The vendor swap landed. The new site runs on Overfuel — and it is a generational leap.

Executive summary

The June vendor migration the May audit warned about has happened, and it went the right direction. dnxmotorsales.com is now an Overfuel-built, Next.js server-rendered site with strong technical SEO, rich structured data, and best-in-class AI-search hygiene. Health score moved from 29 → 85.

Top 4 issues remaining

• Critical — Field CLS 0.43 (poor). Real Chrome users (CrUX p75) experience heavy layout shift. Lab CLS is 0.00, so the shift comes from late-loading content (hero image / web font / inventory cards / cookie or finance widget) that reflows after first paint. LCP (1.72s) and INP (53ms) are both Good.
• Critical — 301 gaps from the migration. Some legacy URLs redirect correctly (/creditapp → /apply-online 308, /blog → /, /inventory/used-cars → /cars), but three categories return 404: /aboutus, /customer-reviews, /chevy-silverado, and the old DealerCenter /bdp/* blog & vehicle detail pages. These were live URLs with equity in the May baseline.
• High — 26 near-town pages are 93% duplicate. The programmatic /used-car-dealership-near-{town}-va pages share ~93.4% identical body content; only the town name in title/H1/intro changes. At 26-page scale this is a doorway-page / thin-content risk under Google's policies.
• Medium — AggregateRating without on-page Review markup. The homepage exposes two AggregateRating blocks (5.0/16 and 4.5/660) but no individual Review items are marked up, and the two ratings conflict. Self-serving aggregate ratings without visible, marked-up reviews may be ignored or flagged by Google.

Top quick wins

• Add 301s for the four legacy paths above (one rule each) — recovers lingering link/brand equity.
• Reserve hero image dimensions / preload the LCP image and font to kill the field CLS.
• Differentiate the near-town pages: per-town intro, distance/drive-time, local landmarks, town-specific inventory or testimonials.
• Reconcile the two AggregateRating values and add Review item markup on /testimonials.

Before / after — May (DealerCenter) → June (Overfuel)

Platform architecture

Detected stack: Overfuel dealer platform, x-powered-by: Next.js, server-rendered HTML (253 KB on the homepage — content is in the markup, not JS-injected). Assets served from static.overfuel.com. Edge sets Cache-Control: public, s-maxage=10, stale-while-revalidate=59 and a finance cookie carrying a credit-tier pre-fill. Apex dnxmotorsales.com → www via 308. This is a purpose-built, AI-search-aware automotive platform — the opposite of the prior DealerCenter setup.

Technical SEO — 88

• robots.txt: Allow: / plus explicit allows for /llms.txt, /inventory.md, /inventory/, /vehicles/; Disallow: /*highlights[]= (faceted-param control); per-agent crawl-delay for ClaudeBot and barkrowler; sitemap declared. Clean and intentional.
• Sitemap: valid sitemapindex → 3 children: latest-inventory (5), inventory (107), page (47). lastmod timestamps fresh (2026-06-11).
• Canonical / meta: self-referencing canonical on www, correct OG tags (title, description, type, url, webp image), title and meta description well-formed and unique per template.
• Security headers: HSTS (max-age 31536000; includeSubDomains), CSP frame-ancestors, X-Frame-Options: sameorigin, X-Content-Type-Options: nosniff. Solid.
• 404 handling: unknown URLs return a true 404. Good.

301 redirect gaps — the one carryover Critical

The May audit's central risk was loss of legacy URL equity. The new platform handles much of it, but spot-checks against the locked May baseline show gaps:

Note: 307 is a temporary redirect. For permanent moves use 301/308 so Google consolidates signals to the target. Full legacy inventory is in baseline-urls-2026-05-25.txt / dropped-urls-need-301-2026-05-25.txt.

Content & on-page — 78

Homepage: real H1 ("Shop used cars in Forest and Altavista, VA"), descriptive copy, search + clear CTAs (Shop inventory, Value your trade, Get approved). VDPs: strong — unique title with year/make/model/VIN, H1, full spec content, and a 39-item FAQ block per vehicle.

