Product Designer · B2B SaaS · Compliance · Data visualization · 2024–2025 · 8 min read
2-minute version ↓Risk Graph
Finding the risk three steps away
Turned a network of 287,000 connected companies into a few clear questions, and one path a compliance team can act on.
Impact at a glance
- Became the product’s signature feature
- Given its own section on the Syenah website
- Its question-first idea spread to the shipment analysis view
One company’s network
Volkswagen, the demo company in the product
- Tier 1612
- Up to Tier 220,338
- Up to Tier 3287,335
What people needed
One path
- Role
- Product Designer, the only designer on the team
- What I did
- User research, problem framing, interaction design, data visualization, UI, prototyping, UX writing, developer handoff
- Team
- Co-founders · knowledge graph architect · AI/NLP engineers · frontend and backend developers
- Timeline
- 2024–2025
- Platform
- Web (desktop)
The 2-minute version
Only designer · Research → UI → handoff
01The ask
“Show the whole supply chain as a network.”
The founders wanted the graph to be the wow moment in demos.
02The digging
I watched compliance teams check their suppliers.
They didn’t want to see everything. They wanted to know if one thing could hurt them.
03The turn
Key insightVersion 1 drew everything. Nobody could use it.
Tier 1 suppliers alone were 292 circles on one screen.
04The idea
A graph you ask questions.
Start with one company. Ask “Any sanctioned companies up to Tier 3?” Get a short list, and one path.
05The build
Counts before clicks.
Filters in compliance language, and a path that explains itself.
06The result
Featured as its own section on the Syenah website.
It was the most distinctive part of the product.
Context
What is Risk Sentinel?
Risk Sentinel was Syenah’s risk platform for companies, banks and asset managers. Every day it reads news, NGO reports and social media, finds incidents linked to companies, and scores how serious they are. It also maps how companies are connected: who owns whom, who supplies whom, who invests in whom.
Key terms in this case study
- Tier 1, 2, 3
- Companies directly connected to yours, the companies connected to them, and the level after that. In a supply chain: your suppliers, their suppliers, and so on.
- Sanctioned company
- A company on a government sanctions list.
- LkSG
- The German Supply Chain Act. Large companies in Germany must check human rights and environmental risks in their supply chain.
- Path
- The chain of relationships that connects two companies.
Screens show Volkswagen because it was a demo company in the product. It was not a client.
The setup
The moment this design is for
Monday morning · A compliance manager at a large German manufacturer
An NGO report mentions a sanctioned company. Her manager asks one question:
“Are we connected to them?”
Her direct suppliers are in a spreadsheet, almost 300 of them. That company isn’t on it. But under the LkSG, a report like this means she has to look further than Tier 1. For a manufacturer the size of Volkswagen, counting every kind of relationship, that’s 612 companies at Tier 1, about 20,000 up to Tier 2, and almost 290,000 up to Tier 3.
She doesn’t need a map of all of them. She needs a yes or no, and if it’s yes, the exact chain of companies, so she can document it.
- Tier 1 · 612 companies · her supplier spreadsheet sits in this ring
- Up to Tier 2 · 20,338 companies
- Up to Tier 3 · 287,335 companies
- The sanctioned company from the report
Not to scale. Counts are for Volkswagen, the demo company in the product, and include every kind of relationship.
Ownership
My contribution
Hands-on design
I ran the interviews, framed the problem, and designed the graph interactions, menus, filters, side panels and empty states. I wrote the interface copy and built clickable prototypes for testing.
Leading without a product manager
I also:
- — turned the founders’ vision into concrete requirements
- — worked with the knowledge graph architect on which questions the graph could answer quickly
- — defined limits and edge cases with developers, like how many nodes a screen can show
Research
People didn't want to explore. They wanted an answer.
The founders’ brief was clear: competitors only show Tier 1, so we show the whole network.
Before designing it, I wanted to see how people actually check a supply chain. I talked with 7 people:
3 compliance and sustainability managers at companies covered by the LkSG
2 procurement risk managers
1 ESG analyst at an asset manager
1 sustainability consultant who helps companies prepare for audits
I asked each of them to walk me through a real case:
“A news report links a company to forced labor. How do you find out if it’s in your supply chain?”
Three things came up again and again:
What I saw: Everyone started from one company, never from “the whole network”
What it changed: The graph starts empty, with one search
What I saw: They asked the same few questions every time: anything sanctioned? who’s riskiest? how are we connected?
