultaneously losing the research phase.

Someone who already knows they want a standing desk used to go to Amazon to figure out which one. Now they don't.

Product research now happens on YouTube, where real owners post long-form comparison videos. It happens on Reddit, where community members give unfiltered assessments.

Amazon's review system has been compromised by years of incentivised manipulation. Reddit hasn't.

It happens through AI tools. Someone asks ChatGPT what's the best robot vacuum under 300 pounds for pet hair. They get a synthesised recommendation without visiting any marketplace.

Being cut out of both discovery and research simultaneously is a qualitatively different problem than losing one phase. If Amazon only lost discovery, they could argue they remain the primary evaluation and conversion channel. Losing both means the platform's fundamental value as the place people go to shop is under structural pressure.

Amazon's own revenue data shows it. Advertising revenue grew 19% year-over-year in Q1 2025, reaching 17.2 billion dollars in a single quarter.

Retail sales grew 7%. The gap is not coincidence.

When organic visibility deteriorates, sellers pay more for paid visibility. Amazon's ad business benefits directly from the degradation of its own organic discovery system.

The platform is charging sellers more to compensate for its own declining utility. Until very recently, sellers had no real choice but to pay.

Four Moves That Show a Platform Acting Under Pressure

When a dominant company starts making reactive product decisions, it signals something structural has changed in its market position. Amazon's product roadmap over the past 18 months reads like a list of patches applied to a cracking foundation.

Blocking 47 AI bots from crawling Amazon was the first signal. In August 2024, Amazon added 47 AI systems to their block list.

OpenAI's web crawler. Anthropic's Claude scraper.

Perplexity's bot. Google's AI-specific crawlers.

The logic is transparent. If ChatGPT answers that question without the user visiting Amazon, Amazon loses the visit, the impression, the conversion, and the customer relationship.

Building a wall is not the same as improving what's behind it. Amazon made AI tools less able to scrape their listings. It did nothing to make Amazon a more appealing starting point for product discovery than those AI tools.

Buy For Me was the most revealing move. Launched February 2025. Amazon's AI bought products from external brand websites when Amazon didn't carry what customers searched for.

Brand owners began receiving orders from a buyforme.amazon.com email address. For products they had never listed on Amazon. Products they had never agreed to sell through the platform.

The same company that blocked 47 external AI systems from crawling Amazon sent its own AI to crawl everyone else without asking. After significant backlash from brand owners, Amazon added an opt-out mechanism.

Amazon Shop Direct told the clearest story. Launched in early 2025, it allows merchants to sell through Amazon's customer interface without listing products on Amazon.

Merchants keep their own inventory, manage their own fulfilment, control their own pricing. Amazon takes a referral cut.

At launch, the programme covered over 100 million products across 400,000 merchants. Read plainly: Amazon is acknowledging they cannot win on product selection through the traditional marketplace model. Rather than competing on inventory breadth, they want to be the traffic intermediary sitting on top of everyone else's inventory.

That's a retreat from the proposition that made Amazon the default starting point for product search.

Rufus was marketed as a solution but functions as a problem. Amazon's AI shopping assistant handles 274 million daily queries.

That's 13.7% of all Amazon searches by late 2025. But 83% of Rufus suggestions direct users to Amazon-branded or Amazon-sold products.

Amazon Basics appears in 41% of results regardless of category or quality ranking.

When Amazon tested expanding Rufus as a popup interface, the backlash was bad enough that multiple technology publications covered the complaints.

Rufus is not a discovery tool. It's a promotional mechanism for Amazon's own product lines, presented with the language of a neutral recommendation engine.

Customers have noticed. The button to access Rufus remains small and tucked into a corner of the app.

That tells you more than any press release.

The Seller Numbers Tell the Real Story

New seller registrations on Amazon fell 44% in 2025. The lowest annual total in a decade and the lowest figure since Marketplace Pulse began systematic tracking in 2015.

American sellers dropped from 26% of new registrations to 16%. Chinese sellers now account for nearly 60% of new signups.

Active sellers fell from 2.4 million in 2021 to 1.65 million by end of 2025. That's a 31% decline over four years.

These numbers don't tell a story of Amazon failing. Amazon's total revenue, its cloud business, and its advertising business are all growing. But the marketplace — the part that depends on independent sellers building sustainable businesses — is contracting.

