What Sidekick Can Actually Do With Shopify Flow
Building a Shopify Flow workflow used to take 30 minutes of clicking through triggers, conditions, and action steps. Now it takes three minutes of typing. That is the headline change in Shopify's 2025 Flow update, and it is why every merchant on the platform should re-examine their automation stack this quarter.
Sidekick, Shopify's in-admin AI assistant, can now author full Flow workflows from conversational prompts. You describe what you want — "when a customer spends over $500, tag them as VIP and send me a Slack alert" — and Sidekick drafts the trigger, conditions, and actions in the Flow editor. You review, test, and activate. No drag-and-drop. No wading through the GraphQL schema to find the right customer field.
This guide walks through the full shopify sidekick automation workflows playbook: how the prompt-to-Flow loop actually works, 10 high-value automations you can build today, when you still need to open Flow manually, and the limits Sidekick hits that merchants keep running into. If you are newer to Flow itself, pair this with our Shopify Flow automation examples post first.
The Prompt-to-Flow Workflow (Step by Step)
The Sidekick-to-Flow loop has four stages: prompt, generate, verify, activate. Skipping verification is where most merchants get burned.
Stage 1: Prompt Sidekick with a clear trigger + action pair
Open your Shopify admin, click the Sidekick icon (top right), and describe your automation in one sentence. According to Shopify's Sidekick Flow changelog, the AI generates the best workflows when your prompt contains:
- A trigger event — "when an order is placed," "when inventory drops below 10," "when a customer is created"
- One or more conditions — "if order total is over $250," "if product tag contains 'seasonal'"
- A concrete action — "send a Slack message," "add the tag 'VIP'," "email the customer"
Example prompt: "When a customer places their 5th order, tag them as 'Repeat Buyer - Gold' and send me a Slack alert in #loyalty-updates with their name and lifetime spend."
Stage 2: Let Sidekick generate the draft workflow
Sidekick uses the full library of triggers, conditions, and actions documented in the Shopify Flow reference. It can also reference Flow actions added by third-party apps — Klaviyo, ShipStation, Smile.io, Gorgias — if those apps are installed on your store.
The output appears as a draft workflow in the Flow editor, not as activated automation. You see each step as a node on the canvas. In January 2026, all workflows automatically render in vertical layout, so the visual flow reads top-to-bottom.
Stage 3: Verify the logic before activating
This is the step merchants skip. Sidekick is a language model drafting against the GraphQL schema — it can miss edge cases, pick the wrong condition operator, or hallucinate a field name. Before clicking "turn on workflow," you need to:
- Read every condition and confirm the operator matches your intent (equal to vs. at least one of)
- Check that tags, channels, and email addresses are spelled exactly as they exist in your store
- Use Flow's preview feature to dry-run the workflow against recent orders or customers
- Confirm any email or Slack actions point to the right recipient
Stage 4: Activate and monitor the first 24 hours
Turn the workflow on, then watch the Run History pane for the next 24 hours. Look for failed runs, skipped conditions, and unexpected recipients. Flow's 2025 redesign added the ability to cancel failing runs in bulk, which prevents one broken workflow from flooding your Slack channel or spamming customers. For deeper automation philosophy, see our guide to what you can automate on Shopify with AI.
10 High-Value Automations You Can Build With Sidekick

Each of these is a real workflow you can prompt Sidekick to generate today. I have ordered them by revenue impact and ease of setup — start at the top.
1. Abandoned Cart SMS Recovery
Prompt: "When a checkout is abandoned and the cart total is over $75, wait 30 minutes, then send the customer an SMS with a 10% discount code if they opted in to marketing."
Abandoned checkout automation is the highest-leverage flow most stores are missing. SMS recovery converts at 3-5x the rate of email because the message lands on the lock screen. The "cart over $75" condition prevents you from discounting low-value carts where margin is already thin. For a deeper play-by-play, see our Shopify abandoned cart automation guide.
2. Loyalty Tier Upgrade Tagging
Prompt: "When an order is paid, if the customer's total spent is over $2,500 and they have 5+ orders, add the tag 'VIP-Gold' and send them a welcome email from our loyalty template."
This is the workflow that powers every VIP program on Shopify without a dedicated loyalty app. Mesa's Flow templates guide highlights the "total spent + order count" condition combo as the cleanest way to segment high-value customers without false positives from one-time big spenders.
