Category Page Optimisation for Car Dealers: 221-Page Study

Category pages are where a car dealer’s organic search is won or lost. The brand page and the model page, “New Kia” and “New Kia Sportage”, are the pages that match the highest-intent commercial queries in the buying journey, carry the internal links to every car in stock, and outlast every car listed on them. A vehicle page disappears the day the car sells. The model page stays, and keeps the links, history and relevance it has built.

Most published guidance on category page optimisation is written for ecommerce, and car retail breaks several of its assumptions. This guide is built for UK dealerships instead, on data from 221 brand and model pages captured from Carwow, Auto Trader, Arnold Clark, Evans Halshaw and Vertu Motors, with more than 120 attributes recorded per page. It shows what the market leaders put on their category pages, which buyer questions they leave unanswered, and what an SME dealership’s pages need.

01Why Category Pages Carry Car Dealer SEO

The page type that matches buying intent, and the one that survives stock turnover.

Car retail sites have two layers of category page. The brand page lists a manufacturer’s range, such as “New Kia cars”. The model page lists one model, such as “New Kia Sportage”. Below them sit trim pages and vehicle detail pages for individual cars.

They match the commercial queries

Queries such as “kia deals” and “kia sportage deals” are transactional: the searcher has chosen a make or a model and is choosing where to buy. In the Semrush UK database, “kia deals” shows around 2,400 monthly searches and “kia sportage deals” around 1,300. On Carwow, both land on category pages, /kia/deals and /kia/sportage/deals, not on individual cars. A vehicle detail page is too narrow for these queries and a blog post too informational. The category page is the only page type that matches them.

They are the permanent layer of the site

Every car in stock is unique, and stock typically turns over within weeks. Their pages appear and disappear with the stock, taking any links and relevance they earned with them. The brand and model pages are the stable URLs in between. They collect external links, internal links and search history over years, and they are the crawl path through which search engines reach new stock as it arrives. A weak model page does not just rank poorly itself. It weakens the route to every car beneath it.

Car retail is not ecommerce

Standard category page advice covers faceted navigation, product grids, pagination and ItemList markup. All of it applies to a dealer’s site, and none of it covers the parts of car retail that decide whether a buyer enquires:

• The price is usually monthly. Most private new car buyers pay through finance, so the monthly figure matters as much as the cash price.
• Finance advertising is regulated. Quoting a monthly figure brings FCA rules on financial promotions into play, which shapes what the page has to show.
• The buyer has a car to trade in. Part exchange is part of the price, and one of the biggest sources of doubt.
• Stock is scarce and local. “Can I get one, and where?” is a real question, not a formality.
• The dealer shares the query with bigger players. The manufacturer’s own site, national marketplaces and brokers all target the same make and model searches.

This guide covers the standard technical attributes, and then the attributes specific to car retail. It judges both against the questions a buyer brings to the page, an approach set out in my Pain Point SEO strategy. Nothing here claims to be a ranking factor. Section 15 explains why.

02What I Studied

Five sites, three brands, one day.

I took the brand page and every model page for BMW, Kia and Toyota from two marketplaces, Carwow and Auto Trader, and three franchise dealer groups, Arnold Clark, Evans Halshaw and Vertu. That gave 221 pages: 12 brand pages and 209 model pages, all captured on 25 September 2026, three weeks into the 76-plate change.

For each page I recorded more than 120 attributes across ten groups: title and headings, price and finance, stock, buyer help, trust, next steps, media, internal links, structured data and page build. I then checked the model pages against eight buyer questions. Section 14 sets out the full methodology and the steps taken to limit bias.

03The Anatomy of a New Car Category Page

Every attribute that matters, who uses it, and whether your page needs it.

This is the core of the guide. Each row is an attribute of a model page. The third column shows how many of the five sites use it on most of their model pages. The last column is my verdict for a smaller dealership: must have means the market treats it as standard or buyers need it, should have means several leaders do it, and edge means few or none do it, so it sets you apart.

Model pages only. “Sites using it” counts sites where the attribute appears on at least half of their model pages. Details can sit in tabs, pop-ups or linked pages this study did not open.

The rest of the guide takes these attributes group by group, with what each big site does and what a smaller dealer should copy or skip.

04Title Tags and Meta Descriptions

The one place where all five sites agree.

Every site builds its model page title from the same three parts: “new”, the make and model, and a buying word. They differ in the fourth part, which tells you what each site thinks the buyer wants.

Titles as captured on 25 September 2026.

Three details are worth noticing. Carwow is the only site with a number in the title, and it updates that saving model by model. Arnold Clark’s titles are the shortest, with a median of 17 characters, and leave out the company name. No site puts a model year or a location in the title.

