Answers hardening and performance budgets
The Answers workflow now includes install-source routing, publish guardrails, performance budgets, and explicit follow-up notes for launch hardening.
Mitigations now in place
- Install source is persisted on the merchant so Answers, Offers, and Watch installs land in the right workflow without breaking the default Rescue path.
- Truth Brief inputs are scrubbed for direct contact details, raw URLs, and record-style identifiers before generation prompts are assembled.
- Publish generation now has a monthly cost circuit breaker and a request-rate limiter to reduce abuse and runaway spend.
- Asset generation already writes to generation cost storage and Answers events so the team can audit who generated and exported what.
Performance budgets
- Scan completion target: under 30 seconds for a 50-product store.
- Asset generation target: under 10 seconds per asset.
- Answers Queue page load target: under 2 seconds.
- Crawl completion target: under 5 minutes for a 200-page site.
Follow-up notes
- Category-pack quality still needs calibration against real scanner telemetry, especially on low-confidence category inference.
- Crawler performance should be benchmarked against larger catalogs before launch to confirm the 200-page budget under realistic storefront conditions.
- Stripe edge cases still need periodic regression checks around trial transitions, webhook retries, and billing-email lead attribution.
- Prompt monitoring should stay event-level and cost-level only; avoid logging raw merchant prompt content beyond the scrubbed prompt assembly path.
- Shopify App Store review materials still need a final compliance pass for AI-generation disclosures and screenshot positioning before submission.
Free Scanner
Think you might have this issue?
Scan your storefront for public Google Shopping blockers before you install. You will get the first findings without connecting Shopify.