3. Low-Stock Slack Alerts
Prompt: "When a product variant's inventory drops below 10 units, send a Slack message to #inventory-alerts with the product title, variant name, and current quantity."
Low-stock alerts let you reorder before you run out, which prevents the silent revenue leak of out-of-stock product pages. Sidekick can also layer in conditions — "only alert me for products tagged 'bestseller'" — so you are not drowned in notifications for dead SKUs. For demand-side forecasting, pair this with our Shopify Sidekick prompts for inventory forecasting.
4. New-Customer Onboarding Sequence
Prompt: "When a customer is created and they have no orders yet, wait 2 hours, then send them a welcome email with our product guide, and tag them as 'New - Pre-Purchase'."
The "created but not converted" window is underused. Most stores only trigger welcome emails off first purchase — by then the customer has already made up their mind. A pre-purchase nudge with educational content (sizing guide, material story, founder video) increases first-order conversion by 10-15% in most categories.
5. Fraud Review Flags for High-Risk Orders
Prompt: "When an order's risk is analyzed as High, hold fulfillment, add the tag 'Fraud-Review', and send me a Slack alert in #order-review with the order number and risk factors."
Critical setup note from Shopify's fraud automation docs: use the Order risk analyzed trigger, not Order created. Fraud analysis takes a few seconds to complete after an order lands, and workflows triggered on Order created will run before the risk score exists. Also confirm your payment settings are on manual capture — otherwise the hold-fulfillment logic will not prevent charge capture.
6. VIP Order Tagging + Concierge Routing
Prompt: "When an order is paid and the customer has the tag 'VIP-Gold', add the order tag 'Priority-Fulfillment' and email our concierge team the order details."
This is how you operationalize a VIP tier: once customers earn the tag (from workflow #2 above), every order they place gets fast-tracked automatically. Your fulfillment team sees the priority tag in the order list, and your customer team gets a heads-up to send a personal thank-you. It is the closest thing a Shopify store has to an enterprise concierge playbook.
7. Subscription Churn Recovery on Failed Payment
Prompt: "When a subscription payment fails, immediately send the customer an email with their updated payment link and add the tag 'Payment-Failed-1st-Attempt'. If the next attempt also fails, tag them 'Payment-Failed-Retry' and send a Slack alert."
Up to 48% of subscription churn is caused by failed payments (involuntary churn), not actual cancellation intent. The fix is a payment-recovery flow that emails a working link within seconds. Smartrr's subscription Flow templates show how to chain retry logic with escalating tags so your CX team can intervene on the third failure before the customer is gone for good.
8. Post-Purchase Review Request Follow-Ups
Prompt: "When an order is fulfilled, wait 10 days, then send the customer an email asking for a product review with a link to our review form."
Ten days is the sweet spot for most categories — long enough for the product to arrive and be tried, short enough that the purchase is still top of mind. If you sell apparel or anything seasonal, bump it to 14 days. Sidekick can also add conditions like "only send if order has no refund" so you do not email unhappy customers.
9. Refund Auto-Approval Rules
Prompt: "When a refund is requested and the refund amount is under $25 and the order was placed less than 30 days ago, auto-approve the refund and send a confirmation email."
Small refunds cost more in CX time than they do in lost revenue. Auto-approving refunds under a threshold (commonly $25-50) frees your team to handle the genuinely complex cases. Pair this with a customer-service automation layer — see our chatbot ecommerce guide — so refund requests get handled end-to-end without a human.
10. Cross-Sell Email After Purchase
Prompt: "When an order is paid and contains any product tagged 'coffee-beans', wait 21 days, then send the customer an email suggesting our grinder and filter bundle."
Category-aware cross-sell is one of the highest-ROI automations on Shopify. The key is tying the wait time to actual consumption — 21 days after a 12oz bag of coffee, most drinkers are about to run out. Use your product tags (the same ones powering your navigation) as the trigger condition, and let Sidekick handle the email template assignment.