Meta descriptions are weaker across the board. None of the 209 model pages put a price in the meta description. Vertu’s run to a median of 234 characters, longer than search results usually show, and some carry template glitches, such as a stray letter after the model name and an HTML entity printed as text. On a site with thousands of pages, one template error repeats on every one of them.

What to copy

• Title: “New [Make] [Model] for Sale in [Town] | [Dealer name]”. The first three parts match what every leader does. The town is the part the national sites cannot use, and it is the one advantage a local dealer has in the title.
• H1: the exact model name, with “new” and your town if it reads naturally.
• Meta description: a “from” cash price, a “from” monthly price and one reason to buy from you, in under about 155 characters. Nobody in this study does this, so it costs nothing to stand out.
• Proofread the template once, on a real page, before it goes out across your whole range.

05The Money Block: Price, Monthly Cost and Finance

The first thing a buyer looks for, and the most unevenly answered.

Share of each site’s model pages. 0% means not found in the main page content at capture, not proven absent.

Arnold Clark is the only site that answers all three money questions on every model page: a cash price, a “from” monthly figure, and the worked example behind it. Carwow leads with the saving against list price and showed no monthly figure at capture. Evans Halshaw’s model pages carry no price at all, and send the buyer to a separate offers page through a “View Offers” button.

The monthly figure matters most. Private new car buyers mostly pay monthly. The Finance & Leasing Association reported its members financed 91% of private new car sales at the point of sale in the twelve months to October 2018. A page with a cash price and no monthly figure skips the question a finance buyer cares about most.

What to copy

• Put a “from” cash price and a “from” monthly price side by side, near the top of the page.
• Show the worked example on the page, not only behind a link. Finance promotions in the UK fall under FCA rules, which in many cases require a representative example once you quote finance figures. Check the specifics with your finance provider or compliance adviser.
• State a saving against list price only if you can keep it current. Carwow updates its saving on every model. A stale saving is worse than none.

06Availability: Stock, Listings and Sorting

“Can I actually get one?” goes unanswered on 70% of model pages.

Arnold Clark states how many cars it has on every model page, in a heading such as “14 new Kia Sportage cars for sale”, and pairs it with a plate-change message. Vertu states stock on 43% of its model pages and is the only site with a sort control over its listed cars. Auto Trader, Carwow and Evans Halshaw almost never say how many cars are available.

For a smaller dealer, stock is the most honest thing you can put on a category page. A national marketplace aggregates stock from across the country. You can write “3 in stock at our Leeds showroom, 2 arriving in October”. That is a sentence none of the five sites can write.

What to copy

• State the count in a heading: “[Number] new [Model] for sale”, as Arnold Clark does.
• List the actual cars with price, monthly figure, trim and fuel on each card.
• Add a sort control if you list more than a handful of cars. Four of the five sites had none.
• If you have no stock, say so and give the order lead time. A buyer can plan around a wait. They cannot plan around silence.

07Answers: FAQs and Buyer Questions

Two sites answer buyer questions on the page. Three mostly do not.

Every Auto Trader model page carried a set of buyer questions, typically ten or more, and 94% marked them up with FAQ schema. Vertu carried them on 74% of model pages, without FAQ schema. Arnold Clark carried them on 8%, and Carwow and Evans Halshaw on none.

The kind of question matters. Auto Trader’s are mostly product questions: top speed, fuel economy, road tax, whether the car is any good. Vertu’s lean towards the purchase: whether the car can be bought on finance, where to buy it, which trim to choose. A dealer’s category page should answer the second kind, because the product questions are already answered by the manufacturer and the motoring press.

What to copy

• Write five to eight purchase questions per model: finance options, delivery time, part exchange, warranty, which trim suits which buyer, and whether the car can be test driven locally.
• Answer each in two or three plain sentences, with real numbers where you have them.
• Mark them up with FAQ schema once the answers are written. This study measured nothing about search results, so treat the markup as hygiene, not strategy.

08Trust: Reviews, Guarantees and Warranty

The questions buyers ask about the seller, not the car.

Customer reviews or ratings appeared in the page content on 92% of Auto Trader model pages and on every Carwow and Vertu model page. At Arnold Clark and Evans Halshaw, review badges sat in the site header or footer rather than in the page content.

Vertu is the only site that tells the buyer what happens if they change their mind. Every Vertu model page opens with a 14-day money-back guarantee and free home delivery within 30 miles, with conditions. No other site states a guarantee on its model pages.

Warranty is the clearest gap. Kia sells its seven-year warranty as a headline feature. Not one of the 24 Arnold Clark and Carwow Kia model pages mentioned a warranty in the page content.

