Shopify Sidekick Tested: 5 Real Jobs on the Basic Plan

Shopify Sidekick is the AI assistant built into every Shopify admin, and on the Basic plan it does real work: it audited our store correctly, wrote clean SEO copy from our own product facts, drafted an automation and built a working custom app. It also got a discount's end time wrong by an hour and told us an automation was "ready" in an app we had not installed.

We gave it five jobs a new store owner actually has, on our own store, and checked every answer against the admin. Here is what it did, what it got wrong, and the two things to check before you click Save.

The short version

The pattern held across all five: Sidekick drafts, you save. Every change sat behind an "Unsaved changes" bar until we clicked Save ourselves. That is the safety net, and it only works if you read what it filled in.

How to open Sidekick

Click the Sidekick icon at the top right of the admin, next to the notifications bell. It opens as a panel beside whatever page you are on, and it can see that page. Every conversation is saved under Sidekick conversations in the left menu.

It shows its work as it goes ("Checked shop details, products, publications", "Running code"), which is the first thing worth reading.

Job 1: "What is stopping my store from making its first sale?"

This is the question every new owner has, and it is where Sidekick was best. In under a minute it came back with a ranked list:

1. The store is password-protected. Correct. Our storefront is still in private mode, and that alone stops every sale.
2. Payment setup couldn't be confirmed. In its own words: "I don't have visibility into your Shopify Payments status from here, so please check yourself."
3. One product is out of stock. Correct. The walnut board had 0 units.

Then "everything else looks fine": three active products, two purchasable, shipping zones set, products published to the Online Store and POS. We suspected the POS line was invented, because Point of Sale is not in our left menu. It was not invented. The product's publishing settings list Online Store and Point of Sale, both switched on. Four out of four checkable facts were right.

The gap is item 2. The Home screen of the same admin shows a "Set up payments / Activate payments" card. Sidekick could not see the one blocker Shopify's own dashboard states plainly. It said so honestly, but a new owner skimming "everything else looks fine" could take it as a green light. It also never mentioned the Agentic channel, which decides whether ChatGPT and other AI assistants can show your products. If that matters to you, we tested those requirements separately.

Use it for: a fast first pass over a new store. Then check payments yourself under Settings → Payments. Our Shopify Payments walkthrough covers what it asks for.

Job 2: a discount code for first orders

We typed: "Create a discount code WELCOME10: 10% off a customer's first order, one use per customer, ending October 31, 2026."

Sidekick opened the discount form and filled every field: the code, 10%, one use per customer, an end date. It did not save. The top bar read "Unsaved discount" with Discard and Save, and nothing existed until we clicked.

It was also honest about a real limit: Shopify's native discounts have no "first order only" switch. The first time, it said so and offered to build a customer segment for it. We ran the same prompt again and it took the other route unprompted: it drafted a "First-time customers" segment (customers with zero orders) to attach to the code.

Here is the catch: the end time. We asked for October 31. Sidekick set October 31 at 22:59, not 23:59. Our store runs on Israel time, and daylight saving there ends on October 25, between the day we asked and the end date. Our best reading is that "end of day" was calculated with today's clock offset. Whatever the cause, the code would stop working an hour before the date customers were promised. If your promotion crosses a clock change, check the time field, not just the date.

Two smaller things. The same prompt produced two different designs on two runs, so do not expect identical output. And on the later runs the card hung on "Loading details" for minutes; on one of them it read "Created discount" (we had clicked to another page once during that run, which may be why). Nothing had been created. "Created" on a Sidekick card means drafted, not saved.

Job 3: SEO title and meta description

On our walnut serving board we asked: "Write an SEO title and meta description for this product, then apply them."

• Title: Hand-Rubbed Walnut Serving Board | Ridge & Oak (46 of 70 characters)
• Meta description: Shop a hand-rubbed walnut serving board from Ridge & Oak Supply. Each 14 × 9 × 0.75 inch board has unique grain and color for a one-of-a-kind table. (148 of 160 characters)

Every fact came from the product: the dimensions and the varied grain. It said it wrote "concise search copy built from those confirmed details". This matters because the day before, the description tool (Shopify Magic) given the same product wrote that it was "food-safe" and "durable for everyday use", claims nobody had made. That test is here. Sidekick, with the full product record in front of it, stayed on the facts.

