AI·May 2025·9 min read

Top E-commerce Trends in 2025: What Online Retailers Need to Know

A year-old trends post is a testable prediction, not evergreen content. Checked against primary sources instead of the original aggregator citations, three of the predicted shifts landed harder than advertised, two never showed up in any platform's own disclosed numbers, and the biggest 2025-2026 shift, AI-driven conversational shopping, wasn't on the original list at all.

Why a trends post needs a one-year audit

Retail trend posts get written and forgotten. Nobody goes back to check whether "2025 is the year of voice commerce" turned into an actual dollar figure anyone can point to.

This is that check. Each claim below is re-tested against the retailer, platform, or standards body that would actually know, not against a secondary aggregator repeating a number nobody re-verified.

The rule applied throughout: if a platform hasn't disclosed a number itself, the claim gets marked unconfirmed, not assumed true because a research firm published an estimate. That standard eliminates most of the flashiest predictions from the original post, which is itself informative about how those predictions were sourced the first time.

What held up, with the receipts

Mobile commerce: the direction was right, the original number is unverifiable

The original post cited a 2023 snapshot, global mobile commerce valued at $2.2 trillion and 60% of all e-commerce sales, sourced to a Shopify statistics roundup rather than a single measured figure. That page aggregates third-party numbers rather than reporting Shopify's own transaction data, which makes the 60% figure itself hard to re-verify against anything Shopify has published directly.

What can be re-verified: Adobe's own Digital Price Index for the 2025 US holiday season puts mobile at 56.4% of online transactions, up from 54.5% in 2024, the first full year mobile has cleared half of online spend (Adobe — 2025 holiday shopping season recap). That's a different metric, US transaction share, not global sales value, but it points the same direction the original claim did: mobile as the majority channel, not a plurality.

US retail e-commerce overall kept growing too: the Census Bureau's Q4 2025 estimate put e-commerce at 16.6% of total retail sales, up from 16.1% for all of 2024 (U.S. Census Bureau — Quarterly Retail E-Commerce Sales). Steady, single-digit-percentage growth, not the discontinuous jump some 2025 forecasts implied.

Generative AI in product development: real, with a corrected number

The original post's Walmart example said Trend-to-Product cut development time "from six months to six weeks." Walmart's own account is more specific: the tool compresses the timeline by as much as 18 weeks, with fashion items reaching shelves in six to eight weeks from trend identification, not a flat six weeks (Walmart Corporate — Technology). Small correction, same direction: this trend materialized in a named, first-party, still-running program.

Augmented reality try-on: shipped, and still expanding

Google's own shopping blog confirms this one directly. Its AI-powered virtual try-on tool, announced at I/O 2025, lets a shopper upload a photo and see billions of apparel listings rendered on their own body, using a custom image-generation model built to understand how different fabrics fold and drape (Google — Shopping in AI Mode, virtual try-on update). Unlike the voice and social commerce numbers, this is a shipped, named, first-party feature, not a projection.

The original post's framing was IKEA-style furniture placement and cosmetics try-on through brand-specific apps. What actually scaled was the opposite pattern: a single platform-level try-on tool spanning billions of listings across many retailers, rather than each retailer building and maintaining its own AR experience.

Resale and re-commerce: accelerated

ThredUp's 14th annual resale report, published in 2026, puts the global secondhand market at $393 billion, growing faster than new apparel, with the U.S. resale segment expanding roughly four times faster than retail overall (ThredUp — 14th Annual Resale Report). The original post's resale prediction held up better than most.

The trends that held up share a pattern: each one has a company willing to publish its own specific numbers. The trends that didn't hold up are the ones no platform has ever disclosed a real figure for.

What didn't materialize as predicted

Voice commerce: the $80 billion figure never got reconciled

The original post cited a third-party forecast of $80 billion in annual voice commerce sales by 2025. No platform, Amazon, Google, or otherwise, has published a figure confirming or denying that number, and the infrastructure side of voice commerce actually shrank in the interim.

Google shut down Conversational Actions, the framework third-party developers used to build custom Assistant shopping experiences, in June 2023, redirecting developer effort to Android-only App Actions instead (Google for Developers — Conversational Actions sunset overview). A platform retiring its third-party voice app framework is a strange thing to see in the middle of a supposed voice-commerce boom.

Social commerce: infrastructure is real, "mainstream" is unproven

Shopping features inside social apps are genuinely built out now. Meta's Commerce Platform gives brands checkout, catalog, and order management APIs directly against Instagram Shopping and Facebook Marketplace (Meta for Developers — Commerce Platform).

What's still missing is any platform disclosing what share of retail actually moves through these surfaces. The original post's "global social commerce sales projected to reach $1.2 trillion" claim traced back to a third-party estimate, and no platform's own reporting has confirmed a number anywhere near that scale.

