UK Car Dealer Content Strategy: Attributes Beat Authority

If you own a car dealership or run marketing for one, read this. There are key points here you can apply to your SEO strategy from day one, and most of them cost nothing to test. My study of 967 AI citations across 69 UK used car buying questions found something that changes how a dealer should spend a content budget: what decides whether you get recommended is not how big or authoritative you are. It is whether the page carries the attributes a buyer in a specific place is actually looking for.

Independent dealers took 3% of citations on national questions and 26.0% on town level ones, ahead of AutoTrader, Motors, CarGurus and Cazoo combined. I measured that in AI search because that is where the sources are visible and countable. The underlying principle is not new, and it applies just as directly to how you rank in Google.

A note on scope before we start. I measured AI citations because ChatGPT and Perplexity show their sources, which makes selection countable in a way Google’s ranking has never been. But nothing in this strategy is specific to AI. Pages that carry the right attributes for a local buyer rank better in Google too, for the same reason: they describe a narrower market with less competition in it and they answer the question more completely. If AI search disappeared tomorrow you would still want to publish these pages.

Two methods sit underneath the work. Pain point SEO supplies the attributes, which is to say what your buyers are actually worried about in their own words. A clear content structure keeps thirty pages from collapsing into mush. Sections 03 and 05 cover each.

01Attributes Beat Authority

The finding that should reallocate your budget.

The standard advice to a small dealer is to build authority. Publish more, earn links, wait for the domain to strengthen, and eventually compete. It is slow, expensive, and against AutoTrader it does not work. At 15.7% of every citation in my study and 152 individual appearances, AutoTrader outweighs the next single domain by more than three to one. You are not closing that gap.

You do not have to, because authority was not what the systems were selecting on. They were selecting on fit, and fit is decided by attributes. The more precisely a page describes a buyer’s actual situation, the smaller and less contested the market it competes in becomes. For a local business, the first and strongest of those attributes is where you are.

At town level, independent dealers took 26.0% of all citations. AutoTrader, Motors, CarGurus and Cazoo together took 25.7%. Ask a general question and the marketplaces own the answer outright. Name a town and the small dealers become, collectively, the most cited thing in the answer.

This was not one lucky website. The study surfaced 67 distinct UK independents, among them Bells Garage, Dudley Motor Company, Clarks of Kidderminster, Coventry Sterling Motors and Walsall Car Centre. The towns in those domain names match the towns in the questions.

The gate is not how much the question knows about the car. The gate is whether the question contains a location.

The practical consequence for your content strategy is immediate. A page targeting “used Ford Fiesta” competes with AutoTrader and loses. A page targeting “used cars in Dudley” competes with the other dealers in Dudley. Same effort, entirely different fight.

02What “Attribute Plus Location” Actually Means

The highest yielding page shape in the study, and why it works.

Location gets you into the conversation. Adding one more thing gets you cited every time.

The study tested six levels of question, each adding one piece of information. Here is what happened, and the two rows that matter are the last two.

n = 69 questions. Spearman’s rho = 0.597, p < 0.0001. Cell sizes are small, so read the direction rather than the decimals.

A model name neither helped nor hurt once a place was present. Place alone scored 81%. Place plus model scored 78%. With nine questions in one cell and sixteen in the other, that gap is noise. The honest reading is that the car model was neutral: it did not add to the place signal and it did not take anything away.

That matters for how you read this whole strategy. Model pages are not a mistake. If buyers in your area search for a specific model and you stock it, build the page. The point is narrower and more useful than “do not build model pages”: a model name on its own is not the attribute that lifts you. On a national question it scored 10%, identical to naming nothing at all, because a model describes a market rather than a buyer.

Adding a requirement to a place hit every question in the set. Budget, first car, warranty, family car, each paired with a town. Eight out of eight.

So what counts as a “requirement”?

This is the part worth slowing down for, because the answer decides what you write.

Look at what the winning questions actually contained. Under £10,000. Cheap. Bad credit. First car. None of those are product specifications. Every one of them is a buyer constraint, which is to say a problem somebody is trying to solve. A budget ceiling is a worry about affordability. “Bad credit” is a fear of being refused again. “First car” is a parent worried about insurance. “Cheap” is somebody frightened of buying a wreck.

So the framework is an attribute framework, and attributes come in two kinds. There are attributes that describe the product, like the model, the trim, the mileage. And there are attributes that describe the buyer’s situation, like their budget, their credit, their licence, their family. Both are legitimate. Only the second kind moved the numbers.

Buyer problem, plus location. That is the page shape. Everything else in this strategy exists to produce those pages and hold them together.

There is a reason this works in both surfaces. A page carrying a buyer problem and a place describes a small, specific market that a marketplace has no structural advantage in. AutoTrader holds more Ford Fiestas than any dealership on earth, so a Fiesta page is a fight you have picked with somebody far larger. Nobody owns “first car for a new driver in Dudley” in the same way, and a local dealer can answer it more completely than a national platform can. That is a competition argument, not an algorithm argument, which is why it holds in Google as well.

03Where the Problems Come From

Not a keyword tool. This is the pain point half of the strategy.

There is a recurring complaint among agencies doing this work: the keyword tool shows zero volume for exactly the queries that seem most promising. The complaint is accurate and beside the point. A tool measures what people type into a search box. The questions in my study were conversational, carried a budget and named a town, and queries of that shape have no volume reading anywhere.

If you pick topics from a volume column, you will systematically miss the pages that win.

The alternative is to take them from buyers. That method is pain point SEO, and its implementation guide covers the extraction process step by step. For a dealership, four sources do most of the work.

