Quality Attribute Scenarios

Performance Efficiency

The support bot must quickly answer user questions, even during peak load hours.

Attribute	Performance Efficiency
Source	Multiple simultaneous users
Stimulus	200 concurrent users ask questions at one time
Artifact	MARS Bot handling requests
Environment	Production environment during peak usage period
Response	The system processes all questions without significant delay and maintains responsiveness
Response Measure	Average response time for bot's answer is under 10 seconds, with no timeouts or failures

Test: Perform load testing — simulate 200 simultaneous requests to the bot and measure the average response time and error rate.

The user should receive a response almost instantly.

Attribute	Performance Efficiency: Time-behavior
Source	User
Stimulus	User sends a question to the bot
Artifact	MARS Bot
Environment	Production
Response	Bot provides an answer to the user
Response Measure	95% of responses are delivered within 5 seconds

Test:
Use logging to measure latency for each request, build a graph, and make sure that 95% fit within 5 seconds.

The user should receive relevant and correct answers. Any incorrect or "broken" answer reduces the credibility of the system.

Attribute	Functional Suitability: Functional Correctness
Source	User
Stimulus	User asks a supported question to the bot
Artifact	MARS Bot
Environment	Production
Response	Bot provides the correct and expected answer
Response Measure	At least 98% of test questions (from the reference set) are answered correctly

Test:
Create a test set of question-answers and run these questions through the bot, calculate the percentage of correct answers.