Claims the original evidence couldn't actually support

Two more of the original predictions weren't wrong so much as unfalsifiable from the start, because the original post's own citation was already a secondary aggregator, not a source capable of confirming or denying anything a year later.

Sustainability as a purchase driver

The original post's sustainability claim linked to a general commerce-trends aggregator site, not a retailer disclosure or a standards body. A year later, there's still no first-party retail data source that isolates "sustainability" as a measurable driver of purchase decisions at scale, which means the claim was never really testable in the first place, not confirmed and not refuted.

Quick commerce, generalized past its actual evidence

The original post's quick-commerce examples, BigBasket and Blinkit, are India-specific grocery delivery platforms. Extending that regional pattern into a general "quick commerce meets instant gratification" global trend overreached what the cited examples actually supported, and no U.S. platform has published comparable instant-delivery adoption figures in the interim.

What the original post missed entirely

The single largest 2025-2026 shift in how people discover and buy products wasn't on the original trends list: AI-driven conversational shopping, delivered through chat and search interfaces, not voice devices.

Google's own announcement describes agentic checkout, where its AI completes a purchase on a shopper's behalf after the shopper sets preferences like size, color, and budget, alongside conversational product discovery inside AI Mode and the Gemini app (Google — Agentic checkout and AI shopping features). This is the automation the original "voice commerce" prediction was reaching for, arriving through a completely different interface than anyone forecast.

Trend-by-trend scorecard

2025 prediction	2026 status	Primary source
Mobile commerce dominance	Direction confirmed, original number unverifiable	Adobe Digital Price Index
GenAI in product development	Materialized, number corrected	Walmart Corporate
AR / virtual try-on	Shipped and expanding	Google Shopping blog
Resale market growth	Accelerated	ThredUp Resale Report
Social commerce "mainstream"	Infrastructure real, scale unproven	Meta Commerce Platform docs
Voice commerce, $80B by 2025	Unreconciled; platform investment retreated	Google Conversational Actions sunset
Sustainability as purchase driver	Unfalsifiable — original source was an aggregator	None available
Quick commerce, global framing	Overgeneralized from India-specific examples	None available outside India
AI-driven conversational shopping	Arrived, but wasn't predicted	Google agentic checkout announcement

What this means for reading the next trends post

A number without a named, dated, first-party source is a guess wearing a percentage sign.
A platform investing engineering effort (shipping a feature, publishing an API) is a stronger signal than a platform being mentioned favorably in a forecast.
A platform retiring infrastructure (like Google sunsetting Conversational Actions) is a stronger counter-signal than any forecast miss.
The biggest shift in any given year is often the one nobody predicted, because it didn't exist as a category yet.
A claim sourced to a secondary aggregator in the original post is usually unfalsifiable a year later too, since there was never a primary number to check against.
A regional example (a single country's grocery-delivery platforms, say) doesn't automatically generalize into a global trend just because a trends post frames it that way.

FAQ

Why did voice commerce underperform its 2025 forecast?

The forecast came from a third-party research firm, not a platform disclosure, and the platform actually pulled back its third-party voice developer framework in 2023, two years before the target year the forecast was made for (Google for Developers — Conversational Actions sunset overview).

Is mobile commerce still growing or has it plateaued?

Still growing. In the US, mobile crossed 50% of online holiday transactions for the first time in the 2025 season, up from 54.5% the prior year to 56.4%, per Adobe's own data (Adobe — 2025 holiday shopping season recap). The original post's global 60% figure isn't directly comparable, since it measured sales value worldwide, not US transaction share.

Did social commerce actually reach the scale originally predicted?

The infrastructure to support it is real and well-documented by Meta and other platforms. No platform has published transaction-volume figures that confirm the original trillion-dollar-scale projection.

What replaced voice commerce as the AI shopping interface?

Conversational and agentic shopping through text-based AI interfaces, chat and search, not smart-speaker voice commands. Google's agentic checkout feature is the clearest example shipped so far (Google — Agentic checkout and AI shopping features).

How should a retailer treat next year's trends list?

As a set of hypotheses to check against primary sources in twelve months, not a roadmap to execute against today. The trend worth acting on now is usually the one a platform has already shipped, not the one a research firm has forecast.

Was AR try-on actually a hype cycle like voice commerce?

No, this one shipped. Google's own I/O 2025 announcement and subsequent updates describe a working, expanding virtual try-on feature built on a dedicated fashion image-generation model, not a forecast (Google — Shopping in AI Mode, virtual try-on update).

Why call the sustainability trend "unfalsifiable" instead of just wrong?

Because the original claim was never attached to a measurable, disclosed figure from a retailer or standards body. Without a baseline number to check, there's nothing to confirm or refute a year later, which is itself the finding.

References

From the Destm engineering archive. For current work on this topic, start at Solutions or the blog.