High: 26 near-town pages are 93% duplicate

The programmatic /used-car-dealership-near-{town}-va set (Lynchburg, Bedford, Roanoke, Amherst, … 26 towns) measures 93.4% body similarity between samples. Title, H1, and the opening line swap the town name; the remaining ~3,600 words are shared boilerplate. Unique title/H1 keeps them out of "exact-duplicate" territory, but at this scale Google can treat them as doorway/thin pages. Differentiate each: per-town intro, drive-time/distance from that town to each lot, local landmarks, and town-relevant inventory or reviews — or consolidate to one strong "service area" page plus a few high-volume towns.

Structured data — 90

Among the strongest areas. Homepage @graph: AutoDealer, Organization, ContactPoint, two PostalAddress + GeoCoordinates (both lots), six OpeningHoursSpecification, AggregateRating. VDPs add Vehicle with Offer (price, priceCurrency, InStock, UsedCondition, VIN, mileage), Brand, EngineSpecification, ItemPage, and a FAQPage.

• Medium: Homepage carries two conflicting AggregateRating blocks — 5.0 / 16 and 4.5 / 660 — with no individual Review items marked up. Reconcile to one source of truth and add Review markup on /testimonials (where 2 AggregateRating blocks exist but zero Review items).
• Note: Per Google's May-2026 update, FAQPage rich results are deprecated for commercial sites. The per-VDP FAQ is still valuable as AI-citable content and on-page UX — keep it, but don't expect FAQ rich snippets.

AI search readiness — 95

Best-in-class and the biggest swing from May (18 → 95). The platform ships llms.txt (structured site map for LLMs: locations, inventory categories, finance, trade, service, about), a live inventory.md (machine-readable stock feed — "Total vehicles in stock: 112", make breakdown, timestamped), explicit robots allows for both, and a ClaudeBot crawl-delay rule. Schema carries the citable data points (price, VIN, mileage, hours, geo, NAP) that ChatGPT, Claude, Perplexity, and Google AI pull. This is exactly the hygiene the May audit said was missing.

Performance / Core Web Vitals — 70

Field data (CrUX, p75 real Chrome users):

The CLS is the problem. Lab CLS is 0.00 but real users see 0.43 — meaning something reflows after first paint that a fast test machine doesn't catch: the hero image rendering without reserved dimensions, a late web font (FOUT), inventory cards hydrating, or the finance/cookie widget injecting. Fix: set explicit width/height (or aspect-ratio) on the hero and all above-the-fold media, preload the LCP image + primary font with font-display: optional/swap, and reserve space for any injected widget. Mobile Lighthouse Best-Practices also scored 73 (6 failed audits) — worth a pass alongside the CLS work.

Accessibility — 92

Mobile Lighthouse: Accessibility 92, SEO 100, Best Practices 73, Agentic Browsing 67. Accessibility is good; remaining items are typically contrast and tap-target spacing. SEO is a perfect 100.

Local & NAP — consistent

Two locations, consistent in schema:

• DNX Motors Altavista — 1101 Burnbridge Road, Altavista, VA 24517
• DNX Motors Forest — 1163 Wards Road, Forest, VA 24551
• Phone: (434) 616-2229; dedicated /location-altavista + /location-forest + /directions pages; 26 surrounding-town pages target the service radius.

Off-site verification (GBP listings, NAP across Bing/Apple/aggregators, review velocity) was not crawlable in this pass — recommend a GBP audit for both lots and a citation-consistency check as a follow-up.

Recommendations — do in this order

1. [Critical] Fix field CLS 0.43. Reserve hero/above-fold media dimensions, preload LCP image + font, reserve space for injected widgets. Re-measure in CrUX after 28 days.
2. [Critical] Close the 301 gaps. Add permanent redirects: /aboutus → /about-us, /customer-reviews → /testimonials, /chevy-silverado → /inventory?make=Chevrolet, and bulk-map old /bdp/* URLs. Convert the /blog and /inventory/used-cars 307s to 301/308.
3. [High] De-duplicate the 26 near-town pages. Add per-town content (drive time, landmarks, town-specific inventory/reviews) or consolidate to a service-area hub + top towns.
4. [Medium] Reconcile AggregateRating + add Review markup. One rating source of truth; mark up visible reviews on /testimonials.
5. [Medium] Best-Practices pass. Clear the 6 failed mobile Lighthouse Best-Practices audits (console errors, third-party cookies, image aspect ratios).
6. [Low] Follow-up GBP / citation audit. Verify both lots' Google Business Profiles, NAP consistency across directories, and review velocity.