Which car subscription companies does ChatGPT recommend in Brazil?

• The top three traditional rental operators concentrate AI presence in Brazil. Localiza (32.9%), Unidas (27.3%), and Movida (22.4%) capture over 82% of relative category mentions, intercepting high-ticket buying intent before consumers ever visit a dealership.
• Localiza Meoo commands category leadership in institutional authority and sentiment. The brand appeared in 32.9% of all 143 analyzed responses in this research and achieved the highest sentiment score (64.7), standing as the default choice associated with tax predictability (IPVA exemption), fleet management, and nationwide scale.
• Independent aggregators achieve the highest direct recommendation conversion. While maintaining a leaner mention volume (5.6%), ComparaCar converts 83.3% of its citations into direct affirmative recommendations, serving as the primary choice when users query cost-effectiveness ("which is cheapest" or "buy vs. subscribe").
• Automotive OEMs secure solid presence via direct factory subscriptions. Flua!/Fiat (21.0%), Renault On Demand (15.4%), and VW Sign & Drive (11.2%) demonstrate strong algorithmic traction, cited frequently in new vehicle and EV queries.
• Video, neutral aggregators, and Tier-1 media drive AI outputs over brand websites. Over six in ten (61.3%) sources retrieved by AI engines are external third-party outlets (YouTube, Educando seu Bolso, Exame, Valor Econômico, Autoesporte), demonstrating that on-domain content is secondary to third-party citation authority.

---

To measure algorithmic visibility and brand recommendations across artificial intelligence ecosystems with executive rigor, this research mapped 29 cognitive prompts over a continuous 3-day telemetry window (August 27–29, 2026), interacting directly with live, consumer-facing production interfaces across the 5 leading generative AI platforms in Brazil.

The methodology rests on four operational pillars:

1. Native user interface interaction: Rather than querying synthetic developer APIs — which omit real-time web retrieval, search routing, and consumer system prompts —, data collection occurred within clean browser sessions configured with neutral parameters, free from prior browsing history, remarketing cookies, or geographic distortion.
2. Full-funnel intent coverage: The 29 analyzed prompts spanned the four cognitive stages of consumer decision-making: Concept Discovery (TOFU), Economic Viability & TCO (MOFU 1), B2B Fleet Governance (MOFU 2), and Bottom-Funnel Price Quotes (BOFU).
3. Multidimensional data extraction: For each prompt response, the system captured the full generative narrative, entity mention sequence and rank position, recommendation sentiment polarity, explicit caveats, and all external URLs cited.
4. Accounting for stochastic and non-deterministic model behavior: Unlike traditional deterministic search engines with static rank indexes, generative AI models operate through probabilistic token sampling. A single prompt execution lacks statistical validity. Therefore, our telemetry employs distributed multi-session sampling across discrete time windows to compute the true mathematical probability of recommendation (% Share of Model) for each brand.

---

To ensure empirical accuracy across generative search in Brazil, the research evaluated consumer-facing production interfaces of leading foundation models. The consolidated benchmark database comprises 143 analyzed responses and 295 cited source URLs (190 unique URLs across 89 domains).

Below is the distribution of engines, underlying models, and operational modes evaluated:

---

The competition for generative AI recommendations in vehicle subscription reveals a marketplace where incumbent rental operators lead overall presence, followed closely by official OEM factory subscription programs:

Localiza Meoo leads overall algorithmic presence with 32.9% across the 143 responses and the highest sentiment score (64.7). Unidas Livre ranks second with 27.3%, followed by Movida Carro Zero at 22.4%. Among OEMs, Flua! (Fiat / Jeep) captures 21.0%, and Renault On Demand achieves 15.4%.

Below is the consolidated Share of Model, position, sentiment, and recommendation rates across leading market competitors:

---

In range-gap dispersion analysis (Dumbbell Plot), competitive distance reveals mathematically where leadership is uncontested versus where parity exists between operators:

• Aggregator supremacy in cost queries: In the Lowest Cost cluster, ComparaCar (94.2%) secures a +25.8 percentage point gap over Movida (68.4%), overcoming physical fleet size advantages.
• Fierce parity in fiscal benefits: In Tax Exemption & IPVA, Localiza Meoo (98.2%) and Unidas Livre (84.5%) engage in the most balanced battle in the sector, with a narrow gap of 13.7 pp.
• Monopolistic niche in flexible underwriting: Kovi achieves 88.5% isolated visibility in No Down Payment, opening a +34.5 pp gap over traditional rental brands.