What it changed: Ready-made questions instead of building filters
What I saw: Finding the risk wasn’t enough. They had to show auditors how the companies were connected
What it changed: The path became the main output, not the network
So I changed the goal
How might we help a compliance team go from almost 290,000 connected companies to the one path that matters, and show how it connects back to them?
The pivot
Version 1 drew everything
Version 1
Version 1 followed the brief. Search a company, and the graph expands its full network.
In testing
In testing, it fell apart fast. Expanding just one relationship type, Volkswagen’s 292 Tier 1 suppliers, drew a ring of 292 circles around it. Sanctioned companies were in there, but people had to zoom in and hunt for them. Nobody went on to Tier 2. Several people said it looked impressive, then asked what they were supposed to do with it.
The flip
The problem wasn’t the visual style. A full network answers no question. So I flipped the model: instead of showing everything and letting people search it, people ask a question and the graph shows only the answer.
Nodes on screen
293
Design decisions
Counts before clicks, filters in compliance language, and a path that explains itself
Start with one company
Problem
Decision
Why
Questions, not filters
Problem
Decision
Right-click any company and choose from ready-made questions, grouped by depth:
Tier 1
- Sanctioned companies
- 10 riskiest companies
- 10 companies with the most incidents
Up to Tier 3
- The same three, plus
- 5 riskiest paths
- 5 most connected companies
The answer opens as a ranked list in a side panel. Each row has its own “find shortest path” button.
Result
Show the count before you click
Problem
Decision
Every expand option shows its size first. For bigger questions, Advanced Expand shows the size of each depth, a live result count that updates as filters change, and a clear display limit of 5,000 companies.
Companies with no incidents are hidden by default. That alone cut Volkswagen’s 612 direct relations to 68 worth looking at.
Why
Expand
Adds 612 companies to the graph
Advanced Expand
Within the display limitLimit 5,000
Filters in compliance language
Problem
Decision
Why
The path is the evidence
Problem
Decision
- Path finding between any two companies, with a swap button
- Labeled lines on every step, like “Supplier” or “Current investment”
- One color per relationship type, with a legend that can expand into subtypes, for example Supplier and Vendor
- Sanctioned companies are marked with a clear stamp, visible at any zoom level
Result
Action · 5 riskiest paths
- VolkswagenStart
- Qatar HoldingInvestor
- VTBBankSanctioned
- VTB subsidiarySubsidiarySanctioned
Details without leaving the graph
Problem
Decision
Company details open in a side panel:
- — who sanctioned the company
- — risk score and number of incidents
- — supply chain act violations
- — recent incidents and top risk factors
A link opens the full profile when needed.
Why
Small things that made exploring feel safe
- Undo · Redo
Undo and redo, so people dared to try things
- Delete
Delete a node from the graph without losing the rest
- 6 layouts
Six layouts, from force-directed to hierarchical, for different questions
- Badges
Badges that switch between risk score, incident count and number of connections
- Nodes on screen
A “nodes on screen” counter, so people always knew how much they were looking at
Testing
Testing the second version
I tested the question-first version with 5 of the original participants, using the same task: “Are we connected to a sanctioned company?”
5 of 5 found an answer
2 of 5 asked to export the path as a report
This time everyone found an answer. Most started with the sanctioned-companies question, then used the path to understand the connection. Two people asked to export the path as a report. That became the top item for the next release.
Outcome
Impact
The graph became the product’s signature feature. It was given its own section on the Syenah website: interactive supply chain analytics with one-click actions, path finding and filtering.
From a network to an answer. In testing, people went from “impressive but confusing” to finding a sanctioned connection in a few clicks.
The question-first idea spread. The shipment analysis view follows the same rule: it starts with only the main company and lets people add more with filters.
Risk Sentinel didn’t find enough paying customers, and Syenah moved on to a new product, SyFin. The graph didn’t save the business. But it taught me what I took into my next AI product.
Reflection
What this project taught me
A big dataset isn’t a feature. A clear answer is. The value wasn’t in showing 287,000 companies, but in hiding almost all of them.
Show the cost before the click. Counts and limits made a heavy tool feel safe to use.
If I continued this work...
- Export a path as an audit-ready report
- Make Tier 3 questions faster, or show progress while they run
- Let people save an investigation and return to it later