Are the points necessarily real prices?

No; they may represent a research budget, and the interpretation must be stated.

Is SIMALTO the same as a simple importance-rating survey?

No; constrained trade-offs are central to its purpose.

Can it predict sales automatically?

No; a forecast requires an explicit model and validation beyond raw survey choices.

Must every respondent choose the same package?

No; differences across users or customer segments can be part of the useful finding.

AI Glossary / Statistics and probability / Intermediate

Simalto

Definition

SIMALTO is a market-research approach that uses a grid of product or service attributes and levels to explore customer priorities under trade-offs. The name is commonly expanded as Simultaneous Multi-Attribute Level Trade-Off. Respondents identify current or desired levels and allocate limited points or another stated budget to improvements. The exercise reveals priorities within the choices and constraints presented. It is a structured preference-elicitation method, not a guarantee of actual purchasing behavior or a single universal statistical test.

What it means

Customers may say that every improvement is important when no constraint is imposed. A SIMALTO-style task makes the constraint visible: choosing one improvement can leave fewer resources for another. The grid translates broad wishes into specific service or product levels, such as response times or reporting frequency. The resulting choices describe trade-offs within that scenario. Their interpretation depends on whether respondents understand the levels, regard the alternatives as realistic and treat the point budget as meaningful.

How it works

Define distinct attributes and ordered levels that respondents can understand. Establish a baseline and assign point costs to moving between levels, with clear instructions about the available budget. Ask respondents to indicate current experience, desired levels or constrained choices as required by the study. Record the choices and explore patterns across relevant groups. More elaborate applications may model alternative packages, but the analysis must state its assumptions. Pilot the grid to detect confusing wording, unrealistic combinations and tasks that exceed respondents' attention.

Three practical examples

Example 1: A business service provider asks clients to trade faster responses against richer reports and additional training. Example 2: An AI software team explores priorities among support, integration and administrative features within an illustrative improvement budget. Example 3: A facilities team asks users to allocate points among clearly described service upgrades. These exercises can reveal preferences among the offered options. They do not establish the value of an omitted feature or ensure that respondents would make identical choices when paying real money.

Think of it this way

Think of shopping for improvements with a limited basket of tokens. Each upgrade uses some tokens, so selecting everything is impossible. The useful information comes from what someone gives up as well as what they choose. However, the exercise is shaped by the token prices and available shelves. Different costs, missing options or unclear descriptions can change the apparent priorities, so the questionnaire design is part of the evidence.

Worked example

Consider a simplified illustrative grid with three optional upgrades from a common baseline: faster responses cost four points, richer reports cost three and extra training costs two. Each respondent receives five points. Two respondents choose faster responses, spending four each; three choose richer reports plus training, spending all five. In this five-person example, response upgrades are selected by 40%, while reports and training are each selected by 60%. All three upgrades would cost nine points and exceed the budget. A package combining responses and reports would cost seven, also infeasible. These counts describe choices under the stated point costs; they are not independent willingness-to-pay estimates or a statistically validated market forecast. The report-and-training package attracts more selections here, but five illustrative respondents cannot establish that it is the best commercial offer.

Illustrative case study

In this fictional case study, Willow Service Cloud plans improvements to an AI document platform. Interviews initially produce a long list of highly rated requests. Researchers build a small trade-off grid, test its wording with users and then ask a broader sample to choose constrained improvement packages. They find different priorities among administrators and occasional users. The team uses those patterns to design further prototypes and real-world tests, while documenting the hypothetical budget and avoiding a claim that survey choices directly predict future revenue.

Common misconceptions and mistakes

Common mistakes include presenting the exercise as a direct measure of actual demand, assigning arbitrary point costs without considering their influence and combining attributes whose meanings overlap. Another error is treating selection percentages as independent importance weights when choices were constrained and interdependent. The method is related to trade-off and conjoint research but should not be assumed to use the same task structure or estimation model as every conjoint design. Small or selective samples also limit generalization.

Business use

Businesses use SIMALTO to discuss product development and service priorities when resources are limited. It can make customer conversations more concrete by showing which improvement levels compete for the same budget. Good reporting includes the grid, baseline, point costs, respondent population and analysis method. AI product teams should distinguish user preferences from technical feasibility, delivery cost and measurable outcomes. A preferred package may still be expensive, unreliable or incompatible with other requirements, so preference research informs rather than replaces product judgment.

Check your understanding

Which two upgrades can be combined within five points in the example?

Answer: Richer reports and extra training, costing three plus two.

What fraction selects faster responses?

Answer: Two of five respondents, or 40%.

Why can the three selection percentages not be treated as separate unconstrained purchase probabilities?

Answer: They arise from linked choices within one limited budget.

Does a nine-point package satisfy the five-point constraint?

Answer: No.

Related terms

Conjoint analysis
Related terms include trade-off analysis
stated preference
attribute level
willingness to pay
choice modelling
market segmentation and product configuration

Learn next

Learn next how to write clear attribute levels and pilot a choice task. Compare unconstrained ratings with budget-constrained choices, then examine how hypothetical preferences differ from observed behavior. Study sampling and uncertainty before interpreting subgroup differences, and test whether conclusions remain sensible when plausible point costs or package constraints change.

Sources

Sources: B2B International, What is SIMALTO?, https://www.b2binternational.com/experience/methods/faq/simalto/; B2B International, Pricing Research: How to Assess What Our Customers Value, https://www.b2binternational.com/publications/price-research/amp/. These practitioner sources describe their use of the method. The token grid, respondent counts and fictional software project are original explanatory examples, not evidence of predictive accuracy or a commercial recommendation.

Previous: Silhouette Coefficient Next: Similarity