How AI Search Is Recommending Online Car Buying Sites: Monthly Trends

Executive Summary

CarMax holds the category lead in October 2026 with 88.3% valid recommendation coverage, a 2.3-point edge over CarGurus at 86.0% and 1.8 points ahead of Carvana at 86.5%. The lead is stable in the sense that matters most to competitive positioning: no brand posted a significant upward move this month, and the leader held its position despite giving back ground across the full series.

The sharpest single-month gains came from the declining group clawing back ground, not from a new challenger. Autotrader rose 3.4 points from 82.4% in September 2026, Cars.com rose 1.6 points to 80.0%, Carvana rose 1.1 points to 86.5%, and CarMax rose 2.3 points to 88.3%. None of those single-month gains cleared the normal variation threshold, so the category read for October is one of partial recovery inside a longer downward drift rather than a new breakout.

Edmunds remains the category's most persistent structural decliner, falling 14.4 points from 75.0% in July 2026 to 60.6% in October 2026 across a three-month streak. TrueCar fell 9.0 points across the same span to 37.1%, also on a three-month streak. Lithia Motors / Driveway fell 5.3 points to 10.8%, and Autotrader fell a significant 3.5 points to 85.8%.

Against the prior month, the picture is different from the baseline comparison: four of the seven stable brands moved upward in October — CarGurus, CarMax, Cars.com, and Carvana — while Autotrader, one of the month's significant baseline-to-current decliners, posted the largest single prior-month increase in the category at 3.4 points. Edmunds added a further 1.6-point decline and TrueCar added 2.0 points of decline. The category's baseline-to-current read is "significant decliners and stable names," but the prior-to-current read is "partial stabilization with continued erosion at the bottom of the mid-tier."

Research scope: Each monthly run begins with 800 prompt-surface observations across the benchmark's defined AI/search surface universe (516 unique questions in July 2026, 555 in August 2026, 576 in September 2026, and 575 in October 2026). All 800 mentioned a tracked brand or competitor in each month. Of those, 794 were relevant and 6 irrelevant in July 2026; 781 were relevant and 19 irrelevant in August 2026; 793 were relevant and 7 irrelevant in September 2026; and 796 were relevant and 4 irrelevant in October 2026. The public metrics use the 731 qualified observations for July 2026, 726 for August 2026, 733 for September 2026, and 738 for October 2026 that survive both qualification stages.

Key Findings

AI Response Inconsistency Alerts

Questions This Section Answers

• Where did AI platforms contradict each other about return policies and inventory?
• Which brands and questions produced the highest-severity factual conflicts?

Six critical or high-severity factual inconsistencies were detected across four AI platforms in the current measurement window. The conflicts concentrate in two companies (CarMax and Carvana) and involve contradictions in how AI systems describe return policies and new-car inventory.

Carvana

Google AI Mode and Perplexity provided directly conflicting information about whether Carvana sells new cars. When asked "Which is the most reliable online car buying platform for new cars?", Google AI Mode stated that "Carvana primarily focuses on used and certified pre-owned inventory, not new cars," citing the Edmunds new-car landing page. Perplexity, answering the same question, described Carvana as one of the most reliable online platforms for new-car purchases, citing Forbes Advisor, CompareTheCarrier, and ConsumerAffairs. The conflict was rated critical with 0.95 confidence. The flagged sources on the Google AI Mode side referenced used-car platform roundups, which supports the positioning that Carvana's inventory is used and certified pre-owned.

CarMax

Five high-severity inconsistencies were detected around CarMax's return window, spanning Copilot, Gemini, and Google AI Mode. The core contradiction is between a 10-day and a 30-day return policy, and it appears across multiple question framings.

Return window (10 days vs 30 days, Copilot vs Gemini). When asked "What's the best website to look at used cars?", Copilot stated a 10-day return window while Gemini stated a 30-day return policy. Copilot cited RoadLancer, Sakaem Logistics, and SmartUpWorld. Gemini cited Monarch TG, Vehicle Transport Hub, and Santander Consumer USA. The flagged sources split on both sides: SmartUpWorld listed "10 days," while Monarch TG listed "30 days."

Return window (10 days vs 30 days, Gemini vs Copilot). When asked "Does Carvana have competition?", Gemini stated CarMax offers a 10-day money-back guarantee, while Copilot stated CarMax has a 30-day return policy. Gemini cited FinMasters, Monarch TG, and Forbes; Copilot cited Yahoo Finance, Sakaem Logistics, and Monarch TG. Monarch TG appears on both sides of this conflict.

Return window (10 days vs 30 days, Google AI Mode vs Gemini). When asked "Are there other apps like Carvana?", Google AI Mode stated a 10-day money-back guarantee while Gemini stated a 30-day money-back guarantee. The Gemini side cited an Indy Auto Man dealership page that referenced a 30-day money-back guarantee as part of a warranty package, suggesting the platform may have conflated warranty and return coverage.

