Mirror the machine: what Audi and Volkswagen get right about platform change

A brand moving to a new content platform is the most instructive thing a dealer marketing department will see all year, and most of them watch it the way you watch a neighbour's renovation: vaguely irritated, mostly about the noise.

There are two moves in the Audi and Volkswagen pattern worth stealing. The visible one is the platform: components instead of pages, one asset library, market variants of the same content, brand sites and dealer authoring environments converging on Adobe Experience Manager. The invisible one is the gate: page additions reviewed by the manufacturer before they go live on the store's own domain.

Dealers treat both as overhead. Both are actually free consultations from a company that spent eight figures learning what a page is.

What actually changes when a brand moves to AEM

Adobe Experience Manager is less a website builder than a component factory. Templates, a component library, a digital asset library with rights and expiry, workflow states, and locale variants of one page definition rather than many copies of it. Audi's dealer and web teams have been authoring pages inside exactly that environment, and by 2022 the group was wiring AEM to Workfront so campaign work moved through a defined pipeline rather than an inbox.

When a brand rebuilds there, three decisions get made that dealers rarely make for themselves: what a page is, what a fact is, and who signs off. Those three decisions are the whole lesson.

The Audi and Volkswagen programs, in plain terms

Both brands run certified-provider dealer programs. Audi has a Certified Program covering digital marketing providers; Volkswagen runs Certified Website, Digital Marketing and Streaming TV programs with approved partners, which is why vendor announcements read like enrollment notices. The platform upstream and the vendor list upstream, the store executing inside both.

That is not sinister, and it is not free either. DealerRefresh has been arguing for years that OEM website policies overreach, and there is a real cost to running a store's only owned property inside someone else's rulebook. The honest read is narrower: the OEM supplies two things most dealers are bad at buying for themselves, an information architecture and a QA function. Take those two, and negotiate the rest.

Copy one: a page is a composition, not a build

Most dealership sites are accumulations. Six hundred URLs, four button styles, three headline fonts, a landing page a vendor built in 2021 that nobody can edit. Every new campaign means a new bespoke page, which means a new review, a new QA pass, and a new thing that will break at the next platform change.

In the OEM pattern a page is a template plus approved components plus content. That single shift does more for a marketing department than any tool purchase: a new page becomes an arrangement of blocks that already passed review, and the schema, accessibility and tracking come with the block rather than being remembered by a person.

Copy two: facts are objects, not sentences

A manufacturer states a model, a trim, a powertrain, an incentive once and references it everywhere. The same fact fills the model page, the comparison page, the offer tile, the dealer locator and the structured data, because it lives in a field rather than in a paragraph.

Dealers do the opposite. The price exists in a hero image, a paragraph and a table, and the three disagree by Thursday. Two consequences, and the second is the expensive one. First, drift: the vehicle detail page, the landing page and the ad all promise different numbers, and the customer notices before the desk does. Second, answer engines cannot quote you. A model that extracts facts from pages trusts the source that states them consistently in machine-readable form; a store whose offer lives in an image is invisible to that extraction, no matter how good the photography.

If you take one thing from AEM, take this: trim, offer, payment and service interval are fields, and every surface renders the same field.

Copy three: expiry is a field

Brand assets carry rights windows and end dates, because a lapsed usage claim is a legal problem rather than an aesthetic one. Dealers run last month's payment special on a live page for a quarter and find out when a customer arrives with a screenshot.

Put a start and end date on every offer object and schedule the unpublish. Memory is not a content management system, and the store that removes expired claims automatically avoids the most expensive kind of compliance finding: the one a customer can prove.

Copy four: variants, not copies

AEM's market pattern is one page definition rendered for many locales. The dealer equivalent is one service page definition rendered for many rooftops, or one campaign shell with a per-store offer. Dealers duplicate instead, so a fix has to be made in nine places and gets made in five.

Consolidation is unglamorous and it is where migration budgets are actually saved, because the pages you do not have to rebuild are the cheapest pages you own.

The review gate is release management, not a tax

Review-before-publish is standard in OEM dealer programs, and the complaint is always latency. Two things are true at once. The gate protects claims you would otherwise make unguarded: pricing, incentive eligibility, brand marks, legal disclosures. And a gate applied to bespoke pages will cost you weeks, every time.

The departments that handle it well run it as a release process:

• A pre-approved library. Eighty percent of new pages should be compositions of blocks that already passed. Only new copy needs a new review, and the review queue shrinks to the part that carries risk.
• Batched submissions. One drop a week on a calendar, not one email per page. Reviewers are people with queues, and a batching rhythm is the only lever you control.
• A page inventory. Owner, approval state, launch date, revenue role. If you cannot list what is pending, you cannot promise a launch date to the store.
• A shelf of approved-but-unlaunched pages so a stalled review does not leave the campaign calendar empty.
• The approval record as evidence. When a co-op claim needs proof of performance, the approved asset being the asset you ran makes the paperwork disappear.

The metric worth reporting is review touches per published page. It should fall every quarter. If it is flat, you are still building bespoke pages and calling it a platform.

What not to copy

The release cadence. A brand can wait for a quarterly train. A store cannot wait nine weeks to correct a payment on its highest-traffic page. Keep an emergency path for factual corrections, and use it only for facts.

Brand voice over inventory. National copy never names a stock number, because it does not have one. Your advantage is that you do.

Gatekeeping that outlasts the incentive. If review takes five weeks and the program runs thirty days, the page is dead before it publishes. That argument is won with arithmetic, not with complaint: show the forfeited window and ask for a fast lane on program-tied pages.

Vendor rigidity. Before renewing an approved provider, check that it can ship structured data, a real redirect map and a sub-three-second page. Approved does not mean capable, and the program will not audit that for you.

The migration checklist we would hold you to

1. URL inventory and the redirect map finished before the build, never after launch.
2. Parity first, consolidation second. Thirty to sixty days of identical pages, then merge.
3. Staging with structured-data validation, so schema errors surface before the crawl.
4. Your twenty highest-revenue pages preserved one to one, including titles and meta descriptions.
5. A baseline thirty days before cutover: organic sessions, impressions by page group, form fills, tracked calls. No baseline means no claim.
6. Index coverage reviewed weekly for six weeks, with the old and new sitemaps submitted.

Mirror the discipline, keep your own speed

Following the manufacturer here is not deference. It is refusing to relearn something at your own expense. The brand already decided that content has to be modeled, that assets need expiry, and that a second pair of eyes on a pricing claim is cheaper than a legal review. Copy all of that.

Then keep the one thing the OEM does not have: the inventory, the local offer, and the ability to change a number on a page the same afternoon a program changes.

That is the rule behind the in-house campaign section on our report page, where every sales and service play names the manufacturer program it has to agree with before it ships.