Find product bundles worth testing from order history
Separate products customers already buy together from combinations that only sound appealing.
What you’ll produce
A ranked list of bundle candidates with co-purchase counts, margin assumptions and a proposed small test.
Inputs & connections
- Order-line export with order IDs, product IDs, quantities, refunds and dates
- Product costs, inventory, shipping constraints and current bundle or promotion rules
Start with a store export. A custom agent or scheduled Slack report requires separately built integrations; this guide does not install the reported SDK workflow.
Your starting prompt
Replace the bracketed fields. This is our suggested prompt, not a quote from the source.
Analyze [order-line export] for bundle ideas over [period]. Deduplicate orders, account for refunds and distinguish repeat units from distinct products. Count how often each candidate pair appears together and show both the count and its share of relevant orders. Flag small samples, existing promotions and unavailable stock. Estimate contribution only from supplied costs, clearly listing missing shipping or fee assumptions. Propose a small bundle test with an inventory limit and comparison metric. Do not present a forecast as guaranteed revenue, create a discount or publish a product. Return the supporting calculations and a decision sheet.
How to run it
- 01
Export a defined period of orders and reconcile orders, refunds and line counts.
- 02
Calculate co-purchase patterns and remove candidates driven by data errors or existing bundles.
- 03
Check stock, margins, shipping and operational fit for a short list.
- 04
Have the owner approve one limited offer and its measurement plan.
- 05
Compare contribution and purchase behavior during the test before expanding it.
Review before you act
Check whether an apparent pairing is common enough to matter and whether the offer still contributes profit after all known costs.
Limitations & success checks
- Products purchased together historically may not sell better as a discounted bundle.
- Small samples and prior promotions can produce misleading recommendations or revenue projections.
Measure: Candidate pairs with sufficient evidence, test contribution margin, stockouts and incremental performance against the chosen comparison.
Sources & attribution
The sources below support the reported example or a tool capability. They do not validate our prompt or prove the outcome in your environment.
Source post and author checked through X public oEmbed on October 17, 2026. The supplied source describes a Claude SDK workflow producing Shopify bundle recommendations and reports. Revenue projections in that report are not validated results; this guide requires a measured trial. The prompt, setup and checks are original editorial additions. The directory has not reproduced the reported deployment or results.
Creator / publisher: @victorpaycro
Editorial review: 2026-10-17. How we label evidence.