What does the holiday catalog research show?#

Holiday gift questions repeat, but the standard product-data fields reviewed by Lily AI do not directly capture many of the facts those questions ask for. In the filtered comparison, 92 of the top 100 December 2024 gift phrases remained in the top 100 in December 2025. Separately, 60.4% of December 2025 estimated gift-search volume had a primary constraint with no dedicated standard attribute in the specification mapping. [C1] [C2]

92 of 100
of December 2024’s top 100 gift searches were back in the top 100 a year later.December 2024 to December 2025, 558 gift searches (Ahrefs US estimates).
60.4%
of gift-search volume turns on a fact with no standard field.Share of December 2025 gift-search volume, 598 searches (Ahrefs US estimates).
87%
of September to December 2025 volume arrived in November and December.Share of September to December 2025 volume, 24 holiday searches (Ahrefs US estimates).

These are three different denominators. The findings do not mean that 60.4% of retailers have incomplete catalogs or that 87% of annual retail demand arrives in two months. The study compares modeled search demand with a product-data specification; it does not audit every retailer’s published listings.

The practical task is to identify recurring questions, attach verified product evidence to them and publish that evidence through supported fields and visible product content. A product can be suitable for a request without its listing explaining why.

How predictable are the leading holiday gift questions?#

The leading phrases changed little between December 2024 and December 2025. Of the top 25, 24 returned; of the top 50, 47 returned; of the top 100, 92 returned. The corresponding retention rates were 96%, 94% and 92% in the filtered 558-phrase cohort. [C1]

Wide table: swipe horizontally or focus the table and use the arrow keys to see every column.

Gift-phrase retention: December comparisons
ComparisonCohortTop 25 retainedTop 50 retainedTop 100 retainedRank correlation
Dec 2024 to 202555824 / 25 (96%)47 / 50 (94%)92 / 100 (92%)0.97
Dec 2019 to 202523120 / 25 (80%)40 / 50 (80%)85 / 100 (85%)0.87

The six-year comparison also shows substantial persistence: 85 of the top 100 remained between December 2019 and December 2025 in a 231-phrase cohort. That result needs a selection caveat. The lists were assembled from recent leading terms, favoring phrases that survived. Treat the long-run overlap as an upper-bound indication within this selected set, not an unbiased estimate for all historical gift searches.

Rank correlation was 0.97 for the filtered consecutive-year cohort and 0.87 for the six-year cohort. Across the broader 575 gift phrases naming no year, the consecutive-year correlation was 0.953. These are separate calculations. The report’s headline uses the filtered 558-phrase comparison.

Your December 2026 question list is, to a first approximation, last December’s. It is public, specific and free, so test the catalog against it before the freeze.

What do gift searches ask for when they name no product?#

They commonly name a person, an occasion, an age, a budget or a situation. In the 598-phrase dataset, 98% of December 2025 estimated US gift-search volume named no product. That is a volume share within the selected gift-search universe, not a share of all shopping searches or people. [C1]

Selected gift phrases: December 2025 estimated US searches
MeasureEstimated searches
White elephant gift ideas964,875
Gifts for mom556,168
Gifts for dad463,044
Stocking stuffers322,667
White elephant gift ideas $2579,830
Gifts for teen boys54,804
Gifts for 10 year old boys40,110
Gifts for men who have everything38,333

Source: Lily AI analysis of Ahrefs US estimates for December 2025, pulled 20 October 2026. All bars share one scale. Searches, not shoppers or sales. [C1]

“Gifts for dad” asks for a recommendation without specifying a product category. “White elephant gift ideas $25” adds an occasion and price limit. “Gifts for 10 year old boys” makes exact age relevant. “Gifts for men who have everything” asks the retailer to interpret a situation.

Which products answer this question, and where is the deciding fact written?

Each phrase still needs judgment. Being related to an occasion does not establish that every product is a good gift for it. Catalog enrichment should express supported facts, such as intended activity, recommended age range, dimensions or included accessories. It should not invent recipient suitability simply because the phrase has search volume.

