GEOly Research · 2026-11-03
The AI gift shelf: who ChatGPT recommends on holiday prompts
When someone asks an assistant what to buy as a gift, a shelf of products forms in the answer. We cut 159,227 ChatGPT shopping cards from holiday-intent prompts to see who is on that shelf, which retailer gets the click, and how much the answer changes from the all-intent shelf.
What the numbers say
- ×11Etsy's share of the shelf multiplies on gift intent — 0.5% of all-intent cards, 5.5% of holiday-tagged ones, moving it from 14th to 3rd.
- −6.6ppHome Depot falls from 11.6% to 5.0%. Gift prompts push the shelf away from home improvement and toward gift and consumer-electronics retail.
- 81.5%Chinese cross-border brands own the power-bank shelf: Anker, UGREEN, INIU, Baseus and CUKTECH together hold 81.5% of brand-matched cards.
- 7× gapConcentration varies enormously by category: Samsung holds 36.1% of tablets while the top jewelry brand holds 5.1%. The same GEO playbook does not fit both.
- Holiday cards
- 159,227
- Holiday prompts
- 4,320
- Categories
- 1,066
- Link domains
- 14,290
A note on Black Friday
This is the holiday baseline, not a Black Friday reading. The numbers describe the ChatGPT run of 2026-09-04 — the most recent date in our public shopping dataset. A Black Friday edition needs the public dataset publishing through the 2026-11-25 → 12-01 window and a Black-Friday-intent prompt set; until both exist we will not label September data as Black Friday.
Where the gift shelf sends the click
Share of holiday-tagged cards whose link points at each domain, next to the same domain’s share on the all-intent shelf. The delta is the part worth reading: gift intent rewards gift and electronics retail and punishes home improvement.
| Domain | Holiday share | All-intent | Delta | Cards |
|---|---|---|---|---|
| walmart.com | 16.2% | 17.3% | -1.1pp | 25,281 |
| bestbuy.com | 6.3% | 4.9% | +1.4pp | 9,924 |
| etsy.com | 5.5% | 0.5% | +5.0pp | 8,667 |
| target.com | 5.2% | 5.7% | -0.5pp | 8,197 |
| homedepot.com | 5.0% | 11.6% | -6.6pp | 7,836 |
| nordstrom.com | 4.9% | 3.7% | +1.2pp | 7,610 |
| wayfair.com | 3.3% | 3.9% | -0.6pp | 5,184 |
| ulta.com | 1.3% | 2.8% | -1.5pp | 2,088 |
| backcountry.com | 1.2% | 1.0% | +0.2pp | 1,830 |
| macys.com | 1.1% | — | — | 1,712 |
| bhphotovideo.com | 0.9% | 1.2% | -0.3pp | 1,435 |
| lowes.com | 0.8% | 1.7% | -0.9pp | 1,286 |
| microcenter.com | 0.8% | — | — | 1,276 |
| scheels.com | 0.8% | 1.0% | -0.2pp | 1,215 |
| guitarcenter.com | 0.7% | — | — | 1,160 |
The shelf, category by category
Ten categories with enough holiday-tagged volume to rank. Brand shares are a share of brand-matched cards in that category — the denominator is stated on each page, because it varies from 35% of cards (jewelry) to 98% (smartwatches).
| Category | Cards | Leader | Leader share | |
|---|---|---|---|---|
| Jewelry | 1,410 | Macy's | 5.1% | View category → |
| Chargers | 1,010 | Anker | 32.0% | View category → |
| Electric Scooters | 951 | Segway | 20.3% | View category → |
| Makeup Gift Sets | 930 | Clinique | 8.1% | View category → |
| Power Banks | 759 | Anker | 29.4% | View category → |
| Tablets | 674 | Samsung | 36.1% | View category → |
| Smartwatches | 645 | Garmin | 33.4% | View category → |
| Sound Bars | 608 | Samsung | 25.0% | View category → |
| Headphones | 588 | Sony | 24.2% | View category → |
| Electric Bikes | 557 | Specialized | 13.5% | View category → |
What to do with this
1. Find your category’s concentration before setting a target. On a shelf like tablets you are fighting one incumbent for second place; on jewelry the shelf is long-tail and winnable with volume of well-structured product pages.
2. Check which domain the card links to, not just whether your brand is named. A card that names you but links to Walmart hands the click to the retailer — see the shelf study for own-site shares by category.
3. Gift intent is a different prompt set from your evergreen one. If you only track “best X”, you miss the shelf that forms around “gift for…”, where the retailer mix and often the winning brands differ.
4. Two hero products can carry a category. In headphones, two models are 21.4% of all cards; concentrating your content and PR on one flagship can beat spreading it across a catalogue.
See your own brand on this shelf
GEOly tracks which products AI assistants recommend in your category, which sources they cite, and where the click goes — daily, across four engines.
Methodology
- Source: GEOly’s public shopping dataset. One row per ChatGPT shopping card from the weekly run of 2026-09-04, US-English prompts. 159,227 cards, 156,423 of them carrying a link domain.
- Holiday intent: a card counts as holiday-tagged when its topic’s own seasonality label contains "holiday". That is the prompt set’s label, not our reading of the prompt text. 4,320 prompts across 1,066 categories qualify.
- Brand shares: share of brand-matched cards within the category, deduped to one row per card by highest match confidence. Cards whose title matches no brand in our brand table are excluded, so the denominator is smaller than the card count — it is printed on every category page. Jewelry matches 35% of its cards; the electronics categories match 93–98%.
- Product shares: share of all cards in the category, grouped by normalised title. The price is the mean of the minimum price read off those cards, rounded to the dollar; it is a card-side reading and not an offer or a street price.
- All-intent comparison: the "all-intent" column is the same domain’s share on the full shelf of 695,137 linked cards from the same run, as published in our shopping shelf study. A blank means the domain is outside that study’s top 14.
- Limits: one engine (ChatGPT), one country (US), one weekly run. Shares describe that snapshot, not a trend. Brand matching is name-based, so a brand sold only under a retailer’s private label will be undercounted.
- How to cite: "GEOly, The AI gift shelf (2026), data from the ChatGPT run of 2026-09-04." Please link to this page. Re-use the tables freely with attribution.