What to copy

• Put your review rating in the page content, near the price, not only in the footer.
• State the manufacturer warranty on every model page, in words. It is free, it is true, and several leaders leave it out.
• Say plainly what happens if the buyer changes their mind, even if the answer is your standard cancellation terms. One site of five does this.

09Next Steps: Part Exchange and Calls to Action

Every site offers part exchange. Almost none explains it.

Part exchange is one of the most anxious parts of buying a car. The buyer worries the online valuation will drop at the forecourt, and that the discount on the new car will quietly shrink to pay for it. A button answers none of that.

The calls to action also show what each site wants the buyer to do next. Carwow asks the buyer to “Get offers”. Evans Halshaw offers “View Offers” and a test drive. Vertu offers “Buy online” and “Reserve online” on its model pages. Auto Trader leans on “Contact us”.

What to copy

• Explain part exchange in three steps on the page: how you value the car, when the price is fixed, and what can change it at handover.
• Offer one main action and one lighter one. For a local dealer that is usually “Book a test drive at [branch]” and “Value my car”.
• Name the place and the person where you can. A national site cannot say who will meet the buyer.

10Internal Links, Breadcrumbs and Structured Data

The technical layer, where the gaps are widest.

Every site shows a visible breadcrumb on every model page, and every model page links back to its brand page. Brand pages link to the full range: Carwow’s BMW page linked 40 models. On model pages, Carwow links to a median of five related models, the most of any site.

Structured data is where the sites diverge most.

Types found in JSON-LD on the captured pages.

No site marked up the dealer itself with the AutoDealer type, customer ratings with AggregateRating, or its listed cars as an ItemList on model pages.

What to copy

• Link every model page to its brand page and to three to five related models.
• Mark up breadcrumbs with BreadcrumbList. Two of the five big sites skip it.
• Mark up your business with AutoDealer, including your town and opening hours, and mark up the FAQs you wrote in section 07. As with all markup, this study cannot show it changes results. It takes an afternoon and removes doubt.

11Page Build: Length, Weight and JavaScript

More is not better, and heavier is not better.

Median main content ranged from 509 words on Carwow to 2,257 on Vertu. Length did not predict answers. Arnold Clark answered more buyer questions than Auto Trader with about a quarter of the words, and Evans Halshaw wrote more than Carwow and Arnold Clark while answering the fewest.

Page weight varied far more. Vertu’s model pages had a median HTML size of about 1.55 MB and around 5,200 page elements. Evans Halshaw’s weighed about 97 KB with around 760 elements, roughly 16 times lighter. Heavy pages are slower to load on a mobile signal, which is where many car buyers browse.

Almost none of Auto Trader’s model page content exists in the HTML its server sends. The median share of words that appear only after JavaScript runs was 99.9% on Auto Trader, against 0% on Carwow and 3% on Arnold Clark. Search engines can render JavaScript, so this is an observation about how the page is built, not a claim about its visibility. For a smaller site, putting the key content in the HTML is the safer choice.

What to copy

• Write until the eight buyer questions are answered, then stop. The leaders that answered the most did it in anything from about 570 to about 2,250 words.
• Keep price, stock and FAQs in the HTML, not loaded later by a script.
• Test the page on a phone on mobile data before and after each change.

12The SME Blueprint

One model page, top to bottom, built from what the leaders do and what they leave out.

The order matters as much as the content. The block at the top answers the questions every buyer asks first: what it costs, what it costs a month, whether it is available, and what to do next. Everything below it earns trust and removes doubt. The product detail sits near the bottom, because the buyer who reached this page has already chosen the car.

The brand page

Brand pages follow the same logic one level up. Among the five sites they ranged from about 490 words at Auto Trader to more than 4,000 at Vertu, and each linked to the brand’s full model range. For a smaller dealer, the brand page needs a clear title such as “New Kia Cars for Sale in [Town] | [Dealer]”, every model you sell as a card with a “from” cash and monthly price, a short note on the manufacturer warranty, and links to every model page.

13How the Five Sites Score, and Where SMEs Can Win

Eight buyer questions, averaged per model page.

I scored each model page on eight yes-or-no questions:

1. How much is it?
2. What will it cost me a month?
3. How is the finance worked out?
4. Can I actually get one?
5. What do buyers usually ask?
6. What do other buyers think?
7. How does part exchange work?
8. What if I change my mind?

No site answers everything. Arnold Clark answers the money and availability questions on every page. Auto Trader answers buyer questions on every page. Vertu is the only site with a stated guarantee, and it still leads at 4.9 of 7 if the guarantee question is removed.