• Your own sales floor. The questions your team answers twenty times a week, in the customer’s words rather than yours. Highest yield source available and the one nobody mines. Ask a salesperson what buyers are nervous about and you will get five page ideas in as many minutes.
• Reviews, yours and your competitors’. A complaint names the thing that mattered and failed. Competitor one star reviews are a list of the objections you should be answering on your own site.
• Google Business Profile questions and live chat logs. Unfiltered, already written down, already in the buyer’s phrasing.
• The AI platforms themselves. Ask ChatGPT and Perplexity your town’s questions and record who gets named. Section 07 turns this into a proper diagnostic.

What you are looking for is the constraint, not the topic. “Warranty” is a topic. “What happens if it breaks down two weeks after I buy it and I have no money spare” is a constraint, and it is the one that produces a page worth citing. Specific claims persuade where general ones fail, and the same appears to hold for what gets selected into an answer.

04Why This Is a Local Play, and Why Now

The audience arrived faster than the competition noticed.

A local content strategy used to mean a Google Business Profile, some citations, and a thin location page. That was enough when local search was one box on one results page. It is not the situation any more.

BrightLocal’s 2026 Local Consumer Review Survey records that use of ChatGPT and generative AI for local recommendations climbed from 6% to 45% in a single year. Buyers are already asking these systems who to visit.

The part that should interest you more is the gap. The SOCi Local Visibility Index 2026 finds that fewer than half the businesses leading Google local search results also appear in AI local recommendations. Winning Google local guarantees nothing here. The two surfaces select sources differently, which means the dealers currently ahead of you in the map pack have not automatically taken the space.

And the stakes per citation are higher than in organic search. BrightLocal’s consumer research finds 72% of consumers assess three or fewer businesses before deciding, and only 1% look at more than ten. An AI answer naming three dealers is not a results page with ten blue links and an implicit invitation to browse. It is a shortlist. You are on it or you are not.

None of this is a reason to abandon what already works locally. Keep the profile accurate, keep collecting reviews, review platforms took 2.8% of citations in the study, so they are feeding the answers directly. The argument is that local now has a second surface, it is less contested than the first, and the content you need for it is different content.

05Structuring It So It Scales

The architecture. Fifteen pages need a filing system or they become mush.

Sections 01 to 04 gave you a page list. This section is about how those pages sit together, and it is the part that decides whether page eleven strengthens the site or dilutes it.

There is a well known complaint in agency circles that specificity is the enemy of scalability, that you can have thirty pages or you can have good pages, but not both. It is not an automotive problem. The Content Marketing Institute’s benchmarks find only one in three marketers report having a scalable model for content creation, and that 42% of those with an underperforming strategy blame unclear goals rather than resources or writing talent.

The resolution is to hold the structure constant and let your own local data supply the difference. Three levels:

The four rules that make it hold

1. Every page points upward. Answer pages link to their service page, service pages link to the hub. Two clicks maximum from anything to the thing it describes. That is how relevance accumulates instead of leaking away.
2. Make local data the variance, not the wording. There is a habit of keeping “about 30% variance” in phrasing between template pages so they do not look duplicate. Rewording the same content produces pages that differ lexically and say nothing different. Use your real price bands, your actual stock mix, the insurance groups you hold, what your finance partner accepts. Then the template scales and the data supplies the specificity.
3. Publish in bursts. A complete cluster shipped together reads as a finished set. The same pages dribbled out over six months read as an unfinished one.
4. Delete what qualifies nothing. One dealer cut their site by 60% and watched traffic rise. Pages answering no question and receiving no links make the whole site harder to understand while contributing nothing. My budget study reaches the same conclusion from the money side: fewer places, more depth.

This is where the brief does its work. The structure only holds if each page is commissioned to answer one question for one buyer in one place, and that has to be settled before anybody writes.

06The Publishing Plan, In Order

This will generate more pages than you can write well. Ordering matters more than coverage.

On choosing which towns. The study does not answer this and I will not pretend it does. In practice three things decide it: whether your stock matches what that town buys, whether the drive is realistic for a test drive, and whether a competitor already owns the answer. Run the section 07 diagnostic on each candidate and you will have a defensible order inside an afternoon.

07Knowing Which Problem You Have

Ten minutes, no cost, and it decides where the budget goes.

Absence from an answer has two causes, and they demand opposite fixes.

A page retrieved and discarded 215 times is not invisible. It is unconvincing.

How to run it

• Ask Perplexity your town’s questions. Five to ten, at problem plus place level. “Cheap used car dealers in Dudley”. “Used car finance with bad credit Dudley”.
• Open the sources panel. Perplexity shows both what it retrieved and what it cited. Note which domains appear in each.
• Find yourself in one column, the other, or neither. That single observation tells you which project you are running and stops you buying the wrong one.
• Check whether the page actually finishes the job. To answer a question about used car finance a buyer needs APR, deposit, term, total payable, and what happens if they are refused. Leave part of that out and the page gets retrieved and dropped, every time.

Most dealers told they are invisible in AI search go and buy technical SEO. Roughly three quarters of the time the page was found and rejected, and technical work changes nothing.

08Three Things to Stop Doing

Each one costs money and the data says it buys nothing.

The second row is a null result. I include it because the opposite advice circulates widely.

On schema markup, briefly. Mark your business up properly with AutoDealer, put your town in addressLocality and the surrounding towns in areaServed, then stop thinking about it. That is an afternoon’s work. My study measured nothing about schema, so I claim nothing either way. It is hygiene, not strategy, and the variable that separated a 3% citation rate from a 26% one was the content on the page rather than the JSON underneath it.

09What This Strategy Does Not Establish

The limits, stated plainly, so you can weigh them before spending anything.

10Frequently Asked Questions