AI and Product Search: Q4 2025 Data

How AI Assistants Changed (Q4 2025)

The fourth quarter marked the integration of the first ad-based monetization models and new access tiers. This transition appears driven by the infrastructural and computational costs required to sustain these models, necessitating a search for economic sustainability beyond mere user base growth.

• ChatGPT and the “Go” Plan: OpenAI will introduce the first advertisements for free-tier users and launch ‘ChatGPT Go,’ a mid-level tier that includes in-conversation ads. These will appear as sponsored links or related suggestions during product searches.
• Google AI Overviews: There has been an increased presence of shopping ads directly within AI-generated overviews. Google integrated Merchant Center feeds to show real-time products and prices in response to transactional queries.
• Perplexity: The platform consolidated “Sponsored Follow-up Questions,” allowing brands to suggest deeper inquiries related to their products.
• Agentic AI: In Q4, execution capabilities (e.g., autonomous navigation) became more stable, allowing assistants to handle price comparisons and availability monitoring more granularly during the Black Friday period.

These developments outline AI’s role as an operative assistant rather than just an informative one, integrating commercial dynamics aimed at overseeing the user journey from search to transaction.

E-commerce Industry Impact (Q4 2025)

Data from the fourth quarter of 2025 highlights a recovery in referral session volumes from AI tools, with a 43.0% increase compared to Q3. This figure shows an acceleration relative to the slowdown observed in the previous quarter (+23.5%), primarily driven by seasonality linked to Black Friday and holiday shopping.

Despite the volume increase, metrics suggest a consolidation of user behavior, utilizing AI as a preliminary filter:

• Engagement: The Bounce Rate further improved, dropping to 28.5% (-4.1% vs. Q3). Parallelly, Average Session Duration rose to 52 seconds (+6.5%). These indicators describe traffic that, while originating from an external interface that tends to resolve queries “in-app” (zero-click), maintains high relevance when it chooses to click through to the e-commerce site.
• Conversion: The Add to Cart Rate (ATC) stood at 5.1%, recording a -10.0% decline compared to the peak of 5.7% reached in Q3.

Note: The contraction in the ATC rate is a recurring phenomenon in the final quarter of the year. Comparison activity typical of sales and gift-hunting tends to generate more exploratory session volumes. Users utilize AI assistants to compare prices and availability across multiple sites, thus diluting the immediate conversion rate compared to non-promotional periods.

Top AI Referral Sources (Q4 2025)

The analysis of traffic sources in the fourth quarter of 2025 highlights a return to the main player. ChatGPT increased its share of referral sessions, rising from 85% in Q3 to the current 89%. During the year-end shopping peak, OpenAI’s platform confirmed itself as the primary entry point for users utilizing AI to land on e-commerce sites.

Among other tools, stabilization or slight contractions were observed:

• Perplexity maintains its share at 3% of sessions.
• Copilot and Gemini both stand at 1% of referral traffic.
• The “Other AI referrals” category, which had risen to 9% in the previous quarter, dropped to 5%, suggesting that during the holiday period, users preferred to rely on better-known and consolidated assistants.

Q4 Revenue Analysis

The distribution of revenue (expressed as a percentage of total revenue generated by AI channels) shows significant shifts in the conversion capacity of different platforms.

While ChatGPT continues to generate most of the economic value, its share of total revenue dropped to 83.0% (compared to 88.4% in Q3). Despite the increase in traffic volume, a partial redistribution of value toward other channels is noted.

The most evident anomaly of the fourth quarter is Copilot: despite generating only 1.2% of sessions, it contributed 11.2% of total revenue. This figure indicates high economic efficiency, suggesting that referral links clicked via the Microsoft assistant were associated with high average order values or users with a higher-than-average purchase intent.

As for the other players:

• Perplexity generated 2.9% of revenue, a share in line with its traffic volume (3%).
• Gemini stands at 1.0% of revenue, maintaining a balanced ratio between sessions and conversions.
• The “Other AI referrals” category saw its economic impact decline to 1.8% (down from 5.7% in Q3).
• Tools such as Claude are beginning to show early signs of monetization, although they remain at marginal levels (0.1%), while models like Mistral did not record any direct revenue.

Industries Most Affected by AI Traffic (Q4 2025)

The distribution by sector in Q4 appears balanced between Retail & Shops (21.8%), Luxury (21.7%), and Sports (19.8%). Conversely, the contraction of referrals for the Hard Goods (electronics) sector continues, falling to 7.2% from the initial 34.6% at the start of the year.

This dynamic may reflect a trend toward the “Zero-Click” model: searches related to consumer electronics, being based on technical data, might be resolved directly by AI assistants, limiting external clicks to the final purchase phase. In contrast, categories such as Luxury, Sports, and Fashion maintain higher referral volumes, likely due to the user’s need for a deeper visual and brand-site experience.

Traffic Distribution Detail (Q4 2025):
Retail & Shops: 21.8% (Increasing)
Luxury: 21.7% (Increasing)
Sports: 19.8% (Increasing)
Fashion: 14.0% (Decreasing)
Beauty: 9.0% (Stable)
Hard Goods: 7.2% (Decreasing)
Home & Design: 5.9% (Decreasing)
Food: 0.3% (Decreasing)
Kids: 0.2% (Stable)

Annual Comparison: 2024 vs. 2025

The year-over-year (YoY) comparison highlights total growth compared to 2024, showing the increasing impact of artificial intelligence on e-commerce acquisition dynamics.

What the Data Tells Us

Year-end data outlines a path where AI, while remaining a numerically secondary traffic source, seems to be integrating more steadily into search habits. Although referral volumes remain low compared to traditional channels, the growth observed in 2025 indicates that artificial intelligence is now a recurring element in the acquisition mix.

The integration of advertising dynamics and the development of operative functions (such as assisted navigation) could further alter the user experience. The “zero-click” trend in certain sectors suggests that AI is beginning to manage the informative component directly in chat, limiting site visits to moments closer to the final choice or brand-specific exploration.

In this scenario, AI tends to orient sessions with a more defined intent toward e-commerce, acting as a natural filter between the research phase and the decision phase. For retailers, the challenge for 2026 lies in monitoring this evolving ecosystem and adapting their presence to a purchase journey that appears increasingly mediated and synthesized by virtual assistants.