Shopify Sidekick Flow Automations at a Glance

| Automation | Trigger | Key Condition | Primary Action | Revenue Impact |
|---|---|---|---|---|
| Abandoned Cart SMS | Checkout abandoned | Cart > $75, SMS opt-in | Send SMS + code | High |
| Loyalty Tier Upgrade | Order paid | Total spent > $2,500 | Add VIP tag + email | High |
| Low-Stock Slack Alerts | Inventory changed | Quantity < 10 | Slack message | Medium |
| New-Customer Onboarding | Customer created | Zero orders | Wait + welcome email | Medium |
| Fraud Review Flags | Order risk analyzed | Risk = High | Hold + Slack alert | Critical (loss prevention) |
| VIP Order Concierge | Order paid | Tag = VIP-Gold | Priority tag + email team | High |
| Subscription Churn Recovery | Subscription payment failed | Any | Payment link email + tag | Critical |
| Review Request | Order fulfilled | No refund | Wait 10d + email | Medium |
| Refund Auto-Approval | Refund requested | Amount < $25, order < 30d | Approve + email | Medium (CX time) |
| Cross-Sell Post-Purchase | Order paid | Contains tagged product | Wait + suggest bundle | High |
When You Still Need to Edit Flow Manually
Sidekick is excellent at first drafts. It is not excellent at nuanced logic. Here are the scenarios where you will still want to open the Flow editor directly after Sidekick generates a draft.
Complex branching logic
Sidekick handles linear flows well — trigger to condition to action. When you need multi-branch logic (if A then X, if B then Y, if C and not D then Z), the AI tends to collapse branches or miss the "and not" negation. Build the skeleton with Sidekick, then open the editor and add the branches manually using the "if/then/else" node.
Custom Liquid in email actions
Flow's Send email action supports Liquid templating for dynamic content — order line items, customer first name, product variants. Sidekick will draft a plain-text email body but will not write the Liquid loops that pull order data. You will need to edit the email step and paste in {% for line_item in order.lineItems %} blocks manually.
App-specific action configuration
Third-party Flow actions (Klaviyo profile updates, ShipStation shipment creation, Yotpo review requests) often require specific field mappings that Sidekick does not know by default. The action will appear in the workflow, but you will need to open it and fill in the app-specific fields. For context on when to integrate Flow with your app stack, see our Shopify Flow vs Zapier comparison.
Scheduled time triggers
Sidekick is optimized for event-driven triggers (order placed, customer created). Scheduled workflows — "every Monday at 9am, run a report" — work but often need manual tweaking of the recurrence pattern and timezone settings in the trigger configuration panel.
Testing and Rollback: The Safety Playbook

A bad workflow can spam customers, leak margin, or flag legitimate orders as fraud within minutes. Treat every Sidekick-generated workflow as untrusted until you have tested it. Here is the safety checklist.
Use Flow's preview mode before activating
The 2025 Flow update introduced a preview feature that simulates your workflow against recent real orders or customers without actually running the actions. Use it on every new workflow. If preview shows "would have sent email to 400 customers" when you expected 40, you have a condition bug.
Start with narrow conditions, then widen
If your first draft is a VIP tagging flow, set the threshold at "total spent > $5,000" for the first week instead of $500. You will catch zero false positives at the high threshold, confirm the logic is sound, and then lower the bar. This is the same principle behind canary deployments in software — small blast radius first.
Monitor the Run History for 48 hours
Flow's Run History shows every execution with the trigger data, the condition evaluations, and the action results. Check it twice daily for the first 48 hours after activation. Failed runs are red and often point to missing fields or invalid recipients. Shopify's Flow optimization docs recommend archiving workflows with chronic failures rather than leaving them half-broken.
Keep a rollback tag
When a workflow adds tags (VIP, priority, churn-risk), include the workflow version in the tag — "VIP-Gold-v1" instead of "VIP-Gold." If you later realize the workflow tagged the wrong customers, you can do a bulk search-and-remove on v1 without affecting tags from v2. Small discipline, huge time saver during incident recovery.
Document every workflow in a shared doc
Flow has 50+ triggers and 100+ actions. After six months you will forget what a workflow does, which overlaps which, and which ones to update when you change your product catalog. A simple Notion or Google Doc with "workflow name, purpose, trigger, conditions, owner" prevents the Flow sprawl that hits every store at year two.
Limitations: What Sidekick and Flow Cannot Do
Sidekick is impressive for a first-generation AI Flow builder, but it has real limits that merchants run into. Understand these before you promise your team "Sidekick can automate that."
