Sidekick / New feature

Sidekick now offers "AI-tested"
storefront improvement recommendations

Sidekick finds areas to improve, drafts the changes, and compares them against your current storefront using AI-simulated shoppers. It doesn't test with real shoppers. The merchant reviews the recommendation and decides whether to apply it.

What's on this page
  1. What it does (in 30 seconds)
  2. How it works: From analysis to recommendation
  3. Changes covered by simulation
  4. AI simulation vs. real A/B testing
  5. Requirements
  6. The merchant flow (3 steps)
  7. 5 key points for developers
  8. 3 practical use cases
  9. One-line summary for proposals

1What it does

Sidekick handles "identify improvements → draft changes → compare against the current storefront with AI simulation" from start to finish.
Merchants review the recommendation along with its resultsand simply decide whether to apply it.

Find

Sidekick identifies areas of your storefront that could be improved.

Draft

Creates the changes as a draft. Nothing goes live right away.

vs

Compare with AI

Compares the draft against your current storefront using AI-simulated shoppers.

2How it works: From analysis to recommendation

① Identify improvements Sidekick reviews your storefront ② Create a draft Prepares a modified version as a theme change ③ AI simulation Current vs. draft Compared using AI shoppers ④ Merchant review Review the proposed changes ⑤ Apply or not The merchant decides and applies ①–③ are handled by Sidekick / ④⑤ are up to the merchant
"AI-simulated shoppers"are used for the comparison, so real visitors never see the draft.
The article doesn't specify the simulation method, evaluation metrics, or the format of the results.

3Changes eligible for simulation

Sidekick simulates changes that meet the following 2 conditions.

Condition 1

Can be tested as a storefront theme change

Applies to improvements that can be expressed as theme changes.

Condition 2

Likely to affect add-to-cart behavior

Changes expected to affect add-to-cart are selected.

The article doesn't say how improvements that can't be expressed as theme changes, or changes tied to metrics other than add-to-cart (such as completed purchases or average order value), are handled.

4AI simulation vs. real A/B testing

ItemSidekick's AI simulationReal A/B testing (for reference)
Who it's tested on AI AI-simulated shoppers Real visitors
Exposure to the live store None Not tested on real shoppers Shown live to a subset of visitors
Positioning Not a live A/B test (stated explicitly) Live A/B test
Creating proposed changes Automatic Sidekick creates the drafts Prepare it yourself
Decision to apply The merchant decides after reviewing Review the results and decide yourself
Additional cost Not specified —

5Requirements

Access to Sidekick

Provided as a Sidekick feature.

?

Eligible plans, regions, and languages

Not specified.

The article points to the Shopify Help Center for details. Before relying on this, check in your store's admin whether Sidekick offers these recommendations.

6Merchant flow (3 steps)

1

Receive a Sidekick recommendation

You get improvement ideas compared against your current setup through AI simulation.

2

Review the proposed changes

Review the draft's contents yourself.

3

Decide whether to apply

Apply it once you're satisfied. Nothing goes live automatically.

The final decision stays with the merchant.Sidekick handles the recommendation and the draft; you decide whether to apply it after reviewing. The article doesn't describe the specific UI steps.

75 key points for developers

1. Changes are made at the theme level

Simulations are limited to changes that can be tested as storefront theme changes. Whether changes outside the theme (apps, checkout, etc.) are covered isn't specified.

2. The metric is add-to-cart

Changes are chosen based on their expected impact on add-to-cart. Keep in mind this doesn't indicate impact on purchase completion rate or revenue.

3. Not validated with real users

The article states clearly that it doesn't test with real shoppers and isn't a live A/B test. Treat the results as an early signal of what might work.

DRAFT

4. Changes stop at a draft before going live

Sidekick creates a draft, which is applied only after the merchant reviews it. How the draft is stored in theme management (e.g., as an unpublished theme) isn't specified.

API ?

5. Relationship to APIs and theme development workflows isn't specified

The article doesn't cover API access, conflicts with Git-connected themes, or behavior with custom themes.For projects that manage themes as code, a process for reviewing the diff before applying a Sidekick-created draftshould be agreed on in advance.

83 practical use cases

AI
USE CASE 1

Narrow down ideas worth testing before running a live A/B test

Challenge
There are plenty of storefront improvement ideas, but low-traffic stores can't afford to validate all of them with real A/B tests.
Approach
Review ideas compared against your current setup with Sidekick's AI simulation first, and only move the promising ones on to real testing or production.
Impact
Narrow down options before using real visitors, saving the time and traffic spent on testing.
Technical note
The simulation is explicitly stated not to be a live A/B test. If you use it to justify revenue impact, review it alongside real-world measurements after the changes go live.
USE CASE 2

A starting point for improvements at small stores without an agency

Challenge
There's no in-house UI/UX specialist, so the storefront goes untouched because no one knows what to change to increase add-to-carts.
Approach
Let Sidekick identify improvements and create drafts, so the merchant only needs to review the recommendations and decide whether to apply them.
Impact
Even without specialist knowledge, merchants can start from concrete improvement ideas backed by comparison results.
Technical note
The flow requires reviewing each recommendation before it's applied. A person should decide whether a change affects brand expression.
diff
USE CASE 3

Building reviewed Sidekick recommendations into client operations

Challenge
If clients apply Sidekick recommendations on their own, the changes may fall out of sync with the theme the agency manages.
Approach
Add a rule to the maintenance contract that the agency reviews Sidekick drafts before they're applied, and only apply the reviewed ones.
Impact
Take advantage of the speed of AI-driven improvement suggestions while keeping theme quality and management under control.
Technical note
The article doesn't cover how drafts are stored or how they work with Git-connected themes. Verify the behavior in a test store before rollout.

9One-line summary for proposals

"A new feature where Sidekick handles everything from identifying improvements and drafting changes to comparing them with AI shoppers.
It lets you identify theme changes likely to boost add-to-carts before showing them to real visitors.
It's not a live A/B test, so the final decision and real-world measurement are up to the merchant."