When a numeric question lands in a Halo Reports Crosstab, the default view is a mean. For many analyses, that's exactly what you want: a single summary number you can compare across columns. But sometimes the mean isn't the answer.
A mean of 7.4 on a satisfaction scale tells you the average, but it doesn't tell you whether most respondents are clustered at 8 or whether you have a polarized split between Promoters and Detractors. A mean engine displacement of 2.6 liters tells you nothing about whether the sample skews toward small sedans or is evenly spread across vehicle classes.
When the shape of the data matters more than the average, Halo Reports lets you display a numeric question as a percentage distribution to show the share of respondents at each response value, not just a single summary statistic. This article walks through how, when, and what to watch for.
When the Distribution Tells You More Than the Mean
Treating a numeric question as categorical responses is especially useful when:
You want to break down responses by value rather than compute a mean.
You're working with a numeric scale that represents selection, for example, a satisfaction or recommendation rating from 1 to 5 or 0 to 10.
You want to define Top Box or grouped response summaries (e.g., Top 2 Box analysis).
The third use case will be familiar to readers of an earlier issue on Net Promoter Score. NPS is, by definition, a Top/Bottom Box framework applied to a 0-10 recommendation scale. The technique in this article is the broader capability that makes NPS-style analysis possible across any rating scale or numeric question in your study.
How to Display Numeric Questions as Percentages in Halo Reports
Step 1.
Add a numeric question to your query. Numeric questions are identified by the ruler icon in the Questions list. By default, they're placed with a Mean statistic description, visible next to the question, and in the Results panel.
Step 2.
Right-click on the numeric question in the Query panel (or use the dropdown caret) to open its menu. Under the Commands section, select "Use as category variable."
Step 3.
The Results panel now displays each response value as its own row, with the default statistic "% of column, sample base," meaning each value's share of the total respondents who answered the question. A "Sample total" row appears at the top, confirming the 100% summation alongside the unweighted count.
A useful note on reversibility: per Halo Reports' documentation, a numeric question converted to categorical can be converted back to numeric, as long as it remains in a panel that accepts numeric calculations (marked with the Σ icon). Practically, that means you can experiment freely, including by transposing the table, without worrying about a one-way change.
Three Core Truths When Working With Percentage Distributions
1. Know your base.
Halo Reports' default percentage statistic is "% of column, sample base." Each value is divided by the total number of respondents who answered the question. That's the right base for most distribution views, but if you're filtering on a subset or comparing to a different denominator, verify the base matches the question you're actually asking before sharing results.
2. Percentages don't equal precision.
A clean-looking percentage from a tiny base can be wildly noisy. A 67% share from a base of three respondents is not the same statistic as a 67% share from a base of 300. Treat percentages on small subgroups with the same skepticism you'd apply to mean scores with small sample sizes.
3. The mean and the distribution answer different questions.
When you need a single summary number to compare across groups, the mean is often the right view. When you need to understand the shape of responses (e.g., clusters, skew, polarization, Top Box concentration), the distribution is.
Beyond the Display: Putting Distribution Views to Work
Once your numeric question is rendering as a distribution, a few follow-on moves pay off consistently:
Build Top and Bottom Box summaries from rating-scale questions to track customer satisfaction or brand affinity over time. NPS is one specific example; the same logic applies to any positive-rating threshold in your study.
Compare distributions across segments. Once you've broken a numeric question down by value, querying it by demographics, geographies, or behavioral segments reveals where the distribution shifts, not just whether the mean does.
Spot polarization that the mean would hide. A symmetric distribution and a bimodal split can produce identical means. The distribution view tells you which one you're looking at and whether your "average customer" actually exists or is a statistical artifact between two real groups.
With these habits and the percentage display in Halo Reports, your numeric questions stop reporting averages and start telling the underlying story.
📚 Related Reading
Customer Loyalty: Calculating a Net Promoter Score (NPS) in Halo Reports
The single question that defines customer loyalty, and how to calculate it in Halo Reports.