Return window (10 days vs 30 days, Copilot vs Gemini). When asked "What are good used car websites?", Copilot stated CarMax now offers only a 10-day return window, explicitly noting a reduction from 30 days, while Gemini stated CarMax has an excellent 30-day return policy. Crediful, cited by Copilot, specifically noted that "CarMax reduced its money-back guarantee from 30 days to 10 days." Indy Auto Man and Monarch TG, on the Gemini side, both listed 30 days.

Return window (10 days vs 30 days, Gemini vs Copilot). When asked "Where is the best place to shop for a used car?", Gemini stated the money-back return window is usually 24 hours to 10 days, while Copilot stated a 30-day money-back guarantee. WhichCar, cited by Gemini, described the return window as measured "in days rather than the moment you drive off the lot," without a specific number.

The pattern across all five CarMax conflicts is that AI platforms are drawing on promotional and review sources that disagree about the current return policy. The public benchmark cannot resolve which value is accurate; it can only record that the disagreement is recurring and appears on multiple surfaces.

Benchmark Context

Questions This Section Answers

• How many prompts survive the funnel from collection to qualified observations?
• What separates the raw 800-prompt collection universe from the qualified benchmark set?

The report separates the raw collection universe from the qualified analysis set. Brand-level recommendation percentages are calculated within the qualified benchmark set.

August 2026 and September 2026 sat between these two months in the series, with 800 source prompts each, 555 and 576 unique questions respectively, 781 and 793 relevant prompts, 19 and 7 irrelevant prompts, and 726 and 733 qualified observations. With the qualification funnel established, the following metrics summarize category-wide performance.

Benchmark-Level Metrics

September 2026 is the intermediate month worth noting for interpretation: the recommendation-shaped answer share bottomed at 54.0% in September before recovering to 77.8% in October, while the valid recommendation shortlist share dropped to 85.1% in September before returning to 91.3% in October. The response mix in October shifted heavily toward recommendation shortlists (574 of 738 observations), up from 372 in September and 450 in July 2026.

AI Recommendation Trend

Questions This Section Answers

• Which brands lead AI recommendation coverage in October 2026?
• Where did the top of the category cluster, and which brands fell away from the leaders?

The Category Is Stable at the Top and Eroding in the Mid-Tier

October 2026 does not show a leadership change or a new breakout, but it does show the mid-tier thinning out. CarMax leads at 88.3%, with Carvana, CarGurus, and Autotrader clustered between 85.8% and 86.5%, within 0.7 points of each other. The gap to the 5th-place brand, Cars.com at 80.0%, is 5.8 points. Edmunds and TrueCar continue to fall away from the top group on multi-month streaks, and Lithia Motors / Driveway shows a baseline-to-current decline driven primarily by the August tracking change.

No brand exceeded normal month-to-month variation in October on the upside, and the baseline-to-current declines at the leadership level were largely realized in August and September. The category-level baseline-to-current change is the combination of one large structural decliner (Edmunds), one moderate structural decliner (TrueCar), and a set of modest single-digit declines across the top four that have partially recovered in October.

What Changed This Month

Questions This Section Answers

• Did CarMax hold its category lead in October 2026?
• Which brands posted significant baseline-to-current declines this month?
• Why is Autotrader's baseline decline different from its September-to-October movement?

CarMax Holds the Lead With a Modest October Recovery

CarMax held 88.3% valid recommendation coverage in October 2026, down 2.3 points from 90.6% in July 2026 but up 2.3 points from 86.0% in September 2026. The baseline-to-current movement is not significant, and the prior-to-current movement is not significant either, making October a stable month for the category leader.

Raw mention presence was 97.7% with 721 of 738 observations mentioning the brand, essentially flat against 97.5% in July 2026. Top-3 placement rose 3.6 points from 27.4% to 31.0%, with 229 top-3 placements. Rank-one placement was 7.0% with 52 rank-one counts, up marginally from 6.8% in July.

The distinction to notice is that CarMax's presence has been near-universal across the entire four-month series, but its conversion to top-3 and rank-one placement has not kept pace with that presence. Its rank-one rate of 7.0% remains lower than CarGurus (26.8%), Carvana (14.6%), and Autotrader (13.0%), meaning CarMax is more often named somewhere in the answer than named first.

Highest-priority diagnostic: Which high-intent prompts convert CarMax's near-universal visibility into rank-one recommendations, and which competitors take the first position when CarMax does not lead?

Edmunds Continues a Three-Month Structural Decline

Edmunds fell 14.4 points in valid recommendation coverage from 75.0% in July 2026 to 60.6% in October 2026, a significant decline spanning three consecutive monthly drops. The October decline of 1.6 points from September was not itself significant, but it extends the streak.

