Rule Testing
Merchant Center rule testing too slow? Try local simulation
In the official flow, every rule change means save a draft, run Test Rules, then wait 10–20 minutes for the report. One small tweak = one round of waiting. RuleGen moves the “verify” step into the browser: the moment a rule is built, it simulates Before/After on your sample data, catching problems before you submit.
Where the 10–20 minute test loop slows down
Google’s official rule test is asynchronous: after you submit a Test, the report is queued for generation — Google’s docs cite 10–20 minutes. That stretches every “edit rule → wait for report → read result → edit again” loop, especially when you’re iterating across multiple rules and attributes.
Local simulation attacks the cost of verification: it’s a purely deterministic TypeScript engine, no LLM, no network. On your sample data (≤10 rows × 20 cols) it returns row-by-row Before/After and PASS / WARNING / ERROR status in under 10 seconds, so you can judge “is the logic right?” before deciding whether to submit the official test.
| Dimension | Google official test | RuleGen local simulation |
|---|---|---|
| Wait time | 10–20 min (async report) | Instant (in-browser) |
| Google account required | Yes (inside Merchant Center) | No (no Google login) |
| Catch issues before submit | Only after the test | Simulate as soon as the rule is built |
| Edit → retest loop | Wait for a new report each time | Retest instantly after every change |
| Iterate locally | Bound by Google’s pace | Unlimited, zero cost |
Go to the Debugger: diagnose an existing rule
If “the rule is written but the result is wrong”, use the Rule Debugger: paste the existing rule or describe the symptom, and it returns a diagnosis (e.g. “your condition may never match”), likely causes, and a fix suggestion — then back to simulation to verify.
Local simulation + Debugger
Catch problems first, then submit. No Google login, no feed upload, no catalog stored.
Open the builder →Frequently asked questions
Why is Google’s official rule test so slow?
The official flow is: save a draft → run Test Rules → wait for the report. The report is generated asynchronously; Google’s docs say it can take 10–20 minutes. In between, all you can do is wait.
Can local simulation replace the official Test Rules?
It can’t replace it, but it can front-run it. Local simulation (a deterministic TypeScript engine) computes Before/After row by row on the sample data you submit, to catch logic problems first (a condition that’s never true, a reference to a non-existent attribute, boundary casing, etc.). Before you actually apply, we still recommend one official Test in Merchant Center.
How much data does local simulation support?
The MVP supports up to 10 rows × 20 columns of sample data — enough to cover the happy path, the negative path, and the edge case (the three-layer accuracy guard from PRD §47).