Starting your first network design project? Here are some best practices to help keep things smooth and get to results faster.
1. Start with a Clear Scope
The foundation of your project is defining the scope. Once you’ve nailed that down, it becomes easier to identify the key information you’ll need later.
How do you define the scope?
Write down the business question—what is the main issue you’re trying to solve? This will help you narrow things down: geography, product types, locations, and transport modes.
Why does this matter?
Because you only want to use the data you actually need. It’s easy to get overwhelmed with all the available data, but more isn’t always better. Being clear on what’s needed (and where to find it) makes data cleaning and transformation faster—and keeps you from wasting time on unused data.
2. Define Your Scenarios
Scenarios are just different versions of how things could work, depending on the decisions you’re trying to test. Once you define them, it becomes easier to figure out what data you’ll need.
In most cases, you’ll want one year of transactional data—basically, how products move from suppliers to demand points. It’s also important to know the physical location of all supply chain points (suppliers, factories, warehouses, customers). Don’t worry about having exact coordinates—zip codes or city names are fine; SCN Navigator can fill in the rest.
3. Start Small, Then Scale
You don’t need to solve everything with your first model. It’s better to define a manageable scope and deliver value quickly. You can always build on it later.
4. Think About Model Size and Granularity
Let’s say your business question is about secondary transport (delivery to customers)—then there’s no need to model your suppliers or factories in detail. This kind of analysis should be done for each scenario to make sure you’re setting the right level of detail.
Remember:
Optimization means finding the minimum or maximum value or in a mathematical model. The smaller and more focused the model, the faster it solves. But you still want enough detail to reflect your supply chain accurately.
Rule of thumb: Keep your data as simple as possible, without compromising the quality of your answers.
Don’t Forget the Limits
Clustering or aggregating elements can dramatically reduce solve time, often with little to no impact on the results.
There are two practical limits to keep in mind:
• Excel threshold: 1 million rows, if you’re not directly integrated with the backend.
• Solver threshold: 50 million “rule.” This is the product of:
Number of Products × Periods × Warehouses × Customers × Modes of Transport.
Stay under this limit to keep your model efficient. Depending on your business question, some of these will need more granularity than others.
Tips for Simplifying Your Data
• Remove noise: ERP data often includes returns, samples, or small volumes. Focus on transactions that reflect actual physical flows.
• Aggregate products: Group products into a manageable number of “model products” that still reflect the complexity you need. Use ERP product families if possible, or create modelling groups that make sense for your project.
• Aggregate demand nodes: Use city or zip code-level data.
• Use proxy assumptions for capacity: Historical data can work well.
• Simplify transport rates: Average out freight costs to reflect directional differences without going too granular.
5. Start optimizing!
Have fun using all the capabilities of SCN and start running, comparing and creating new and more fun scenarios!