Understand what drives outcomes

When a client asks "why?" or a score moves between waves, AddMaple has several tools for driver work. This guide maps each one to the job it does, so you can pick the right starting point.

Pick your outcome column, then use the table at the bottom to choose where to start.

Related columns for discovery

Related Columns scans the dataset, tests relationships with the appropriate method for each column type, and orders results by significance and effect size.

Use it when you have an outcome column and want to see what else is associated with it—or when you are early in analysis and do not yet know which cuts will matter. Click a related column to open a pivot chart and inspect the relationship.

Significance testing for segment differences

Significance Testing highlights cells meaningfully higher or lower than expected in a pivot table.

Use it when you already have a comparison in mind—NPS groups across experience attributes, regions or customer types against the average, or any table where you want z-score shading on segment differences.

Key driver analysis to rank what matters

Key Driver Analysis ranks candidate drivers across mixed survey data—numbers, opinion scales, categories, and multi-select columns.

Use it when the question is "which factors appear to matter most when considered together?" Before running the model, remove columns that duplicate the outcome or reveal the answer directly—for example, do not use "NPS category" as a driver of "NPS score."

Regression or correlation for focused relationships

Correlation compares two numeric columns (Pearson or Spearman, chosen automatically). Regression models a numeric or binary outcome with a small set of variables.

If you have many possible drivers, start with Related Columns or Key Driver Analysis, then use regression or correlation to inspect the strongest relationships more closely.

Clustering for respondent segments

Clustering groups respondents who answer similarly across several columns—useful when the question is "what natural segments exist?" rather than "what is associated with this one outcome?"

Use it after you have cleaned scales and key banner fields, when you want personas or segments for targeting, or to sanity-check whether significance patterns align with broader group structure. For method choice (K-Means, HDBSCAN, and more), see Clustering methods explained.

NPS as an outcome

NPS is often the outcome teams want to explain. AddMaple calculates NPS automatically for suitable 0–10 opinion scale columns.

A typical workflow in AddMaple:

1. Confirm the NPS column is detected correctly.
2. Use Related Columns to find what is associated with NPS or NPS group.
3. Pivot the strongest relationships and turn on Significance Testing where useful.
4. Run Key Driver Analysis if several candidate drivers compete.
5. Filter coded open-text themes for detractors or promoters—see Connect open-text feedback with scores, segments, and outcomes.

Which tool should I use?

Deeper dives on individual tests

AddMaple picks the right test automatically in Related Columns and pivot workflows. When you want the full definition, formulas, or step-by-step UI for one method, use these guides—or see How does AddMaple perform statistical calculations? for the full map.

• Chi-Square test
• T-Test
• ANOVA
• Kruskal-Wallis
• Correlation
• Clustering

See also

• Exploring your data
• Connect open-text feedback with scores, segments, and outcomes
• From survey file to client-ready findings