Raw mention presence fell 8.0 points from 82.8% to 74.8%, with 552 of 738 observations mentioning the brand. The brand recorded 447 valid recommendations in October, down from 548 in July. Top-3 placement fell 3.0 points from 10.3% to 7.3%, and rank-one placement fell 1.4 points from 2.6% to 1.2%, with only 9 rank-one counts.

The distinction to notice is that Edmunds is losing presence and placement simultaneously, which is a compound erosion pattern rather than a stage-specific weakness. It appears in fewer answers, and when it appears, it lands lower.

Highest-priority diagnostic: Which prompts and AI surfaces account for the 8.0-point drop in Edmunds' presence, and which brands are taking recommendation slots in the answers where Edmunds no longer appears?

TrueCar Continues a Steady Three-Month Decline

TrueCar fell 9.0 points in valid recommendation coverage from 46.1% in July 2026 to 37.1% in October 2026, a significant three-month decline. The October decline of 2.0 points from September was not itself significant, indicating continued steady erosion rather than an abrupt drop.

Raw mention presence fell 7.1 points from 62.1% to 55.0%, with 406 of 738 observations mentioning the brand in October. Valid recommendation count fell from 337 in July to 274 in October. Top-3 placement was 4.3% with 32 placements, and rank-one placement was 0.4% with only 3 counts.

The distinction to notice is that TrueCar's decline is broad-based across presence, coverage, and placement, with the brand now registering occasional negative sentiment (1 of 738 observations). This is not a placement-only or presence-only issue.

Highest-priority diagnostic: Which competitor pairings and prompt categories account for the 7.1-point presence decline, and is the erosion concentrated in a specific AI surface?

Autotrader's Baseline Decline Masks a Sharper October Recovery

Autotrader fell 3.5 points in valid recommendation coverage from 89.3% in July 2026 to 85.8% in October 2026, crossing the significance threshold. The prior-to-current movement, however, was sharply upward: Autotrader rose 3.4 points from 82.4% in September 2026, falling just short of also qualifying as a significant move.

Raw mention presence held essentially stable at 98.5% with 727 of 738 observations, the highest presence rate in the benchmark. Top-3 placement eased from 46.7% to 43.8%, while rank-one placement rose 1.7 points from 11.3% to 13.0%, with 96 rank-one counts.

The distinction to notice is that Autotrader's October movement is the sharpest upward reversal in the category this month, but it remains a partial recovery relative to its July baseline. The brand is closer to its baseline than it was in September but has not erased the earlier decline.

Highest-priority diagnostic: Which prompt categories drove the September-to-October recovery, and are those the same prompts that drove the July-to-September decline?

Cars.com Posts the Largest Top-3 Gain in the Series

Cars.com fell 1.7 points in valid recommendation coverage from 81.7% in July 2026 to 80.0% in October 2026, within normal variation. The prior-to-current movement was 1.6 points upward from September, also within normal variation.

Raw mention presence rose 2.6 points from 93.3% to 95.9%, with 708 of 738 observations mentioning the brand. Top-3 placement rose 8.9 points from 17.5% to 26.4%, the largest baseline-to-current top-3 gain in the series, with 195 top-3 placements. Rank-one placement eased slightly from 3.3% to 3.2% with 24 rank-one counts.

The distinction to notice is that Cars.com improved its presence and its top-3 placement without converting that into rank-one leadership. The brand is appearing more often and being placed higher in shortlists, but it is rarely the single best answer.

Highest-priority diagnostic: Which prompt categories drove the 8.9-point top-3 gain, and why has that gain not translated into rank-one placements?

CarGurus, Carvana, and Lithia Motors / Driveway Move Modestly

CarGurus fell 3.1 points from 89.1% in July 2026 to 86.0% in October 2026, just below its significance threshold. Its top-3 rate fell 6.1 points from 53.9% to 47.8%, even as its rank-one rate held at 26.8%, the highest in the category by a wide margin. CarGurus recorded 635 valid recommendations and 198 rank-one counts.

Carvana fell 2.1 points from 88.6% in July 2026 to 86.5% in October 2026, within normal variation. Its rank-one rate fell 4.0 points from 18.6% to 14.6%, with 108 rank-one counts and 638 valid recommendations. Its top-3 rate fell 3.6 points to 33.9%.

Lithia Motors / Driveway fell 5.3 points from 16.1% in July 2026 to 10.8% in October 2026, a significant baseline-to-current decline. Raw mention presence fell 6.4 points from 17.8% to 11.4%, with 84 of 738 observations mentioning the combined entity. The brand recorded 80 valid recommendations and a top-3 rate of 4.3%, up slightly from 3.6% in July. The baseline-to-current decline reflects the August entity-tracking change as much as a competitive shift, since the parent was tracked under a combined label in July and split in August.

