Verified against ChatGPT · 2026-08-08
Turn one purchase into a specific cross-sell pitch instead of a generic 'you may also like' rail
Builds cross-sell copy and product pairings tied to the exact item just bought and the problem it doesn't fully solve on its own, for use on the order-confirmation page or a post-purchase email.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are writing cross-sell recommendation copy for one specific purchase moment — not a generic "customers also bought" widget, but a short, reasoned pitch for 2-3 companion products that make the item just bought actually work as intended. ITEM JUST PURCHASED A 4-person dome tent, single unit, no footprint or stakes bundle included. WHAT IT DOESN'T FULLY SOLVE ALONE No ground protection — the tent floor voids its waterproofing warranty if pitched directly on rocky or wet ground without a footprint. CANDIDATE COMPANION PRODUCTS Tent footprint ($24), inflatable sleeping pad ($59), citronella lantern ($18), a second identical tent ($140). PLACEMENT Order confirmation page, directly under the order summary, before the buyer closes the tab. MARGIN OR INVENTORY PRIORITY We're overstocked on the footprint after a supplier minimum order — prefer it over the sleeping pad if both are equally relevant. RULES For each companion product, write one sentence that names the specific gap left by the original purchase and how this item closes it — never a vague benefit statement that would apply to any add-on ever bought alongside anything. If a candidate product doesn't actually address the stated gap, say so explicitly and drop it rather than including it anyway to fill out a list of three. Rank the 2-3 you keep by how directly they close the gap, not by which one has the highest margin — if there's a margin or inventory priority that should break a tie between two options that are otherwise equally relevant, apply it only as a tiebreaker and say where you applied it. Write the actual on-page or email copy in the buyer's voice of urgency: someone who already decided to buy once should be pitched as "complete what you started," not as a fresh sales pitch starting from zero. Do not suggest more than 3 items regardless of how many candidates were given — a longer list reads as upselling everything in the catalog rather than a considered pairing. WHAT NOT TO DO Do not write "frequently bought together" as a justification on its own — that's a statistic, not a reason, and it doesn't tell the buyer why this specific gap exists. Do not recommend a product that competes with or replaces the item just bought. OUTPUT FORMAT 1. A table: companion product | one-line reason tied to the specific gap | where it ranks and why. 2. The actual copy block ready to paste onto the stated placement surface, sized appropriately (a two-line blurb for a checkout page, a short paragraph for an email). 3. Any candidate you dropped and the one-line reason it didn't actually close the gap.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
The instruction to name the specific unsolved gap rather than lean on "frequently bought together" forces the model out of its default pattern-matching mode, where GPT-5.1 will happily generate plausible-sounding companion-product copy purely from category co-occurrence (tents pair with camping chairs, camping chairs pair with coolers) without ever checking whether the pairing addresses an actual functional hole in the first purchase — that co-occurrence reasoning produces technically true but persuasively empty copy, since "other campers also bought this" gives the buyer no new reason to add it to this specific order. Requiring the model to explicitly drop a candidate that doesn't close the stated gap counters a related failure mode: left unconstrained, GPT-5.1 tends to keep every item offered in the prompt because removing one feels like discarding user-supplied information, even when a candidate is actually irrelevant or competitive with the original purchase; stating the drop rule upfront gives it permission to prune instead of rationalizing inclusion. Capping the list at three and ranking by gap-closure rather than margin also blocks a subtler bias — language models trained on marketing copy tend to over-index on whichever product in a list has the strongest sales-adjective density (bundle deals, premium tiers), which silently promotes margin-heavy items over the pairing a buyer would actually recognize as necessary. Framing the copy as "complete what you started" rather than a fresh pitch matters because a buyer who just converted is in a different psychological state than a cold browser, and generic upsell phrasing written for browsing-stage shoppers reads as tone-deaf immediately after checkout.
What you get back
1. Footprint ($24) — protects the tent floor from the exact ground damage that voids its waterproof warranty; ranked first, it's the only item that prevents an outright product failure. 2. Sleeping pad ($59) — the tent keeps you dry, not comfortable; ranked second as the next most direct gap. Lantern dropped — a nice-to-have, not a gap this purchase actually left open. Copy: "Your tent's floor isn't warrantied against rocky or wet ground on its own — add the matching footprint before your first trip out."
Verified against
ChatGPT GPT-5.1 · 2026-08-08
Changelog
- 2026-08-08 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — Google Ads management, built and maintained for you — that's Scult's day job.
EXPLORE GOOGLE ADS MANAGEMENT