The sellers who remain are squeezed harder each year. The platform needs advertising growth to compensate for declining organic reach.

What April 2025 Actually Proved

The boycott worked because of a specific, quantifiable threat. 400 sellers doing 2 billion dollars in annual revenue represented a meaningful fraction of Amazon's advertising revenue. Turning off that spend simultaneously created a number Amazon's finance team couldn't ignore.

The lesson isn't that collective action works in principle. The lesson is this: data gives you a position. A position is the only thing that moves this platform.

The sellers who led that boycott knew their numbers. They knew what their combined ad spend meant to Amazon's quarterly reporting. They understood the asymmetry well enough to calculate that the threat was credible.

Most sellers don't have that. Most sellers couldn't tell you what percentage of their Amazon revenue is genuinely incremental.

Driven by advertising, not organic. Versus what Amazon would have sold without paid support.

They spend on ads because stopping feels risky, not because they've modelled the actual contribution.

The sellers protected from Amazon's next policy change are the ones who own their numbers. They know their true blended margin.

They have channels operating independently of Amazon's decisions. They can calculate the impact of a policy change in hours, not find out weeks later in their payout.

That's not luck. That's data. And it's available to anyone willing to stop depending on Amazon's dashboard to tell them how they're doing.

What to Do With This

Amazon is still the largest e-commerce marketplace in most Western markets. It will be a viable, important channel for years. The argument here is not to leave it.

The terms of the deal have changed. They now favour sellers who treat Amazon as one channel among several.

The sellers who treat it as their entire business are the ones absorbing every policy change with nowhere to go. The organic reach advantage that justified platform dependency is smaller than it was.

The fee structure is higher. The competitive pressure from Chinese sellers is more intense.

And the consequences of any of those changes are more severe for sellers who have nowhere else to go.

The sellers who came through 2025 strongest had two things. Meaningful revenue on channels outside Amazon. A data view that showed their whole business, not just the version Amazon chose to show.

Building both, starting now, is the most important decision most Amazon sellers can make. Frankly, it's also the most overdue one.

How to build a data strategy that doesn't depend on any single platform

See how Alpomi unifies your cross-channel data and book a demo

Frequently Asked Questions

Why did Amazon sellers boycott in April 2025?

Amazon rolled out three policy changes in rapid succession in early 2025. A 7-day payout delay after delivery confirmation.

A ban on credit cards for advertising spend. A new 3.5% fuel and surcharge fee.

A coalition of 400 sellers generating a combined 2 billion dollars in annual revenue coordinated a simultaneous ad shutoff. Amazon reversed the ad payment change within 48 hours.

Is Amazon losing e-commerce market share?

Amazon's share of online shopping journey starts dropped from 61% in 2022 to approximately 50% in 2025. The decline is concentrated among younger consumers. 56% of Gen Z now start product searches on Google, YouTube, or AI tools.

Amazon's total revenue continues to grow. But the marketplace's role as the primary product discovery channel is declining structurally.

What is the Amazon Rufus AI assistant?

Rufus is Amazon's AI shopping assistant integrated into the Amazon app, handling over 274 million daily queries by late 2025. Independent analysis found 83% of Rufus recommendations direct users to Amazon-branded or Amazon-sold products. Critics describe it as a promotional tool wearing a discovery costume.

What was Amazon's Buy For Me feature?

Amazon's Buy For Me let Amazon's AI purchase products from external brand websites on behalf of users. For items Amazon didn't carry.

Brand owners received orders from a buyforme.amazon.com address. For products they had never listed on Amazon.

Products they had never consented to sell through the platform. Amazon added an opt-out mechanism after significant backlash.

Why are fewer sellers registering on Amazon?

New seller registrations fell 44% in 2025, to the lowest level since 2015. Rising fees played a role.

So did growing competition from Chinese sellers, who now represent nearly 60% of new registrations. And the simple fact that alternatives — Shopify, TikTok Shop, Walmart Marketplace — have matured into real options.

Should I diversify away from Amazon?

The evidence supports building meaningful revenue outside Amazon. Not because Amazon is failing.

Because single-platform dependency creates business risk that's more acute than at any point in the past decade. The goal for most sellers is channel diversification.

Shopify, TikTok Shop, other channels — built to the point where Amazon represents a significant but not existential share of revenue.