The Friction of Manual Scale
Every launch meant building pages from scratch. Shoots produced 35–40 assets with fewer than half making it to production. Copy was written before images were confirmed. Marketing and ecommerce were claiming ownership of the same pages. Teams finished hours before go-live.
The same 3-week process, across 22 markets, running continuously for a full year.
Building a shared content structure
Getting five teams onto a single content structure meant 20+ alignment sessions — product, engineering, brand, SEO, and marketing, each with a different idea of what the content model needed to do.
SEO needed subtitles under collection titles. Marketing needed a benefits section. Paid social needed fewer products per page. Each requirement made sense in isolation. The sessions weren't about compromise. They were about finding the constraints that could hold across all use cases and building the naming convention around those: prefix, purpose, size.
It's still in use today.
Five collection templates, sized by product complexity. One naming convention to keep every team and every market building from the same structure.
The content framework
Most content problems get solved with more process. I wanted to solve it with less. That meant designing a framework tight enough to create consistency, but flexible enough that teams wouldn't route around it.
Understood the backend structure
with BE & FE dev
The platform's distinction between single reusable values and structured reusable units shaped the entire content model. In Shopify, these are metafields and metaobjects: one stores a single piece of content, the other stores a structured, repeatable unit.
Mapped products & collections
with E-commerce
We sized templates by content complexity, not by SKU. A bottle needs to explain material, thermal properties, and care instructions. A coaster doesn't. The Lucy smart cap needed the most explanation of all. XS to XL reflects how much a customer needs to understand before buying.
Defined template logic
with Brand, Copy and SEO
We split every template into two variants: launch and evergreen. Launch templates carried space for temporary USPs, promotional mechanics, and limited edition messaging. Evergreen templates were built to last.
Collection and product templates architecture.
How the system works
What makes this system work isn't the components, it's the hierarchy.Each layer does one job: content units hold the source information, sections shape how it's grouped, templates define how it's presented. The result is a live store that feels tailored, built entirely without custom design work.
From content unit
Atomic content blocks
Small, clearly defined content units hold the source information. These are created once and reused wherever needed.
To section
Building blocks
Metaobjects are grouped into sections that can be combined and reordered based on context and use case.
To template
Structured storytelling
Sections are assembled into templates that define how a product or collection is presented, without hardcoding content.
To live store
Consistent output at scale
Templates render dynamic pages that feel tailored, while relying on the same underlying content system.
From a single content block to a live store - the same structured hierarchy, scaling across 22 markets without custom design work.
Driving Adoption
A framework nobody uses is just documentation. For this to stick, it needed to become the default, not an optional resource teams could ignore under deadline pressure.
That meant building it with teams, not for them. Every team knew what was coming before it launched. No surprises, no features forced onto anyone. A final alignment session brought everyone to the table before anything went live.
The hardest negotiation was convincing teams that a shorter page would outperform a longer one. Every team had valid reasons to add more. Each section had to earn its place.
The templates have evolved since. Sections added, sections removed as needs changed. The framework hasn't. That's the proof it held.
waterdrop® Shopping App
In 2024, waterdrop launched a standalone shop app. The content architecture built for the website — the content model, the naming conventions, the structured templates — became its foundation. The shop app now generates approximately €1M/month.
The content system did not just fix a website problem. It became infrastructure for a new product.
One content model, two digital products.
Impact & Metrics
The framework became the foundation for how waterdrop® scales content globally. But more than the system itself, the lasting impact was a team that no longer needed design to move.
The same 3-week process, before and after. Requirements that used to pile up in week three are now defined before the work begins.
€300k/year in recovered capacity
Calculated with the CTO based on assets produced but never shipped, and hours lost to rework across teams.
50% reduction in production lead times
Requirements predefined and templates in place before the launch clock started.
Design off the critical path
Launch ownership shifted to product, brand, and marketing. Teams plan and execute without waiting on bespoke design work.
25+ custom content sections
Enough flexibility to assemble any page, within a structure that stays consistent across 22 markets.
Reflections
The hardest part wasn't the design. It was holding the line on shorter pages when every stakeholder's instinct was to add more content. In a DTC company, more always feels safer. Convincing teams to trust the data over their instincts took longer than building the system itself.
If I were starting again, I'd establish the governance model earlier: who can add new sections, what the approval process looks like, and what happens when a new market needs an exception. The system worked, but it relied on me being in the room to protect it. That's a fragility I'd design out from the start.
What it taught me: the most durable design work is often invisible. Nobody sees the naming convention or the Metaobject structure. They just notice that launches stopped being chaos.
©2026 Teodora Gheorghe