Technical SEO for car dealerships, inside the platform you cannot rewrite.

Technical SEO for car dealerships breaks into four workstreams, and they are not equally urgent on every site. The audit decides the order, because fixing Core Web Vitals on a site the engines cannot parse is money spent on the wrong layer.

An SRP is a search results page, the filtered inventory listing a shopper lands on when they ask for used trucks under $35,000 within twenty miles. It is the highest-intent page type on a dealer site and the one managed platforms govern worst, because the platform will render any filter combination a URL asks for and nobody at the group ever decided it should.

• Facets generate pages nobody chose to publish. Make, model, trim, year, body style, price band, mileage, drivetrain, and color combine into a URL space that is effectively unbounded. Crawlers find those combinations through the filter UI and work through them. SRP SEO starts with deciding which combinations match real search demand, making those indexable and internally linked, and closing the rest with directives instead of hoping crawlers lose interest.
• Templated titles and metas make every SRP compete with every other one. A single pattern applied across thousands of filter states produces thousands of pages an engine reads as near-identical. Distinct patterns per indexable facet tier, written the way shoppers phrase the query, are what separate an SRP layer that earns impressions from one that competes against itself.
• Empty SRPs stay live after the stock turns. A filter state that held a dozen vehicles last quarter holds none this quarter and still returns a 200 with an empty grid. Engines treat those as thin pages and discount the layer they sit in. Zero-result states need a status decision rather than a default.
• The listing grid often arrives after the HTML does. When inventory is injected client-side, what a crawler parses on first pass can be an empty template with no vehicles in it. Whether that resolves depends on the platform and on the crawler, and AI answer engines are less patient about it than Googlebot. How each platform handles this →
• Sort and pagination parameters accumulate underneath. Every sort order and page index is another URL pointing at largely the same inventory. Canonical logic and parameter handling on this layer are usually vendor-ticket work rather than something the group can change directly, which is why the scoping conversation comes before the fix.

SRPs are also where geographic demand lands, since a filtered listing is the closest thing a dealer site has to a page about a make and model in a city. That overlap is why this work runs alongside local SEO for car dealerships rather than separately from it.

A VDP is a vehicle detail page, one URL per unit in stock, generated by the platform from the inventory feed. That single fact decides where VDP SEO happens. The work lives in the feed and the template, because nobody is editing four thousand pages by hand and the page that exists today may be gone by the end of the month.

• Sold units get no indexation policy. The unit leaves the lot and the page does one of four things, depending on how the platform was configured years ago: returns a 404, redirects to the homepage, redirects to the SRP, or stays live as a listing for a vehicle nobody can buy. Each choice has a different cost, and most groups have never been asked which one they are living with.
• Feed-generated descriptions make near-identical units into near-identical pages. Three same-trim units in the same color carry the same generated copy at three URLs. The work is producing differentiation from the fields the feed does vary, at template level, rather than writing prose per unit.
• Structured data ships partial. A vehicle node with no mileage, a VIN that never made it out of the feed, a trim string that disagrees with the title on the page. Those gaps decide whether an engine reads the URL as a listing for a specific unit or as a page about a model. Vehicle schema markup, in detail →
• VDPs are the heaviest pages on the site. Large photo sets, chat widgets, payment calculators, trade tools, and every tag added since launch land on the same template. Slow VDPs get crawled less often, and on an inventory site that means a smaller share of stock is ever seen.
• VDPs go orphaned. A unit reachable only through a filter state that is blocked or never crawled has no internal path pointing at it. Sitemaps built from a stale snapshot make it worse, because the one signal that could have carried the page is listing inventory that already sold.

VDP SEO done properly is an indexation policy that survives turnover: correct status codes on sold units, redirects that resolve to the nearest live equivalent rather than the homepage, template-level content that differentiates, and sitemap output built from current stock. Where this sits in the wider program →

The schema layer is where AI search visibility and traditional SEO meet, which is why it gets oversold in both directions.

• What schema does. It removes ambiguity. An engine reading marked-up inventory knows what the page is, what the business is, where it operates, and what is in stock, instead of inferring all four and occasionally inferring wrong.
• What schema does not do. It does not confer rank, and it does not make a site citable on its own. A perfectly marked-up group with no corroboration anywhere else still loses to an equal competitor that third-party sources describe. Why corroboration decides citations →
• Why it matters more now. A generated answer has to justify itself from sources it parsed. Markup that resolves cleanly is the difference between being a usable source and being skipped, and skipped is not a ranking you can see in a report.
• Where dealer platforms fall short. Inventory markup is routinely shipped partial: offers without price, listings without availability, organization markup that disagrees with the Google Business Profile two clicks away. Each of those is an ambiguity an engine has to resolve, and it resolves them conservatively.

If neither classic search nor AI engines can crawl and render your content properly, there is no content strategy good enough to compensate. That is the whole argument for doing this layer first. What runs on top of it →

• The pages that convert are the heaviest pages on the site. A vehicle detail page carries a photo carousel of thirty to sixty images, a payment calculator, a trade-in tool, a chat widget, a finance application and frequently a video walkaround. Every one of those is third-party JavaScript loading on the page a buyer reaches from a phone at the end of a search. The homepage passes Core Web Vitals and the VDP does not, which is the wrong way round.
• Interaction delay is the metric dealer sites fail. Layout shift gets attention because it is visible, but the failure that costs a dealership is a page that renders and then does nothing for two seconds while widget scripts contend for the main thread. A buyer tapping a gallery that does not respond leaves, and the session records as a bounce that looks like a content problem.
• Mobile-friendly design is mostly subtraction on a dealer site. The useful work is removing or deferring what is not needed on first paint: lazy-load the carousel past the first image, defer chat until interaction, load the payment calculator on tap rather than on render, and serve modern image formats at the size the viewport actually uses. Almost nothing on that list requires a redesign, and most of it is available inside the platform.
• Tap targets and forms decide whether the lead submits. A lead form with fields sized for a mouse, a phone number that is not tappable, and a sticky bar covering the submit button will pass every automated mobile test and still fail the only interaction that matters. This is worth checking by hand on a real phone, on the three pages that produce leads.
• Parity between the mobile and desktop rendering. Content hidden from the mobile template is content the index does not see. Dealer platforms sometimes drop specification tables, dealer comments or schema from the mobile view to save space, which removes exactly the detail that makes an inventory page worth citing.

The practical order is to measure the VDP and SRP rather than the homepage, because platform demos and vendor reports both default to the homepage and it is the least representative page on the domain. What those two page types need → The widget question also has a cost side, since most of the scripts slowing those pages are rented tools the store pays for. How that cost gets counted →