How much dealership SEO costs, and why almost nobody will tell you

A dealer can request five proposals for search work, read all five carefully, and finish the exercise without a single number they can put next to another number. That is not because the work is hard to price. It is because a discovery call does something for the vendor that a published rate card cannot.

This is the piece the category does not write. What follows is what dealership SEO actually costs, the four pricing models a dealer will get quoted, what genuinely moves the number up or down, and our own rates stated plainly rather than held back for a call. The broader scope of the work those rates buy is set out on the car dealership SEO page.

Why almost nobody publishes a rate

Start with the honest reason, because it is not complexity. Dealership SEO is not harder to price than a service contract or a floorplan line. Vendors price it internally with tiers and per-rooftop rates like everyone else. The number exists. It is simply not shown.

The discovery call is doing three specific jobs. It lets the vendor size the quote against the perceived budget, so a fifteen-rooftop group and a single store hear different numbers for work that differs less than the gap suggests. It lets the vendor handle objections before a figure is anchored, which is worth real money in a negotiation. And it keeps the vendor out of a spreadsheet where a dealer could line up five quotes against each other on the same axis.

None of that is unusual and none of it is dishonest. It is standard practice in a category that still runs on bespoke quotes. But a dealer should understand what the missing number is buying, and who it is buying it for. Transparent pricing is a differentiator here precisely because most platform and boutique competitors do not publish real rates, which tells you something about how much the opacity is worth to them.

The practical consequence lands on the dealer. Without comparable numbers, the evaluation collapses onto whichever proposal reads best, and proposals are written by people who write proposals for a living. The question of whether the work is any good gets decided by the quality of the sales document. The way out of that is a set of questions the answers to which cannot be dressed up, which is the exercise in evaluating an automotive SEO agency.

The four pricing models a dealer actually encounters

Almost every quote a dealership receives is one of four shapes. Knowing which one is in front of you matters more than the headline figure, because two of the four cannot be compared to anything.

Platform-bundled

SEO appears as a line item inside the website contract. It might read $700 or $1,200 a month, sitting under the site fee, the hosting, and the inventory feed. It looks like the cheapest option on the page and frequently is not, because that line carries a share of the platform whether or not the dealer needs the platform underneath it.

The deeper problem is legibility rather than price. Bundled SEO cannot be separated from the website contract, cannot be independently audited, and cannot be taken along when the contract ends. A dealer asking what the SEO line specifically bought last month is asking a question the invoice was not designed to answer. That structure, and what it costs to leave it, is the subject of the managed SEO platform explainer.

Flat retainer, per rooftop

A stated monthly figure per location, usually tiered by scope. This is the only one of the four models that lets a dealer put two vendors side by side and compare them on the same axis, which is presumably why it is the least common in a category that would rather not be compared.

It is also the model that makes underperformance visible. A flat rate against a defined scope produces an obvious question every month: did the scope get delivered. Percentage and bundled models both blur that question into something harder to ask.

Percentage of media spend

The fee floats as a percentage of the advertising budget. A dealer spending $40,000 a month on paid pays more for search work than one spending $15,000, for reasons that have nothing to do with the search work.

Treat this as a red flag rather than a preference. The fee rises when the ad budget rises and falls when it is cut, and neither movement corresponds to anything happening on the earned side. It also makes the invoice unreadable in exactly the season it matters most: when the paid line swings, a blended fee hides which half of the budget did anything. Earned search should be priced on scope, not on somebody else's media plan.

Project and à la carte

Individual deliverables priced one at a time. A schema implementation here, a batch of location pages there, an audit as a one-off. Genuinely scoped one-time projects are a legitimate way to buy specific work, and we price and publish those too.

The failure mode is using them as a substitute for a program. Search work compounds or it does not work, and a sequence of disconnected projects tends to produce a site with several good pages and no coherent structure underneath them. Buy a project to fix a specific thing. Do not buy twelve projects instead of a retainer and expect the result to add up.

What actually moves the number

Four variables account for most of the spread between a $1,000 quote and a $5,000 one, and only one of them is about effort in the ordinary sense.

Rooftop count, first and largest. Most of the work repeats per location, but the foundation does not. One schema implementation, one set of templates, one competitive picture, and one technical baseline serve the whole group, so the marginal cost of the fifth rooftop sits well below the first. Any pricing model that charges the fifth rooftop the same as the first is either overcharging at volume or was overcharging at the start. The group-specific half of that work, including the cannibalization problem that only appears above one location, is covered on the multi-rooftop dealer group page.

Platform cooperativeness, second and least discussed. A site that accepts structured data, template changes, and new page types converts a retainer into output. A site that does not converts the same retainer into workarounds. This is the variable that most often explains why two dealers paying the same rate get visibly different results, and it is rarely named in a proposal because naming it means telling a prospect their website vendor is the constraint. The specific version of that problem is SEO for groups on a managed platform.

Inventory volume and turn rate. A store moving 400 units a month with fast turn is a feed and schema maintenance problem as much as a content problem. A specialty store with 60 units and long turn is the opposite. Both are legitimate, they simply consume the retainer differently, and a vendor who has not asked about turn rate has not scoped the work.