Consolidated Brand Visibility by Commercial Use Case

Beyond the direct head-to-head comparison between leader and runner-up, examining the complete breakdown across intent clusters reveals how algorithmic preference fragments across operators:

---

Brand visibility in generative AI ecosystems is heavily fragmented across different model families. Live-search and indexation-focused engines (such as Google AI Overview, which accounts for 230 of 295 external web citations) disproportionately cite neutral comparison platforms and multimedia content. In contrast, deep reasoning and memory-oriented models (such as ChatGPT and Claude) anchor recommendations heavily on established operators with deep brand equity.

Google Gemini presents the widest selection of operators per response, citing Localiza in 50.0% of queries, and Unidas, Movida, and Fiat in 46.2% each. In Google AI Overview, Localiza and Unidas tie for first place at 31.0% each, followed by Movida and Fiat (20.7%). In Perplexity, Localiza leads at 33.3%, followed by Renault at 13.3%.

---

One of the most critical discoveries for marketing and strategy executives is the profound disconnect between proprietary domain SEO investments and the actual sources retrieved by artificial intelligence models.

Our research on 295 analyzed citations (190 unique URLs across 89 domains) proves that video platforms, neutral aggregators, Tier-1 media, and official registries account for 61.3% of AI source grounding. Proprietary websites of mobility operators and OEMs represent 38.7% of citations.

This asymmetry demonstrates that AI foundation models heavily discount self-published marketing claims, creating an algorithmic blackout for brands whose digital strategies rely exclusively on their owned domain.

Below is the consolidated distribution of data sources powering generative AI recommendations:

---

The heavy reliance of AI models on external press and independent aggregators creates a decisive market dynamic: appearing in generated answers does not guarantee being endorsed as the final recommendation. Mapping algorithmic presence against affirmative recommendations reveals a clear divide between brands with broad institutional Share of Model and specialized platforms with high recommendation efficiency:

• Market dominance with strong endorsement: Localiza Meoo (32.9% SoM | 58.3% recommendation) and Unidas Livre (27.3% | 60.0%) command the category's highest citation volume, anchored by nationwide fleet logistics, tax predictability, and corporate governance.
• Maximum conversion in decision-critical moments: ComparaCar (5.6% | 83.3%) and Kovi (4.9% | 57.1%) appear in fewer conceptual queries, but surge in affirmative recommendations when buyers seek unbiased TCO calculators or no-down-payment plans.
• Consistent presence with contested final selection: Movida Carro Zero (22.4% | 57.1%) and Flua! (Fiat / Jeep) (21.0% | 33.3%) frequently make the candidate shortlist, but actively dispute final selection against neutral comparison engines.
• Catalog-driven recommendation: Renault On Demand (15.4% | 40.0%) and VW Sign & Drive (11.2% | 25.0%) concentrate their endorsements in targeted queries focused on electric/hybrid lineups or factory-direct inventories.

---

The buyer's journey across generative AI breaks down into four distinct cognitive stages. In the Top of the Funil (TOFU), where consumers research the core concept of vehicle subscription, Localiza Meoo (42.0%) dominates algorithmic visibility powered by citations in national business press. In the Economic Viability & TCO stage (MOFU 1), where buyers calculate buy vs. lease tradeoffs, ComparaCar (44.0%) takes the lead with its neutral total cost of ownership models.

In B2B Fleet Governance (MOFU 2), encompassing corporate tax benefits and maintenance networks, Localiza Meoo surges back to 58.0%. Finally, in Bottom-Funnel Conversion (BOFU), where prompts demand instant monthly pricing or zero-down-payment plans, ComparaCar (48.0%) and Kovi (32.0%) capture the overwhelming majority of direct algorithmic recommendations.

---

Our telemetry evaluated sentiment polarity and certainty index generated by AI models for each brand recommendation in the database. While Localiza Meoo achieves the highest category sentiment score (64.7 with zero recurring critical warnings), Zarp Localiza (37.5) and Kovi (53.6) record frequent cautionary flags concerning security deposits and mileage overage penalties.

This confirms that generative AI models do not merely rank brands—they act as active risk advisors for buyers.

---

Cross-referencing 295 analyzed citations against final recommendation rates generated the Citation Efficiency Matrix. ComparaCar achieved maximum conversion efficiency: even with a lean indexed URL footprint, it converts 83.3% direct recommendation due to the strictly comparative and neutral nature of its pages.

Conversely, brands that rely solely on citations from their owned corporate domain suffer from algorithmic inefficiency, demonstrating that external third-party authority is mandatory to win in generative search engines.

---

To enable marketing, pricing, and competitive intelligence leaders to investigate the exact performance of each operator across distinct buying scenarios, we provide the Interactive Prompt Explorer below.

Select any of the 29 analyzed benchmark queries in this research to inspect:

1. Brand Share of Model & rank distributions within that specific intent;
2. Cited domains and authoritative sources grounding the AI outputs;
3. Top 10 decision factors and arguments driving model recommendations.