ChatGPT Repeat Hoodie Purchase: Build a Dated Difference Log

A ChatGPT repeat hoodie purchase workflow can help compare today's supplied product record with an old order without assuming either change or sameness. Give the model two labelled sources and ask for a dated difference log. The output should distinguish a documented difference from information that was never recorded.

Use the practical repeat-purchase comparison guide to gather the evidence. This companion handles the text comparison and next-question draft. It does not fetch a guaranteed current price from memory or prove that repeating a letter size will reproduce physical fit.

Prepare the two-source worksheet

Remove unnecessary order identifiers, addresses and payment details. The comparison usually needs product identity and relevant fields, not the full receipt. Keep an undated screenshot labelled uncertain rather than treating it as a complete historical specification.

Copy the difference-log prompt

Compare my old and current hoodie records. Preserve source IDs, dates, product identity, currency and selected variant context. For each field, classify it as documented match, documented difference, not recorded in the old source or unresolved. Cite the relevant source text. Do not turn a missing field into proof of a product change. Do not infer measurements from size labels, claim current availability from an old receipt or invent prices and seller terms. End with one precise seller question for the most decision-relevant gap. Do not place an order.

Paste the sources under clear headings. If the current record has not been verified, say so before asking for the comparison. The model's organised table cannot make an outdated source current simply by putting it in the right-hand column.

A filled fictional log

Imagine the old receipt confirms a product name and L size but contains no garment dimensions. The current record also lists an L option and supplies a measuring method. A useful log marks the label as a documented match and the old dimensions as not recorded. It does not conclude that the product grew, shrank or stayed exactly the same.

The next question could concern comparable actual measurements of the owned garment and the current candidate. This is a fictional example, not a claim that any TryBuy product has changed. Historical evidence and current purchase facts remain separate.

Use narrow correction prompts

The old source omitted composition; it did not state a different material. Change that row to “not recorded in the old source” and preserve the current source wording.

Keep the exact currency and variant beside each price. Remove any conclusion about the new final total that the supplied sources do not establish.

Check the revised log against both originals. The model may make a tidy narrative from incomplete records, so inspect any word such as “changed,” “same” or “guaranteed.” Those claims need evidence stronger than a familiar title or photograph.

Review before buying

Confirm the selected product and option on the actual current page before checkout. Use relevant garment measurements and a real try-on where needed for comfort. The log supports a focused question; it does not replace the physical decision or the current transaction record.

When a new seller answer resolves a gap, add it as a dated source and update only the relevant comparison row. Keep the earlier unknown visible in the record history. This makes the final decision traceable without claiming that the assistant independently verified the product or the seller's current terms.

Save the checked comparison with its sources so it can be updated later. For general instruction and verification principles, consult official OpenAI prompting guidance.