A prediction tells you what’s likely to happen. A scenario tells you what could happen.
“If we increase the marketing spend by 20%, what happens to conversions?”
“If supplier prices rise 15%, what does that do to our margins?”
“If we hire three more support staff, how does resolution time change?”
These questions are where predictions become actionable. You’re not just forecasting the future—you’re exploring options.
Beyond point predictions
Most ML tools stop at the forecast. You train a model, you get predictions, you’re done.
But decisions require comparison. You need to see:
- What happens under current assumptions (the baseline)
- What happens if you change key inputs (the scenarios)
- How different those outcomes are (the decision space)
Previously in Kanva, exploring scenarios meant preparing new input files for each variation. That worked, but it was tedious. Every “what if?” required a spreadsheet.
Scenario mode
Kanva now lets you define and compare scenarios directly in the interface. No files required.
Here’s the workflow:
1. Start from your trained model
Open any forecasting project with a working model. The scenario builder appears in the prediction section.
2. Create a scenario
Click “New Scenario” and give it a name—“High Growth,” “Conservative,” “Price Increase,” whatever makes sense for your question.
3. Modify driver values
For each scenario, adjust the drivers (features) you want to change. Increase marketing spend. Raise the price. Change the seasonal pattern. You’re telling the model: “Assume this instead.”
4. Compare outcomes
Run predictions for all scenarios at once. Kanva shows them side-by-side: charts, tables, and the differences between them.
What this looks like in practice
Say you’re forecasting monthly sales. Your drivers include:
- Marketing spend
- Product price
- Competitor activity
- Seasonality
You create three scenarios:
| Scenario | Marketing | Price | Assumption |
|---|---|---|---|
| Baseline | $50K | $99 | Current plan |
| Aggressive | $80K | $99 | Increase spend |
| Premium | $50K | $129 | Raise prices |
Run predictions. Now you see three forecast lines, not one. You can answer questions:
- Is the extra $30K in marketing worth it?
- How much volume do we lose if we raise prices?
- Which path gives better margins?
The model does the math. You make the decision.
Building intuition
Scenarios aren’t just for formal planning. They’re also useful for building intuition about your model.
What happens if I double this input? Does the prediction double, or is the relationship non-linear?
What’s the most sensitive driver? Change each one by 10% and see which moves the needle most.
Are there thresholds? Sometimes predictions jump when inputs cross certain values.
Exploration helps you understand what the model learned—and whether it makes sense.
Practical notes
Some details on how scenarios work:
- Scenarios are saved with the project (you can return to them later)
- You can create as many scenarios as you need
- Comparison charts show all scenarios together
- Export includes scenario labels so you can analyze in other tools
- Scenarios work with cross-project predictions too
This feature is available for forecasting projects now.
Try it
If you have a forecasting model, try creating a few scenarios:
- Open your project’s prediction section
- Click “New Scenario”
- Adjust one or two drivers
- Run and compare
The question isn’t just “what will happen?” It’s “what could happen, and which path do we choose?”
Prediction scenarios are available now in forecasting projects. Questions? Reach out at hello@human-driven.ai