Survey Simulator — Pre-field QA for Quotas & Routing

What the simulator solves

Four classes of launch-day bug, verified cheaply against the real engine:

Quota over-shoot — concurrent completes hitting a cell past its target.
Broken routing — every meaningful respondent type exercised, not just a hand-clicked few.
Unreachable segments — target cells that no respondent path can actually hit.
Export surprises — the analyst's day-one view, verified today.

→ Same engine as production

The simulator runs your survey against the same engine live respondents will hit — not a mock. What it catches is what live traffic would hit.

Two run modes

Logic-aware — does the routing work

Draws respect every if: condition, skip pattern, and cross-question dependency. Options are weighted by how empty the quota cell they feed is, so cells fill evenly without hand-tuned percentages — but every option keeps a floor so screen-out branches still get exercised.

Scenario — reach a specific segment

Pin specific answers to specific questions and verify how many respondents reach a target segment or quota cell. Everything you don't pin still uses logic-aware weighting. Pins are checked against declared options before anything runs — no phantom answers, no silent fallbacks.

The reports you get back

Six linked reports per run:

Summary

Completion rate and outcome mix: complete, terminate, exit, quotafull, stuck.

Coverage

Per-question hit counts. "Never reached" markers surface dead branches.

Quota

Live cell fill as the run completes. Edit targets in-place — never touches live config.

Audit

Blank + out-of-range probes verify your required and range validations actually fire.

Weights

Derived %, qualify / quota / novelty, "why" per option. Pin to override.

Issues · Trace · History

Mid-run flags, step-by-step respondent walk, previous-run comparison.

Coverage tab — per-question hit counts across the survey, with 'via script' markers on hidden questions and 'Never reached' flags for branches no respondent took

Test safely — zero risk to live data

Simulator respondents live in a separate test database. A sim run physically cannot touch live data; a live export physically cannot pull in sim respondents. Safe to run against a live-fielding survey.

For quotas, the simulator fires parallel completes against the real quota-engine locks and verifies cells hit target exactly — not a mocked lock, the actual production path.

Run modes

Reports per run

Risk to live data

See the exports before you field

Every sim run lands in a Test environment that the Export wizard and Reports read via ?env=test. Same analysis surface the analyst uses after launch, pointed at sim data — download the CSV, open the SPSS .sav, preview the datamap, run a crosstab.

Export wizard with the Test / Live environment toggle — the same export surface the analyst uses after launch, pointed at sim data

For CBC and MaxDiff pre-fielding analysis — Design Health and Precision scorecards, D-efficiency, implied-N — see Choice modeling →

Catch problems before you field.