No access to incoming inventory
Per a recurring Shopify community thread, Flow cannot read incoming inventory (purchase orders in transit). So you cannot build a "low stock but PO incoming" condition directly. Workarounds require a third-party inventory app that surfaces incoming as a product metafield.
No control over app-store apps
Sidekick manages native Shopify functionality, but eesel AI's Sidekick overview notes it cannot configure or directly control apps installed from the App Store. If Klaviyo, Recharge, or Gorgias expose Flow actions, Sidekick can use those actions — but it cannot go into the app itself and change settings.
Cannot talk to your customers
Sidekick is a merchant tool, not a customer-facing agent. It will not respond to support tickets, chat with shoppers, or handle Gorgias conversations. For customer-side AI, you need a separate helpdesk integration. Our Sidekick limitations deep-dive covers this in more detail.
Workflow complexity ceiling
Sidekick generates workflows with up to around 10-15 steps reliably. Beyond that, it tends to hallucinate steps or reuse conditions incorrectly. If you need a 30-step workflow (rare but real for Shopify Plus merchants), split it into two chained workflows or build it manually.
GraphQL schema drift
Shopify's GraphQL Admin API evolves quarterly. Sidekick's training may lag the latest field additions by weeks. If you prompt Sidekick to use a condition that only shipped last week, it may not know the field exists and will pick a close-but-wrong alternative. Always verify condition fields against the current API reference.
Common Mistakes Merchants Make With Sidekick Flow Automations

The seven mistakes below account for roughly 80% of broken Sidekick-built workflows I see in the Shopify community and the Talk Shop community.
| Mistake | Why It Breaks | Fix |
|---|---|---|
| Using "Order created" for fraud workflows | Runs before risk analysis completes | Use "Order risk analyzed" trigger instead |
| Not setting payment to manual capture | Payment auto-captures before fraud hold fires | Switch to manual capture in payment settings |
| Vague prompts ("set up VIP stuff") | Sidekick invents thresholds and tags | Prompt with explicit thresholds, tags, and recipients |
| No preview before activating | Hidden bugs spam customers on day one | Always run Flow preview on a test order or customer |
| Duplicate tags across workflows | Two flows both add "VIP-Gold" | Version your tags ("VIP-Gold-v2") and audit monthly |
| Missing opt-in conditions on SMS | Sends SMS to unsubscribed customers (compliance risk) | Add "customer accepts SMS marketing" condition |
| Over-automating refunds | Auto-approves fraud-adjacent refunds | Cap auto-approval at low-dollar, recent orders only |
The prompt specificity rule
The single biggest predictor of whether a Sidekick workflow works on the first try is how specific your prompt is. "Tag VIPs" fails. "When an order is paid, if the customer's total spent is over $2,500 AND order count is 5 or more AND average order value is above $250, add tag 'VIP-Gold' and email loyalty@mystore.com" works. Specificity in, correctness out.
The "test order" habit
Create a test customer and place a test order through Shopify's Bogus Gateway. Every new workflow should run against this test order first. It takes five minutes and catches the dumb mistakes (wrong tag, missing email, broken condition) before they hit real customers.
Pulling It All Together
Sidekick turns Shopify Flow from a power-user tool into something your entire team can use. A customer-service rep can describe a refund automation in plain English and get a working draft in 30 seconds. An inventory manager can build a low-stock alert during a standup. The bar for "who can automate" just dropped by an order of magnitude, which means your store should be running more automations than it was six months ago, not the same five you set up in 2023.
The four-step discipline — prompt clearly, verify carefully, test safely, monitor actively — is what separates merchants who get 10x productivity from merchants who get 10x chaos. Start with two or three of the automations above, prove the loop works on your store, then scale up.
If you are building out your automation stack from scratch, the wider Talk Shop automation resources and our AI emerging tech library walk through the adjacent pieces: chatbots, inventory AI, agentic commerce, and the newer Sidekick features shipping in Shopify's Editions releases.
Which of these 10 workflows will you prompt Sidekick to build first? The one that saves you the most time usually is not the one that sounds flashiest — it is the quiet, compounding tag-and-route workflows that run 1,000 times a month without you noticing.

About Talk Shop
The Talk Shop team — insights from our community of Shopify developers, merchants, and experts.
