Chelsi Teo

Recipe Customization 


Redesigning HelloFresh’s global system for scalable personalization

— SCOPE


CPS (Culinary Planning Service)

MPS (Menu Planning System)

CCM (Customer Content Manager)


— DESIGN TEAM


Lead Product / Service

Designer

+ Junior Visual Designer

HelloFresh customers want variety such as protein or ingredient swaps, not a single fixed option per recipe. Scaling that sounds straightforward, until you try to support it across markets, internal tools, and operational constraints!


I led the UX and service design for a cross-tool revamp of how recipe customization is created, planned, and changed. By introducing a unified model for base recipes, variants, and versions, we turned customization from a fragile workaround into a scalable system capability, leading to:


3–5 variants per base recipe

100% customizable recipes

€34–35M CVA enabled

20–30 hrs/month manual ops saved

13 markets rollout

THE PROBLEM (And why it was bigger than it looked!)

Existing

Process


  • Recipe customization technically existed, but it was held together by manual work.
  • To support variants, teams relied on spreadsheets, copied data between tools, and recreated recipes that already existed. Variants were not linked, visibility was low, and changes were risky.
  • As customization scaled, these workarounds broke down, not because teams lacked discipline, but because the system lacked structure.

[Before: spreadsheets maintained variant recipes + manual workarounds]

My Approach (Discovery & Alignment)

Service Blueprints

I led cross-market workshops and mapped the end-to-end recipe journey using service blueprints. This made one thing clear: most “errors” were not user mistakes, but gaps in how recipes and variants were represented in the system.

Running a workshop across all markets to align on current processes.

What Became Clear (The real problem)

Peeling another

layer of the onion

The biggest problem was that recipe variants were treated as completely separate recipes. There was no real connection between a parent / base recipe, its variants (e.g. protein swaps) and their operational versions.


Additionally, there was no system logic on the backend that defined these relationships, and markets increasingly created workarounds and relied on memory and Google sheets (our most dreaded enemy in HelloFresh!).

Strategy (Reframing the problem)

Instead of designing “better customization features,” I reframed the challenge as a system-design question: How should our internal tools understand recipes so customization becomes safe, visible, and scalable?

Four Guiding Principles

Define a shared recipe structure




Everyone works from the same recipe foundation instead of creating disconnected copies

Create one universal guided way to create variants


  • No spreadsheets, no ambiguity

Make relationships visible to users



  • Teams should not have to guess which recipes are connected

Let the system do the work



Data should flow automatically across tools

The Solution (What Changed)

Guided Variant

Recipe Creation




Before, creating a variant meant leaving the product – referencing spreadsheets, copying SKUs, and maintaining unlinked recipe records.


We moved variant creation directly into the recipe workflow.


Teams can now create variants from a base recipe, define the customization type (swap / add / double), and perform ingredient changes inside the system — saving variants as linked entities rather than copies.


Result: Variants are created once, intentionally, and correctly, eliminating duplication and hours of manual coordination.




Clear Recipe Hierarchy



  • Previously, base recipes, variants, and versions existed as isolated records, making it difficult to know what already existed or how options were related.

  • We introduced an explicit recipe hierarchy:
  • Base recipe → Variants → Planned versions

  • This made relationships visible in the workflow. Teams can now see which variants already exist, how recipes relate across weeks, and what is live, planned, or unscheduled.

  • Result: Reduced duplication, cleaner and more accurate data, faster menu planning, and stronger inputs for demand planning and forecasting


Decisive Change Management



Updating recipes with variants used to be risky. Teams couldn’t clearly see what differed between base and variant, or how changes would propagate.

We introduced guided change management. Teams can now compare base and variant recipes side by side, see exactly which ingredients are swapped.


Explicit update paths replace guesswork:

– Sync all: apply changes across linked variants

– Partial sync: apply changes within a specific context


Result: Teams gained clarity, confidence, and agency, while operational risk dropped significantly.



Recipe

Business Impact

  • 100% of core recipes customizable
  • 3–5 variants per recipe
  • €34–35M CVA enabled
  • 20–30 hrs/month saved per market


Product & User Impact

“Big shoutout to Enterprise UX for this work. These hierarchy visualizations look super sharp even from a ‘noob’ perspective :) Excited to see this power future work!”


  • Principal Product Manager (Active Journeys Alliance)


“…big round of applause from the DACH team to the successful 2in1 rollout of versioning 0.5 and change mgt. Having these means some of the biggest inefficiencies and lack of traceability in menu planning are a thing of the past.”


Sr. Program Manager (Strategic Product Development, DACH)

Awards

  • It’s Nice That
  • AIGA
  • Fonts In Use
  • The Dieline

Contact

email@domain.com

000-000-000


— Instagram

— Twitter

— Facebook

Looking Ahead (What’s next!)

Thinking Beyond

With a unified recipe model now in place, this work will unlock several high-impact next steps:


  • AI-assisted variant creation
  • Automatically generating common variants (e.g. “swap to chicken,” “make it veggie”) based on structured recipe data.

  • Automated recipe step adjustments
  • Reducing manual editorial work when ingredients change, while keeping instructions accurate.

  • Personalized defaults
  • Surfacing the most relevant variant by default based on customer preferences and past behavior.

  • Seasonality-based variant families
  • Allowing recipes to adapt intelligently to ingredient availability and regional context.

This project intentionally focused on building the foundation, enabling future personalization without adding operational complexity.