A smaller dealer does not need to outspend these sites. It needs to answer the eight questions on one page, which only 6 of their 209 model pages manage.

That is the opportunity. The national sites are built for scale, so their pages come from templates, and a template answers only the questions its designers thought of. A local dealer with thirty model pages can write each one properly, name the branch and the people, state real stock, explain part exchange honestly and show the date the prices were checked. None of that needs a bigger budget, and most of it is missing from most of the market.

---

14Methodology

What was studied, how, and the steps taken to keep bias out of the results.

Study type

This is a cross-sectional, descriptive audit of publicly available web pages: a structured content analysis of category pages captured at a single point in time. It records what the pages contain and compares sites. It does not test what causes rankings or sales, and it makes no statistical claims about the wider market, because five sites chosen on purpose are not a random sample of anything.

Which category pages were selected

Five UK new car sites were chosen to cover the two business models a dealer competes with: marketplaces and brokers (Carwow, Auto Trader) and national franchise dealer groups (Arnold Clark, Evans Halshaw, Vertu Motors). Each has dedicated brand and model pages for new cars with a consistent URL pattern. Three brands were studied: BMW, Kia and Toyota, chosen to span a premium brand, a value brand and a volume hybrid brand.

Pages = brand pages + model pages analysed. Total 221: 12 brand pages and 209 model pages.

Included: brand pages listing a manufacturer’s new range, and model pages listing one model. Excluded: trim and derivative pages, individual vehicle pages, separate offers, lease and used car pages, manufacturer websites and editorial reviews. Auto Trader’s brand pages serve new and used cars together; they are included because they are the site’s brand-level category page, and the mix is noted wherever it matters.

How the sample was built: 15 brand URLs were generated from each site’s pattern and checked. Three were dropped because the site does not sell that brand new: Arnold Clark’s Toyota page redirects to its general new cars page, and Evans Halshaw’s BMW and Toyota pages return 404. Model pages were then taken from the links on each valid brand page, which found 223. Of the 235 pages in total, 8 redirected because the model had no page of its own, 2 were blocked by bot protection and 4 failed to load, leaving 221.

Data collection

Every page was captured on 25 September 2026, between midnight and noon, with the same instrument: a headless Chromium browser using one mobile device profile, the same waiting rules, automatic cookie banner dismissal and a scroll to the bottom of each page to load lazy content. Both the raw HTML and the rendered page were saved. The crawler read each site’s robots.txt, skipped anything disallowed, paused between requests, and did not attempt to get past bot protection. It did not click into tabs, pop-ups or linked pages.

Measurement

More than 120 attributes were extracted from each page’s text, HTML and JSON-LD structured data, using written rules applied identically to every site. For content attributes, navigation, header and footer were removed first, so site-wide menus and footer badges could not inflate a page’s score. Banners placed inside the main content, such as Vertu’s guarantee banner, do count.

The eight buyer questions were written from common new car buyer complaint themes, not from a review dataset: hidden costs, confusing finance, long waits, part-exchange disputes and fear of commitment. Each scores yes or no per model page, with equal weight. A site’s score is the average across its own model pages. The attribute verdicts in section 03 are my judgement, based on how widely the market uses each attribute and which buyer question it answers.

How bias was limited

Analysis

The analysis is descriptive: shares of pages with each attribute, medians for counts such as words and page weight, and average buyer-question scores per site. Brand pages are reported separately and kept out of the headline figures, because 12 pages are too few to summarise. No significance tests are reported, because the sites were not randomly sampled and the study makes no claim beyond the pages it measured.

15What This Guide Does Not Establish

The limits, stated plainly, so you can weigh them yourself.

What the big sites do is not proof of what works. These five sites rank and sell largely because of their size, brand and history. Copying an attribute from them does not guarantee their results.

No ranking data. I set out to link these attributes to keyword positions from Semrush. My plan could not export full position reports, and copying positions by hand captured only around 7% of the keyword rows for a single site. Rankings on these sites are also dominated by domain strength, and five sites is too few to separate the page from the domain. So this guide describes pages rather than claiming ranking factors.

Category pages are not whole websites. Evans Halshaw’s offers page, Carwow’s configurator results and Auto Trader’s lease pages sit beyond the category page, and this study did not measure them.

Three brands, one day, in a plate-change month. Other brands may be presented differently on the same sites, and offers, stock and messaging change quickly in September.

Automated detection misses things. Content inside tabs, click-to-load sections, pop-ups and images of text can be invisible to the capture. A 0% means “not found”, not “proven absent”.

Not a compliance review. Nothing here judges whether any site meets FCA or advertising rules.

16Frequently Asked Questions