A 90-day triage plan you can run without a replatform

Every dealer group we audit has a platform replacement document. It has a timeline, a vendor shortlist, and a sponsor who is careful not to commit. It is three years old.

Waiting for it is a decision, and usually the wrong one. Here is the sequence we would run in parallel, in ninety days, with people you already have.

Days 1 to 14: establish an honest baseline

Before anything, measure the thing nobody measures.

• Twenty questions your buyers actually ask, written down verbatim.
• Each asked in ChatGPT, Gemini, Perplexity, and Google, once.
• Logged: mentioned? cited? which competitor? which marketplace?

This is your only real baseline, and it costs nothing. Do it again on day ninety with the same questions. Two comparable snapshots beat a subscription you cannot interpret.

Days 15 to 30: fix legibility

Server-render product and category templates. No redesign. The same page, with the facts in the HTML.

Audit crawler access. Decide deliberately which retrieval crawlers you allow, then publish the policy.

Ship Product schema. sku, brand, offers, applications. Generated from the same data that renders the page, never hand-maintained.

If you can only do one item from this entire plan, do the structured data. It is the cheapest permanent improvement available to a legacy stack.

Days 31 to 55: fix the catalog

• Reconcile the Merchant Center feed against the page for the top 200 SKUs.
• Resolve brand, part number and vehicle naming to one format everywhere.
• Pull ACES applications out of the PIM and into both the page and the feed.
• Kill any generated description that asserts fitment it cannot source.

This is the unglamorous quarter that decides whether the next three years work.

Days 56 to 75: build the answer surface

• Twenty question-answer pages, each answering one question completely in the first two sentences.
• Spec and fitment tables on the top category pages.
• FAQ blocks with FAQPage schema where the questions are real.
• An llms.txt that points at what matters.

Write them like a human who knows the catalog. Do not publish machine-written fitment.

Days 76 to 90: rewire the paid side

• Audit which Search campaigns were auto-migrated to AI Max.
• Check final URL expansion against your top-spend destinations.
• Write the AI Brief with fitment and brand constraints stated plainly.
• Add text disclaimers for legal copy.
• Re-run the twenty questions and compare.

What to show the committee

Not a redesign mockup. Three numbers:

1. Crawlable product pages: before and after.
2. Mentions in the twenty-question set: before and after.
3. Paid landing page quality on the top twenty destinations: before and after.

A committee cannot argue with a before-and-after table the way it can argue with a strategy deck. That is your replatform business case, generated for free, in ninety days, on the platform they are refusing to replace.

You do not need permission to make the catalog legible. You need permission to spend money, and you mostly do not need that either.

The honest caveat

None of this fixes a platform that cannot publish structured data or server-render. If you hit that wall in week two, you have your answer about the replatform, and you have it with evidence.

Either way, you stop guessing. That is worth ninety days.

Week thirteen is the report. The report page shows the shape we hand back: paid separated from organic, every denominator stated, and campaigns ranked by booked appointments.