EvaluationAdvanced

Is version B actually better?

“How would you determine whether version B of an agent is actually better than version A?”

What this tests

  • Understanding that a single demo run is not evidence for probabilistic systems
  • Golden datasets, metrics, repeated runs, and statistical thresholds
  • Multi-dimensional comparison: quality, cost, latency, safety
  • Awareness of judge bias, dataset drift, and online validation

Answers by level

Read the beginner answer first and notice what is missing.

Because outputs vary run to run, I need a golden dataset (representative inputs with expected outcomes or rubrics, including hard cases and past failures), metrics that fit the task (exact match or deterministic checks where possible, LLM-as-judge with a rubric where not, plus tool-selection accuracy and task completion), and enough runs to separate signal from noise: several runs per input, reporting mean and variance, not a single pass. See Evaluating Agents: Testing Probabilistic Systems, Golden Datasets, and Eval Metrics: What to Measure and How.

Then compare on more than quality: cost per successful task, p50/p95 latency, steps per task, error and cap-hit rates, and safety evals (injection resistance, destructive-call rate). B that is 3% more accurate and twice as slow may not be better for the product. Regression tests catch cases A handled that B breaks; the aggregate can rise while specific important cases fail. See Regression Gates and Online Evaluation.

Finally, validate online: a shadow or A/B rollout with the same metrics plus user signals (thumbs, escalation rate, task abandonment), because the golden set never fully matches production traffic.

Green flags · Red flags

Green flags
  • Golden dataset with hard cases and past failures
  • Multiple runs per input, variance reported, statistical threshold
  • Guardrail metrics: cost, latency, safety, cap-hit rate
  • Slices by category and regression checks on specific cases
  • Calibrated judge, deterministic evaluators preferred
  • Online validation via shadow or A/B
Red flags
  • Compares a handful of single runs
  • Uses an uncalibrated LLM judge as the sole arbiter
  • Only looks at accuracy
  • No held-out data or awareness of tuning on the test set

Follow-up questions

F1
B wins 54 to 46 on 100 items. Ship it?
F2
How do you know your judge is trustworthy?
F3
What goes into the golden set?

Practical scenario

A teammate swapped the model powering your support agent and reports "it feels much better" after trying ten conversations. Leadership wants to ship tomorrow. Design the evaluation you would run in the next 24 hours, name the metrics and thresholds, and say what result would make you block the release.

Related concepts · Learn this topic