Troubleshooting guide

Overview

Use this guide when a node formula returns unexpected results, for example:

• empty output

• wrong granularity

• wrong numbers

The goal is to identify where the issue starts and fix it in a structured way.

Before you start

Make sure you can:

• open the Model Editor

• edit the node formula in the Function Editor

• inspect the result in Data Preview

Troubleshooting steps

1. Fix syntax or validation errors first

If the editor shows a validation or syntax error, resolve that before troubleshooting the formula logic.

After fixing the error, check the result again in Data Preview.

2. Define the expected result

Before changing the formula, note:

• one specific intersection you expect to be correct

• the expected output shape, including dimensions and levels

This gives you a clear reference while debugging.

3. Identify the type of problem

Use Data Preview to decide which kind of issue you are seeing:

• Empty output: values are missing where you expect data

• Wrong shape: the result is too aggregated or too granular

• Wrong numbers: the shape looks correct, but the values are wrong

4. Simplify the formula

Start with the smallest possible version of the formula:

• begin with a single node reference, for example 'Sales'

• then add the remaining parts back one by one

If the formula is large, move parts into helper nodes so you can inspect intermediate results more easily.

5. If the shape is wrong, check dimensionality first

Check whether the connected nodes or data sources have the dimensionality you expect.

If the result is too aggregated:

• check whether + or - aggregated the inputs to their common dimensions

• use ROLLUP or ROLLUP_TO if you want to make aggregation explicit

If the result is too granular:

• check whether * or / expanded the result across the combined dimension set

• check whether functions such as EXPAND or EXPANDSINGLE introduced additional detail

6. If the output is empty, check filters and missing intersections

If you use FILTER(...), confirm that:

• dimension names match exactly

• level values match exactly

• the intended level is being filtered

If values disappear after comparisons, filtering, or join-like logic, check whether one input is missing values where the other input is defined.

7. If the numbers are wrong, check arithmetic and assumptions

If the output shape is correct but the values are wrong, check:

• whether constants such as + 1 or - 100000 behave as intended

• whether you should use ADDEACH(...) for element-wise adjustments

• whether assumptions or scenarios affect the result upstream

• whether you need BASELINE(...) or NONSIM(...) for comparison

• whether operator order changes the result

8. Validate with spot checks

Pick at least two representative intersections and confirm:

• the shape is correct

• the values match your expectation

This helps avoid fixing one case while missing another.

9. Clean up the final formula

Once the result is correct:

• split complex logic into helper nodes if that improves readability

• add short # comments where they clarify intent

Still need help?

Contact Valsight Customer Support: support@valsight.com

Related sections

• Formula basics: syntax rules, constants, and how formulas evaluate

• Operators: arithmetic, comparisons, and alignment behavior

• Function catalog: A–Z index and categories for detailed behavior

• Browse by category: choose the right function family by intent