Executive summary
- What: SimGym can now run single-theme simulations against a specific goal, covering the entire shopping journey, product discovery, or add-to-cart confidence, alongside theme comparison.
- Why: Comparing two themes showed which performed better but not why, and a goal-scoped run produces findings you can act on without a second theme.
- Who: Ecommerce and UX teams on Liquid storefronts wanting cheap pre-launch signal before committing to a live test.
What changed
SimGym sends AI shoppers through your store to browse collections, view products, add to cart, and describe the experience. Until now the primary mode compared a published theme against a draft. Three goal-scoped single-theme focuses are now available: entire shopping journey runs the store end to end to surface the largest friction points, product discovery tests whether the home page and collections lead a shopper to a relevant product, and add-to-cart confidence tests whether product detail content gives enough information to buy.
Recommendations have also been reworked. Each carries a numbered marker tying it to the page location it applies to, which removes the interpretation step that previously ate most of the review time. Every recommendation links to a summary of that shopper's visit, with session playback showing where friction occurred and what the shopper responded to well.
The constraints are specific. Your storefront must be Liquid, since Hydrogen and headless are unsupported. Shopify Network Intelligence must be active, the store needs an active plan and at least one eligible priced product, and a permanent password page blocks the app entirely. Runs take one to five minutes. A monthly allowance does not roll over, after which single-theme runs cost 0.75 USD and comparisons 1.50 USD, with comparisons consuming twice the allowance.
Why it matters
The value here is hypothesis generation at trivial cost. A proper experiment needs traffic and weeks of patience, so most teams run far fewer tests than they have ideas. A simulation costing well under a dollar is a filter: it will not tell you what converts, but it will tell you which of nine proposed product page changes deserve the queue slot. Used that way it strengthens a conversion optimisation programme rather than competing with it.
Two cautions matter for Plus merchants. The Liquid requirement excludes headless builds outright, removing a meaningful slice of the enterprise base. And bot protection or rate limiting can block AI shoppers and leave journeys incomplete, which behind an aggressive edge configuration is likely rather than exceptional. If your first simulation returns thin results, suspect your own perimeter before blaming the tool.
Interpret the output as signal, not measurement. These are simulated shoppers whose described reasoning is the useful artifact. Treat each finding as a hypothesis to validate with real traffic, and resist shipping a redesign on simulated evidence.
Role-specific impact
- Marketers: Run a product discovery simulation before a seasonal campaign to check a new landing structure leads shoppers to the promoted catalog.
- Developers: Confirm the store is Liquid and check whether edge bot protection or rate limiting will block simulated traffic. Theme comparison needs at least one draft theme.
- Store admins: Watch the monthly allowance, since it does not carry over and comparisons consume it twice as fast.
Use-case example
Real-world scenario
An apparel brand has eleven proposed product page changes and capacity to properly test two per quarter. The team runs add-to-cart confidence simulations against the published theme and reviews the playback behind each recommendation. Seven proposals attract no shopper friction at all. Three return consistent findings, with simulated shoppers repeatedly unable to locate fit and sizing guidance before adding to cart, and one surfaces an issue nobody had proposed: a size chart in a modal the shoppers never discovered. Total cost is a few dollars and two hours of review, and the quarter's two experiment slots now go to changes with evidence behind them.
Implementation checklist
- Confirm eligibility: Liquid storefront, active plan, Network Intelligence enabled, no permanent password page.
- Check whether bot protection or edge rate limiting will block simulated shopper traffic.
- Pick a goal-scoped focus that matches a real open question, and review the playback behind each recommendation rather than the summary text alone.
- Run one entire shopping journey simulation as a baseline audit, then change one thing at a time and re-run the same focus.
- Promote only validated findings into your testing roadmap, and track which predictions held up.
FAQ
Q: Can we use this on a headless or Hydrogen storefront?
A: No. SimGym requires a Liquid storefront. If your customer-facing experience is headless, the tool is unavailable regardless of plan.
Q: Does this replace conventional conversion testing?
A: No. Simulated shoppers approximate behaviour and results may differ from your real customers. Use simulations to prioritise what you test, then test it properly.
Resources
SimGym simulations and test focuses
Need guidance? Talk to Makro.