Audit your own site first, then call the agency

Most dealership briefs arrive as a feeling. Traffic is down. The store is not showing up. The competitor is everywhere. Someone has already decided the answer is SEO, or paid search, or a new site, and the agency's first job is to agree in a way that can be invoiced.

The agency will run a discovery audit anyway. You will pay for it, or you will pay for it inside the retainer, and it will tell you a story about your own website that you could have established yourself in a week with nothing but a browser, your analytics login and your DMS.

Do the audit before the call and the first meeting becomes a strategy conversation instead of a sales conversation.

Why this is your job, not theirs

An outside auditor sees the public site. You see everything behind it: what the DMS feed actually publishes, which OEM program constrains the page, who is allowed to publish, what the CRM records as a lead, what was tried in 2023 and abandoned, what the margin per unit and per repair order actually is. Those facts decide which recommendations are possible, and no agency can discover them from a crawl.

There is also a commercial reason. An agency that leads with your own numbers cannot sell you a narrative. You become a buyer of judgment rather than a buyer of the baseline.

Part one: what you can check without buying anything

1. Count what is indexed. Pull your sitemap and count URLs by type: vehicle detail pages, model pages, service pages, location pages, campaign pages, blog, tag and filter permutations. The ratio is the story. A store with 4,200 URLs where 1,900 are inventory filter combinations has a crawl problem, whatever the traffic chart says.

2. Read your own site with JavaScript off. Open a VDP, a service page and your hours page with scripting disabled. Whatever remains is what a fetcher that does not execute JavaScript can see. If price, availability and hours vanish, that is a finding, and no amount of content strategy fixes it.

3. Search your own site in Google. Run site: queries for your top services and models. Look at the titles that come back. Do they read like a template, or do they say what the page is? Are there three near-identical pages for the same service?

4. Ask the twenty questions your customers ask. Not keyword tools. The real questions: what does a brake job cost on this model, who has this truck in stock, what are your hours on Sunday, do you take trade-ins with negative equity. Ask them in Google, in ChatGPT and in Perplexity on a clean session and record whether your store is named, quoted or absent. That recording is your baseline; keep the date.

5. Read robots.txt. Look for accidental blocks: an inventory folder disallowed by a template rule, a staging path left open, a campaign folder blocked by whoever built it. This is one of the most common findings in an audit and it takes four minutes.

6. Check whether analytics is telling the truth. Is the tag firing on VDPs and on mobile? What exactly is a conversion: a form submit, a phone tap, a chat open? Does the CRM agree with the platform? A channel cannot be judged on numbers that were never collected.

7. Load speed with field data, not a lab score. Look at the real-user core web vitals for your highest-value templates, mobile only. Note the templates, not the homepage.

8. Hunt for the dead and the duplicated. Retired campaign pages returning 404s. Parameter URLs that got indexed. Redirect chains three hops long. Canonicals pointing at the wrong URL. Write down ten examples with their URLs.

9. Inspect the structured data. View the source and find the JSON-LD on a VDP: vehicle, offer, price, availability, location. Then compare it with what the page visibly says. When price and schema disagree, every surface downstream inherits the error.

10. Inventory the content and its owners. Which sections exist, when each was last updated, who can publish. If the answer to "who can publish" is "anyone," that is a governance finding and it will outrank every tactic in the proposal.

Part two: the internal questions only you can answer

• What is the real lead-to-sale picture by source, from the CRM, not the ad platform?
• Which OEM programs constrain the site: approved vendor lists, creative pre-approval, co-op documentation?
• Who signs off on a claim, and how long does that actually take?
• What is the gross per unit and per repair order that sets your tolerable acquisition cost?
• What was tried before, by whom, and why did it stop?
• What is committed in existing contracts: website vendor, call tracking, review tooling, chat? What cannot be moved this year?
• Who touches the site today, and who would have to be coordinated for anything to change?

Part three: turn it into one page

Numbers, not adjectives. Something like: 1,842 indexed filter URLs; 31 percent of VDPs render no price without JavaScript; 14 service queries tested, cited twice; robots.txt disallows the accessory catalog; last content update on the service hub was March 2024.

Attach the examples. Add the constraint list, the decision rights and the calendar of programs that affect the site this quarter. Do not prescribe the solution. State the problem, the evidence and the limits, and let the proposal answer them.

Part four: what the audit buys you in the room

You can tell diagnosis from pitch. An agency that opens by restating your evidence and extending it is diagnosing. One that opens with an invented list of problems you have never seen is pitching, and the list will be shaped to fit a retainer.

You can price scope honestly. Page count, market count, template count, language count. Scope arguments disappear when both sides count the same URLs.

You can ask better questions. What changes in the first thirty days and what will be measured. Which templates you will not touch and why. What happens to a campaign URL when it expires. What does the reporting look like in month four, when nothing dramatic has happened yet.

You can spot the red flags. Guaranteed rankings. A proprietary AI nobody can describe. Reporting that only shows impressions. No named owner for the account. A migration plan that omits redirects. None of these stand out to a buyer who has no baseline; all of them stand out to one who does.

The audit is also the handover document

Every good agency relationship starts with someone handing over what is known. When you do that first, the engagement begins at month three instead of month one, and the first invoice buys work rather than orientation.

The store that audits itself is not doing the agency's job. It is establishing the facts that only the store can establish, and then buying judgment on top of them.

The report page is the version of that handover we would want from a store: stated denominators, paid kept separate, and campaigns ranked by appointments rather than impressions.