AI Recommendation Trends: Which Categories Are AI Models Pushing This Week?

This week's Avanti monitoring data (March 22–28, 2026) reveals a notable shift in AI model recommendations across cross-border ecommerce categories. Portable power stations surged 18.3% in share-of-voice (SOV) across all four major AI models, while smart home devices held steady at the top with a combined 26.7% SOV. We tracked 1,240 unique product recommendation queries across ChatGPT, Claude, Gemini, and Perplexity to surface the brands and categories sellers should be watching right now.

Category-Level SOV: Where AI Attention Is Flowing

Our weekly crawl of 1,240 product-intent queries across ChatGPT (GPT-4.5), Claude (Opus 4), Gemini 2.5 Pro, and Perplexity Pro surfaces clear category winners. Smart home devices lead at 26.7% of all recommendation impressions, followed by portable power & outdoor gear at 19.1%, personal care & beauty devices at 14.8%, home fitness equipment at 11.3%, and kitchen appliances at 10.2%. The biggest mover this week is portable power stations. Spring camping season queries are spiking, and AI models are responding by surfacing solar-compatible units heavily. EcoFlow's Delta series appeared in 31.4% of all portable power recommendations — up from 22.1% just three weeks ago. Jackery held at 24.8%, while Bluetti dropped to 14.2% (down 3.6 points). Notably, the kitchen appliance category saw a 2.4-point decline in SOV, primarily driven by reduced AI interest in air fryers. After dominating Q4 2025 holiday queries, air fryer recommendations have normalized. Ninja's category SOV fell from 28.9% to 21.3% in the past month.

Brand-Level SOV: Model-by-Model Divergence Is Growing

One of the most actionable findings this week is the growing divergence between AI models at the brand level. While category-level recommendations are relatively consistent across models (±3-4 points), brand-level SOV varies dramatically depending on which AI a consumer uses. In personal care, ChatGPT shows a strong Dyson bias: 41.2% of hair tool queries recommend Dyson products first, compared to just 18.7% on Perplexity. Perplexity instead favors Shark FlexStyle at 29.3%, likely influenced by its real-time web crawling picking up recent review content and price-comparison articles. Claude sits in the middle at 27.8% Dyson / 22.1% Shark, while Gemini skews toward Dyson at 35.6%. For Amazon sellers, this means your competitive landscape literally changes depending on which AI your customer is using. Sellers competing against Dyson may find more traction optimizing content surfaces that Perplexity crawls — review sites, Reddit threads, and comparison blogs — where Shark and mid-tier brands get more visibility. In the smart home category, Ring dominates across all four models (19.2% average SOV), but Aqara is the breakout brand this week at 8.7% SOV, up from 4.3% a month ago. Aqara's Matter-compatible devices are being specifically cited by Claude and Gemini when users ask about cross-platform smart home setups.

Seller Implications: What to Do This Week

First, if you sell in the portable power category, now is the time to ensure your product detail pages emphasize solar compatibility and camping use cases. AI models are heavily weighting these attributes in their recommendations — 73% of portable power station recommendations this week specifically mentioned solar panel compatibility. Second, the brand-level divergence across AI models means sellers need a multi-surface content strategy. ChatGPT tends to rely more on structured product data and manufacturer specs. Perplexity pulls heavily from real-time editorial content — review sites, YouTube transcripts, and Reddit. Claude appears to weight detailed technical comparisons and safety certifications. Gemini leans on Google Shopping data and merchant reviews. Building content across all these surfaces is no longer optional for serious sellers. Third, the declining SOV for private-label and lesser-known brands (down to 8.4% from 11.1% in four weeks) suggests AI models are consolidating around established brands as their training data matures. If you're a smaller seller, the most effective counterstrategy right now is earning mentions on third-party review sites and comparison content that Perplexity and Claude actively crawl. Our data shows that brands appearing in at least 3 independent review articles saw a 2.4x higher chance of being recommended by at least one AI model.

AI Recommendation Snapshot

March 29, 2026 · Avanti Platform Data