Whether AI-answer work is included or extra. This is the largest hidden variable in 2026 quotes. Generative engine work shares almost all of its technical foundation with ordinary search work: the same structured data, the same entity consistency, the same crawlability. Quoting it as a separate retainer on top means charging twice for one implementation. It is included in every tier we run, from month one, for that reason rather than as a promotion. What the discipline actually consists of is set out on the generative engine optimization page.

What we charge, stated plainly

Three tiers, priced per rooftop per month, month to month, each with a flat one-time onboarding fee charged once.

• Core, $1,165 per rooftop per month. The technical foundation, structured data, local profile and category accuracy, the content map, generative engine work, and monthly reporting.
• Compete, $1,615 per rooftop per month. Core plus a heavier content and competitive cadence, with reporting at a biweekly rhythm.
• Command, $2,425 per rooftop per month. The full program, with executive-format reporting as a distinct deliverable rather than a longer version of the same export.

A single-location dealer pays the single-rooftop rate with no band applied. An operator running several stores moves down a banded volume discount that starts at 10 percent off for two or three locations and reaches 32 percent at fifteen or more, applied to every location rather than only the ones past the threshold. Enterprise and large multi-store operations are quoted rather than published, because at that size the scope genuinely does vary and a published band would be a guess dressed up as a rate. The current tier breakdown lives on the pricing page, and what lands in each engagement is itemized under deliverables.

No ad spend passes through the retainer and no percentage of media is billed on top of it. The rate is for the earned search and answer layer only. That boundary is deliberate: a dealer should be able to read one invoice and know exactly what the search work cost, without reverse-engineering it out of a media buy.

What the number should include, and what should be billed separately

The line between retainer and project is where a lot of quotes get slippery, so it is worth stating.

Inside a real monthly retainer: technical work on the site as it exists, structured data that validates against what engines actually consume, content mapped to an owning page rather than published to a quota, local profile and category accuracy for each rooftop, competitive monitoring, and reporting that names which pages and which queries moved. In 2026 it should also include the AI-answer layer, for the reason above.

Legitimately separate: genuinely scoped one-time projects, priced and stated as projects rather than as retainer modifications. A page system build, a migration, a structured-data implementation on a platform that needs one. The discipline that protects the dealer is that a project has a defined deliverable and an end, and does not quietly become a recurring line.

Not a thing we sell at all: content by the batch. Article packs, comparison pieces, and location pages priced as standalone units are the mechanic by which a retainer becomes a quota, and a quota is how a site ends up with a hundred pages nobody assigned an owner to. Content is part of a program or it is noise with an invoice attached.

The pricing red flags

Six things that should slow a dealer down, none of which require any technical knowledge to spot.

• No number without a call. The first and most reliable signal. If the rate cannot survive being written down, it is not a rate, it is an opening position.
• A fee that floats with ad spend. Covered above. It prices the search work against a variable that has nothing to do with the search work.
• An annual term with no exit. Long contracts protect the vendor in exactly the scenario where the dealer most needs the option to leave.
• AI or GEO quoted as a separate retainer. It shares its foundation with the technical work already being billed. Two retainers, one implementation.
• An onboarding fee that recurs. A one-time setup charge is reasonable. The same charge appearing again under a different name is not.
• The site and the data stay with the vendor. A price is only meaningful if the dealer keeps what it built. Ask what happens to the content, the schema, and the site at the end of the term, and get the answer in writing.

The last one is where the most money quietly goes missing in this category, because it is not a line on any invoice. A program that cannot be taken along at the end of the contract has to be rebuilt from zero by whoever comes next, and the dealer pays for that twice without either charge ever being labelled. The three honest ways out of that structure are laid out in the managed platform alternatives piece.

What to budget, by size

Rough shapes rather than promises, and stated as budget planning rather than as a market survey.

One rooftop. Expect to pay the undiscounted per-rooftop rate. Below roughly $1,000 a month the arithmetic stops working for anyone doing genuine technical and content work, and what is usually being sold at that level is a reporting subscription with some blog posts attached.

Two to five rooftops. The band starts applying and the per-rooftop number should visibly drop. This is also the size at which cannibalization between locations becomes a real cost, so a quote that treats five rooftops as five identical single-store engagements has missed the main problem.

Six to fifteen rooftops. The foundation is shared across everything and the per-rooftop rate should reflect that clearly. Reporting structure matters more than it did: a group at this size needs to see performance per rooftop and in aggregate, and most vendors supply only one of the two.

Above fifteen. Quote territory, genuinely. The scope varies enough between a fifteen-store domestic group and a fifteen-store import group with three websites that any published band would be fiction.

One thing that does not change with size: the number should be legible. A dealer at any of these tiers should be able to say what they pay per rooftop per month and what that buys. Surprising numbers of them cannot, and that is the actual finding underneath this whole piece.

Where to start if you are comparing quotes

Put every proposal into the same shape before reading any of them: monthly cost per rooftop, term length, what is included versus billed separately, whether AI-answer work is inside the rate, and who owns the assets at the end. Four of those five are yes-or-no questions and the fifth is a number. Any proposal that cannot be reduced to that grid is telling you something.

Then check the one thing pricing cannot tell you, which is whether the vendor is currently getting anyone named in the answers buyers actually read. That is a diagnostic rather than a sales exercise and it takes days rather than months.

FAQ