How to Audit Your Active Shopify Discounts (and Where AI Helps)

Discount codes accumulate. A welcome code from 2023, a one-off for an influencer, last year's Black Friday code that someone set without an end date, a code created to fix one customer's order. Each made sense on the day. Together they're a slow leak.

An audit takes about twenty minutes and is worth doing before every big sale. Most of it happens in the Shopify admin, because that's where the discount list lives. AI earns its place in the thinking: working out which combinations are risky and what to switch off.

What you're looking for

Five kinds of problem code:

1. No end date. A promotion that was meant to last a weekend and is still live.
2. Leaked. A private code that's ended up on coupon sites.
3. Stacking. Codes that combine with each other, or with an automatic discount, into a bigger cut than you planned.
4. Too generous for today. A percentage that made sense at old prices or old margins.
5. Dead weight. Codes nobody uses that clutter the list and hide the ones that matter.

Step 1: Pull the active list

In your Shopify admin, go to Discounts. Shopify's guide to managing discounts lists the filters you can apply:

• Status
• Method
• Type
• Combines with
• Start date
• Times used
• Eligibility, including customer segment and specific customer
• Tag

Start with Status set to active. That's your working list. Everything you care about is on it.

If you'd rather ask than click, Sidekick can answer questions about your store's discounts from inside the admin: "Which discount codes are currently active, and when does each one end?" Check its list against the filtered Discounts page the first time, so you know it's complete.

Step 2: Find the codes with no end date

Go down the active list and look at the dates. Anything without an end date should have a reason. A standing welcome code is a fair one. "BFCM2024" is not.

For each open-ended code, decide: give it an end date, or deactivate it.

Step 3: Check what's actually being used

Filter or sort by Times used.

• High use on a code you thought was private suggests it's leaked. Search the code name with the word "coupon" and see whether it's listed on a deals site.
• Zero use on an old code means nobody needs it. Deactivate it.
• Steady use on a code you'd forgotten means it's doing something, good or bad. Find out which.

To see the money, ask an assistant with access to your analytics. In Sidekick, or in a connected Claude or ChatGPT:

Show orders and total sales by discount code for the last 90 days.

That turns "times used" into revenue and tells you which codes matter enough to think hard about.

Step 4: Check what stacks

This is where a careful audit pays for itself, and where the rules are less obvious than people expect.

Per Shopify's discount combinations guide:

• Discounts don't combine automatically. Each discount has combination settings, and several apply to one order only when the setting on each allows it.
• Every discount is one of three classes: product, order, or shipping.
• They apply in that order. Product discounts first, then order discounts on the reduced subtotal, then shipping.
• A customer can use up to 5 product or order discount codes and 1 shipping code on one order.
• You can have up to 25 active automatic discounts.

To find the codes that can stack, use the Combines with filter on the Discounts page and choose a class. What comes back is every discount set to combine with that class.

Then think it through. This is a good job for AI. Copy the list of combinable discounts, with their values and classes, into your assistant:

Here are my active Shopify discounts that are set to combine, with their type and value. Product discounts apply first, then order discounts on the reduced subtotal, then shipping. What's the largest total discount a customer could get by stacking these on a 100 dollar order? Show the working.

You'll get a worst case in seconds. If the answer is "62% off plus free shipping," you've found something to fix before a customer does.

Don't forget automatic discounts. A sitewide automatic sale plus a 20% welcome code that's set to combine is the classic way to give away more than you meant to.

Step 5: Check who can use each code

Filter by Eligibility. A code meant for one customer segment that's open to everyone is a leak waiting to happen. For anything valuable, limit it to the segment it was made for. Building those segments is quick with AI, as covered in customer segments with AI.

Step 6: Clean up

Shopify gives you three actions, and the order to reach for them is deliberate:

• Deactivate. Temporary and reversible. Open the discount and click Deactivate, or tick several on the Discounts page and use Deactivate discounts. This is the right default.
• Edit. Change the end date, the value, the eligibility, or the combination settings, then save. Editing doesn't affect orders where the code was already used.
• Delete. Permanent. It removes the discount from your list. Only for codes you're sure about.

Deactivate first. If nobody complains in a month, delete.

Do all of this in the admin. Shopify's official Claude and ChatGPT integrations can't change a discount code that already exists.

Where AI helps, and where it doesn't

Before a sale

Run this audit a week ahead of any big promotion. The specific risks then:

• Last year's sale code is still live and will be found.
• A standing welcome code combines with your sale price.
• An automatic discount and a code both apply.

Five minutes on the Combines with filter covers all three. It's step 3 of the BFCM prep plan for good reason.

A naming habit that makes the next audit faster

Most of the pain in an audit is working out what a code was for. Fix that at the source:

• Put the purpose and the period in the name: WELCOME10, BFCM26-VIP, INFL-JAMIE-OCT.
• Use tags on discounts for the campaign, so the Tag filter can pull a whole campaign at once.
• Set an end date on everything that isn't meant to be permanent.

The short version

1. Filter Discounts by active status.
2. Give every open-ended code an end date or deactivate it.
3. Look at times used, and revenue by code.
4. Use Combines with to find what stacks, and have AI work out the worst case.
5. Tighten eligibility on valuable codes.
6. Deactivate first, delete later.

If a code came out wrong in the first place, see why AI created the wrong discount.