The journeys are identified and built. Seven branches, each triggered by behavior, each tested. Loyalty tier, booking window, past purchases, channel preference. The orchestration platform could run them tomorrow.
Then the request goes to the content team: forty-nine variants ASAP.
I’ve seen versions of this in hospitality, retail, and healthcare. The details change. The shape doesn’t. The stack was ready. The content wasn’t.
Delivery went modular. Content often didn’t.
Over the last decade, marketing technology has become remarkably composable. Decisioning engines, orchestration tools, work management platforms, customer data platforms. All built to recombine. A customer’s journey can branch as finely as the data allows.
Content stayed a deliverable. A page. An email. A banner. An asset in the DAM with a file name and, if you’re lucky, a campaign tag. Each one made whole, reviewed whole, and approved whole.
That model works when you ship a handful of versions. Personalization breaks it. Every branch multiplies the variants, and a deliverable model makes variants by hand, one at a time. The math stops working long before the technology does.
I’ve been arguing for months that content should be treated as a platform, not a deliverable. Personalization is where that argument stops being theoretical.
It shows up as a workflow problem
This is why the problem is hard to see. The bottleneck doesn’t appear in the content. It appears in the queue.
Requests pile up in the work management tool. Reviews stack. Legal reads the same claim forty times in fort-nine slightly different emails. Launch dates slip. Everyone agrees the process is broken.
So the process gets fixed. Intake is redesigned. Approval paths get shorter. Routing is automated. These are real improvements, and the people who make them are right that the problem is systemic.
But the queue fills back up, because the work entering it is still shaped wrong. Forty-nine finished emails will always take longer to approve than one set of approved parts assembled forty-nine ways.
It’s a content-model problem wearing a workflow costume.
What the missing layer looks like
Treating content as infrastructure isn’t abstract. In practice it comes down to three things.
- A content model: breaking content into structured parts with defined jobs—headline, offer, proof point, legal line, image—instead of finished compositions. The model is what lets a journey assemble variants instead of commissioning them.
- A shared taxonomy. The DAM, the CMS, and the decisioning engine have to describe content the same way. If the engine knows a customer is a lapsed loyalty member traveling with family, the content has to be tagged in terms the engine can match. Most organizations run three vocabularies and translate between them by hand.
- Governance at the component level: Approve the claim once, not every email that carries it. Review moves from pages to parts. That’s where the queue actually shrinks.
None of this is new. Content strategists have called it “create once, publish everywhere” for years. What’s new is that the delivery layer finally outran it. It addresses all these previous points.
AI makes this more expensive, not less
Generative tools look like the answer. If variants are the bottleneck, generate the variants.
But generation solves production, and production was never the hard part. Without a model, generated content is just more deliverable: faster, but harder to trust. :A generated claim still has to be checked every place it appears**.** More emails or campaign pages in the queue. More claims to review in more places. More assets in the DAM that nobody tagged in terms the engine understands.
Cheap production makes the missing model more costly. Output outruns anyone’s ability to say what each piece is for.
The question under the stack
Most personalization roadmaps assume a content layer that doesn’t exist. They plan the data, the journeys, the platforms, and the process. Content shows up as a line item: produce assets.
The stack is ready for personalization. The open question is whether the content is a system or a stream of deliverables.
Content as platform was never only an argument for content teams. It describes what every system downstream of content already assumes is there.
Want to talk about content as a platform to power your business? Schedule a free consultation.