It filled the Search engine listing fields and the page showed "Unsaved changes" until we saved. This is the job we would hand it without hesitation.

Job 4: an automation, and the app it assumed you had

We asked: "Set up an automation: when a customer's order is over $100, tag that customer VIP."

It answered with a card, "Tag VIP customers on high value orders", and said the workflow "is set up but not yet turned on, so head to Shopify Flow to review it".

Shopify Flow was not installed on our store. The Preview button led to "You don't have this app installed. Get Flow and try again." A new owner stops there.

So we installed Flow (free, made by Shopify) and looked. The workflow was not in Flow's list of workflows, active or inactive. It opened only through the Preview link in the Sidekick conversation, as a Draft with "Add to store". Lose the conversation and you lose the easy way back to it.

The logic itself was right: Order created → if order total is greater than 100 → add customer tag VIP. One detail to know: it checks the order's total, which includes shipping and tax. A $95 order with $10 shipping gets tagged VIP. If you meant $100 of products, that is a different field, and you need to say so.

Job 5: a custom app, on the Basic plan

The most surprising result. We asked: "Build me a custom admin app that lists every product with fewer than 5 units in stock."

Sidekick asked permission to use App Generation, then wrote a complete app, more than 270 lines of code, "Low Stock Products", and showed a working preview: one variant under 5 units, one out of stock, the walnut board listed, the two stocked products correctly left out. Clicking Install brought up a plain confirmation with no plan warning. We cancelled rather than install it.

Why that is surprising: Shopify's Help Center, read the same day, says app generation "is available only to stores on the Grow plan, Advanced plan, and Shopify Plus plan." Our store is on Basic. We confirmed it on the Plan page with the app preview open beside it. We cannot tell you whether this is a rollout ahead of the documentation or something that will be switched off. What we can say is exactly what happened: on September 24, 2026, on Basic, it built the app and offered to install it.

It was also candid about the app's limits: a live view, not an alert system, and "for scheduled or email notifications, Shopify Flow is the better fit". That is accurate advice.

What this means for which plan you pick

Sidekick itself is included on every Shopify plan, and everything in jobs 1 to 4 ran on Basic. For a new store, that is the plan to start on: the assistant that audits, drafts discounts, writes SEO and sets up automations comes with it.

Custom app generation is the one feature Shopify documents as Grow and up ($105/month, or $79 billed yearly). It worked for us on Basic, but we would not upgrade to Grow for it, and we would not rely on it staying on Basic either. Start on Basic, and if app generation disappears from your store and you need it, that is the moment to look at Grow.

Who should not rely on Sidekick

If you want an assistant you can hand a task to and walk away, this is not that. Every job needed a human to read what it filled in, and on two of five the thing to catch was subtle: an end time an hour early, a VIP threshold that counts shipping. It is a fast, capable drafter. It is not a second pair of eyes on its own work.

Five rules for using Sidekick safely

1. Read the page, not the chat. The chat says "Done" or "Created". The page bar says "Unsaved changes". Believe the page.
2. Check dates and times on anything with an expiry, especially across a daylight-saving change.
3. Ask what exactly a condition checks. "Order over $100" can mean total, subtotal or after discounts.
4. Install what it assumes you have. Automations need Shopify Flow, which is free but not pre-installed.
5. Keep the conversation. Drafts it creates, like Flow workflows, are easiest to find from the conversation that made them.

How we tested this

All five jobs ran on our own Shopify store, Ridge & Oak Supply, on the Basic plan ($1/month introductory price until December 23, 2026), in the English admin, on September 24, 2026. Every prompt is quoted exactly as typed. Each answer was checked against the admin screens it described. We did not save the discount, did not install the generated app and did not switch on the workflow. The SEO fields and a "First-time customers" segment were the only changes we kept. Shopify Flow was installed so we could inspect the workflow. The plan-gate quote is from Shopify's Help Center page on generating apps with Sidekick, read September 24, 2026. Our store has no payment provider yet, so nothing here involved a real order.

If you are starting from zero, here is how we opened the store, and how selling a service on it works.