Highest-priority diagnostic: For CarGurus and Carvana, which prompt categories drove the rank-one declines while top-3 rates held closer to baseline? For Lithia Motors / Driveway, how should the current 10.8% be reconciled with the Driveway standalone series, which showed a 0.0% reading in October after an 8.7% appearance in August?

Small-Count Brands Move Within Their Historical Bands

CarsDirect rose 0.4 points to 1.9% coverage with 14 valid recommendations and 22 observations mentioning the brand. DriveTime rose 0.5 points to 3.8% coverage with 28 valid recommendations and 48 observations mentioning the brand. Both movements are within normal variation, and both brands remain near the bottom of the coverage distribution.

Driveway as a standalone label registered 0.0% coverage in October 2026, unchanged from September, after an 8.7% reading in August 2026 when it was first broken out separately.

The distinction to notice is that these small-count brands sit in a different measurement regime from the top of the category. Their percentage movements rest on dozens of valid recommendations rather than hundreds, so single-digit swings are not directly comparable to swings at the top of the benchmark. The counts are named here so the smallness of the base is visible.

Highest-priority diagnostic: Are CarsDirect and DriveTime presence patterns concentrated on specific AI surfaces, or are they distributed evenly in a way that suggests stable niche positioning?

Buyer-Intent Interpretation

Questions This Section Answers

• Which buyer-intent clusters can the benchmark currently answer?
• Why can't the public benchmark explain pricing or head-to-head comparisons yet?

In October 2026, all 738 qualified observations fell into the Brand Recommendation cluster, with zero observations classified under Pricing & Value or Multi-Brand Comparison. The public benchmark can currently answer which online car buying sites are present and recommended inside AI answers, but it cannot answer how AI systems frame pricing differences, value trade-offs, or direct head-to-head comparisons between brands.

The response-type distribution does show 24 comparison analyses and 9 pricing analyses in the October collection, indicating the raw prompt set contains comparative and pricing-intent queries. Those observations were not separated into distinct buyer-intent clusters in the qualified output, so the commercial questions about price and value remain open for company-level analysis.

Brand Opportunity Summary

Questions This Section Answers

• Which diagnostic question is highest-priority for each brand in October 2026?
• Which brands show declining coverage versus movement within normal variation?

The benchmark identifies where attention is warranted; a company-level analysis is needed to explain why.

Evidence Behind the Benchmark

The aggregate metrics are built from prompt-level observations (query, surface, recommendation outcome, rank, sentiment, and citations where exposed). Company-level analysis can go deeper into prompt, competitor, surface, and evidence patterns. Source presence is not automatically treated as proof of causation.

About This Benchmark

This report is part of the CiteWorks Studio AI Visibility Industry Market research program.

• AI Visibility Industry Market
• AI Visibility Industry Market Methodology
• AI Visibility Industry Market Metrics
• AI Visibility Industry Market Standards

Report-Specific Interpretation Notes

• Small-count movement: CarsDirect (14 valid recommendations), DriveTime (28 valid recommendations), Driveway (0 valid recommendations), and Lithia Motors / Driveway (80 valid recommendations) have small observation bases. Their percentage movements are valid but should be interpreted with the counts named alongside them.
• Structural tracking changes: The Lithia Motors / Driveway series carries the effect of an entity-attribution split in August 2026, when the parent was separated from the Driveway standalone. The baseline-to-current comparison for that combined entity reflects both competitive movement and the tracking change.
• Qualified denominator versus raw collection: All brand-level percentages are calculated within the 738 qualified observations for October 2026, not the 800 raw prompts collected. The public funnel covers both qualification stages.
• Directional analysis: Month-over-month movement identifies changes worth investigating; it does not by itself establish the cause of those changes.

Next Step

The Public Benchmark Shows Where a Brand Is Winning or Losing. A Company-Level Audit Shows Why.

Beneath the aggregate percentages sit the questions that matter for competitive positioning: which high-intent prompts are won and lost, which competitor takes the recommendation when a brand loses, what attributes AI systems associate with each option, and which external sources shape those answers. The October 2026 data shows a stable leader, a partially recovering top tier, and continued erosion in the mid-tier, but the public benchmark cannot explain the prompt-level dynamics behind these shifts, the source patterns that produce the return-window and inventory inconsistencies flagged above, or why Cars.com's top-3 gain has not translated into rank-one placements.

A company-specific AI visibility audit maps those prompt, surface, competitor, ranking, sentiment, and evidence-source patterns into a prioritized visibility strategy. This moves from knowing a brand's coverage percentage to understanding the specific queries, surfaces, and evidence sources that drive it.

Request an AI visibility audit