What Is a Constant Sum Question?
A constant sum question asks respondents to distribute a fixed number of points, typically 100, across a set of items to reflect relative importance, preference, or allocation. If you're measuring which product features matter most, a respondent might assign 40 points to reliability, 25 to price, 20 to ease of use, and 15 to design. The constraint that all allocations must total exactly 100 forces trade-offs: giving more to one item means giving less to others. This produces ratio-level data showing not just rank order but the magnitude of differences between items, something that standard rating scales and simple ranking questions can't deliver.
Why Constant Sum Questions Matter
Standard rating scales let respondents rate everything as highly important, which produces flat data with little differentiation. Constant sum questions eliminate this problem by introducing scarcity, there are only 100 points to go around. This mirrors real-world decision-making, where budget, time, and attention are finite resources that must be allocated. The result is a sharper picture of priorities that's directly useful for product development, resource allocation, and strategic planning. When a stakeholder asks "But which features matter most?" a constant sum dataset has a clear answer, while a Likert dataset often doesn't.
How Constant Sum Questions Work
Structure
A typical constant sum question presents 4-8 items with input fields next to each, plus a running total that must equal the target sum:
Allocate 100 points across the following features based on their importance to you when choosing a project management tool. The total must equal 100.
| Feature | Points |
|---|---|
| Task management | ___ |
| Team collaboration | ___ |
| Reporting and analytics | ___ |
| Integrations with other tools | ___ |
| Mobile access | ___ |
| Total | 100 |
The respondent fills in values that sum to 100. The survey platform validates that the total is correct before allowing the respondent to proceed.
What the Data Tells You
Constant sum data provides three levels of insight:
Relative importance. The average allocation across all respondents shows which items are most important overall. If "Task management" averages 28 points and "Mobile access" averages 8, the relative priority is clear.
Magnitude of differences. The gap between items is meaningful. A 28 vs. 22 split suggests two similarly important features. A 35 vs. 5 split shows one feature dramatically outweighs another. Rating scales can't express these magnitudes.
Segment differences. Comparing allocations across customer segments reveals different priority structures. Enterprise buyers might allocate heavily to reporting and integrations, while small-team users prioritize collaboration and mobile access. These differences become the basis for product tiering and messaging.
Design Guidelines
Number of items. Keep it between 4 and 8. Fewer than 4 doesn't produce enough differentiation. More than 8 overwhelms respondents, the cognitive load of distributing 100 points across 12 items while maintaining a running total is substantial. If you need to evaluate more items, split them into separate constant sum questions or use MaxDiff instead.
Point total. 100 is the standard because percentages are intuitive. Some researchers use 10 for simplicity, but smaller totals reduce the resolution of allocations and force larger minimum increments. Stick with 100 unless your audience would struggle with the math.
Item order. Randomize the order for each respondent. Items listed first tend to receive higher allocations (a primacy effect), and randomization distributes this bias evenly.
Clear instructions. Explicitly state that points must total 100, that higher numbers mean greater importance, and that respondents can assign 0 to items that don't matter. Many respondents are unfamiliar with the format, so clear instructions prevent confusion and reduce errors.
Constant Sum vs. MaxDiff
Both formats measure relative importance through forced trade-offs, but they work differently:
| Aspect | Constant Sum | MaxDiff |
|---|---|---|
| Respondent task | Distribute points | Choose best/worst from sets |
| Output | Point allocations (ratio data) | Preference scores (interval data) |
| Number of items | 4-8 optimal | 8-20+ works well |
| Cognitive load | High (math involved) | Lower (simple choices) |
| Ties allowed | Yes (equal allocations) | No (forced differentiation) |
| Best for | Budget-style allocations, small item sets | Large item sets, ranking without numbers |
Use constant sum when you have a small item set and want respondents to think in terms of allocation or budgeting. Use MaxDiff when you have a larger set of items and want a clean rank order without burdening respondents with arithmetic.
Analysis
Calculate the mean allocation per item across all respondents to see overall priorities. Use standard deviations to identify items with high consensus (low SD) vs. Items that are important to some segments but not others (high SD). Compare means across segments using ANOVA or t-tests. Because the data sums to 100 for each respondent, the allocations are compositionally dependent, an increase in one item mathematically decreases others. This doesn't invalidate standard analyses for most practical purposes, but it's worth noting in methodological documentation.
When to Use Constant Sum Questions
- Feature prioritization where you need to know not just what matters but how much more one feature matters than another
- Budget allocation exercises where you want respondents to simulate how they'd distribute limited resources
- Benefit testing to determine which product benefits resonate most strongly relative to each other
- Time allocation studies to understand how professionals distribute their work hours across activities
Common Mistakes
- Including too many items (10+), which turns the question into a frustrating math exercise and increases the rate of errors and abandonment
- Not validating the total in the survey platform, allowing respondents to submit allocations that don't add up, this data is unusable
- Forgetting to randomize item order, which lets primacy effects inflate allocations for items listed first
How Quali-Fi Supports Constant Sum Questions
Quali-Fi's survey builder includes a dedicated constant sum question type with real-time total validation, item randomization, and configurable point totals across all plan tiers starting at $89/month. The platform's analysis dashboard displays mean allocations by item with segment comparisons and exports the raw data for advanced statistical analysis.
Frequently Asked Questions
Can respondents assign zero points to an item?
Yes, and they should be told they can. Zero is a valid allocation that means the item has no importance to that respondent. Without explicitly stating this, some respondents feel obligated to give at least a few points to every item, which dilutes the differentiation the format is designed to produce.
What if respondents find the math too difficult?
Some respondents will struggle with the arithmetic, especially on mobile. Use a survey platform that provides a running total and prevents submission until the total is correct. If your audience skews less numerate, consider MaxDiff as an alternative that produces similar insights without the math.
How do I handle ties in the data?
Ties (equal allocations to multiple items) are valid and informative, they indicate the respondent truly views those items as equally important. Report them as-is. If you need forced differentiation, MaxDiff is the better format.
Related Topics
Force the trade-offs that flat rating scales hide. Start a free trial of Quali-Fi Surveys and use constant sum questions with real-time validation, randomization, and segment analysis to uncover what really matters to your audience.