#Why the numbers vary so much
Because 'SEO' covers everything from publishing four blog posts a month to rebuilding how a site handles fifteen thousand inventory URLs. Those are different jobs with different costs, sold under one word.
The cheap end is usually content production on a schedule with no technical work. That is not fraudulent — it is just narrow, and it will not touch the problems that most commonly hold dealership sites back.
The expensive end usually includes technical work, platform negotiation, content, and someone senior thinking about it. Whether you need all of that depends entirely on what is currently wrong.
#What actually drives the price
Inventory volume and platform. A store with 400 vehicles on a platform that generates a URL for every filter combination has a fundamentally larger technical surface than one with 90 vehicles on a well-behaved site. More URLs, more ways to go wrong, more work.
How much control you have. On some platforms you can change titles, canonicals and internal linking freely. On others you file a ticket and wait a quarter. Constrained platforms cost more to work on, because effort goes into working around limits instead of fixing things.
Competitive density. Ranking in a metro with nine same-brand stores inside twenty miles is a different problem from being the only franchise for sixty miles. Same work, different amount of it.
Number of rooftops and whether they share content. Groups have a duplication problem single stores do not.
#What to ask before you compare two quotes
Ask what is explicitly excluded. This is the single most useful question and most proposals do not volunteer it. Technical fixes? Content? Google Business Profile? Reporting? The gap between two prices is usually sitting in this answer.
Ask who does the work and how senior they are. 'A dedicated account manager' often means one coordinator across thirty stores. That is not automatically bad — it depends what the work is — but you should know.
Ask what happens if the constraint turns out to be the website platform. If the answer is vague, you may pay a retainer for a year to optimise around a ceiling nobody will name.
Ask what they would need to see before committing to a number. A firm that will not quote before looking is behaving correctly.
#The thing that is not on the invoice
Your website platform is often the largest cost in the programme, and it does not appear in the SEO line item. If the platform will not let you fix how sold vehicles are handled, or will not expose the templates, then a portion of whatever you spend on SEO is buying workarounds.
That is worth knowing before you sign, not eighteen months in. It also reframes the decision: sometimes the highest-return marketing spend available is a platform change, and no SEO retainer at any price substitutes for it.
A good proposal will tell you this even though it is inconvenient, because it changes what the engagement can honestly promise.
#How to tell if you are getting value
Not rankings. Ranking reports are easy to make flattering by choosing the terms. Ask for non-brand organic sessions, indexed page counts by page type, and Search Console clicks segmented by intent — model, service, parts, local.
Look for the split between branded and non-branded. If all the growth is branded, something else in your marketing created the demand and SEO is capturing credit for it.
Expect the first few months to look like cleanup rather than growth, if the technical work was genuinely needed. Index quality usually gets worse-looking before it gets better, because removing thousands of junk URLs is the fix.
#Terms worth negotiating before you sign
Data ownership. If the engagement ends, what do you keep? Analytics access, the Search Console property, content produced, and documentation of what was changed. Content you paid for but cannot take with you was rented, not bought.
Notice period, and what happens during it. A ninety-day notice on a monthly retainer is worth knowing about before month eleven, particularly once a relationship has stopped working.
A change log as a deliverable. Not a report — a record of what was changed on the site, when, and by whom. This costs the vendor almost nothing and is the single most useful artefact when something breaks, or when you eventually work with someone else. Its absence is why so many stores cannot explain their own website.