Which shopping constraints lack a dedicated product-data field?#

The specification mapping placed 60.4% of December 2025 estimated gift-search volume in constraints without a dedicated standard attribute, 7.7% in constraints represented only by a coarse age band, and 31.9% in constraints with an existing attribute. Each query was assigned to one primary constraint. [C1] [C2]

Primary-constraint coverage: share of December 2025 estimated volume
MeasureShare of volume
No dedicated standard attribute60.4%
Coarse age band only7.7%
Attribute exists31.9%

Wide table: swipe horizontally or focus the table and use the arrow keys to see every column.

Exclusive primary constraints: 598 queries, December 2025
Primary constraintQueriesEstimated searchesShare of volumeSpecification mapping
Occasion1154,236,63632.7%No dedicated attribute
Relationship1242,859,01422.1%No dedicated attribute
Gender only972,738,02821.2%gender
Age band51866,4406.7%age_group
Exact age83649,7035.0%Coarse band only
Budget38529,4574.1%price
Passion or role44507,8233.9%No dedicated attribute
Teen27345,4832.7%Coarse band only
Situation or feeling19206,5721.6%No dedicated attribute
Total59812,939,156100%Each query counted once

Occasion and relationship account for the largest no-field categories, followed by passion or role and situation or feeling. Their displayed shares add to 60.3%; the reported 60.4% aggregate uses unrounded values. The individual rows use rounded percentages.

Exact age and teenage needs are only partly represented by broad age groups. A coarse label does not communicate the same information as a verified recommended age range. Gender, a supported broad age group and price have standard attributes in the reviewed specification.

In Google’s age_group, a 10-year-old can only be “kids” and a 15-year-old only “adult”. Occasion, recipient and hobby have no dedicated field.

Do not combine this classification with the companion AI answers report. This report counts each query once, using the constraint that decides it. The AI answers report allows multiple intent tags on one query. Its person, occasion and budget shares therefore answer a different question and can overlap.

Plain “Christmas gift” searches, about 11% of the dataset’s volume, count as occasion searches. Excluding them reduces the no-field share to 56%. Excluding “gifts for kids,” which shows an apparent tracking change, gives a split of 62.5% no dedicated field, 8.0% coarse band and 29.6% existing attribute. Those rounded sensitivity shares total 100.1%.

A missing dedicated attribute is not a prohibition on describing a fact. Supported detail or highlight fields and visible product copy can carry information that has no standalone general field. The applicable platform and category requirements still govern where it belongs.

How can a listing answer a specific gift request?#

Break the request into verifiable facts, then map each fact to an appropriate publishing route. For “a gift for my 10-year-old nephew who plays soccer, under $50,” price alone does not answer the request.

Wide table: swipe horizontally or focus the table and use the arrow keys to see every column.

Illustrative mapping of a gift request to product evidence
RequestVerified fact to supplyPossible route in reviewed Google specification
Under $50Current priceprice
10 years oldRecommended age range, for example 8 to 12product_detail (age_group can only say “kids”)
Plays soccerThe activity it is made forproduct_highlight
A gift for my nephewWho it suits, plus gift facts such as no sizing neededproduct_highlight or question_and_answer

This is an illustrative content brief, not an observed product or a performance result. Recommended ages 8 to 12 are example values that require product evidence. A relationship such as “nephew” is not a product specification; useful supporting facts may include intended junior activity, included equipment or whether sizing is required.

Only the price has a standard home. The age, the activity and the gift facts exist only if someone writes them in.

Illustration, not a real product. Field names from Google Merchant Center’s product data specification, checked 29 October 2026. [C2]

When does the recurring demand arrive?#

In the selected 24-query basket, November and December accounted for 87% of September to December 2025 estimated searches. September contributed 5%, October 8%, November 31% and December 56%. These are shares of that four-month window, not annual shares. Monthly data cannot identify the busiest individual week. [C1]

24-query basket: share of September to December 2025 estimates
MeasureShare of four-month volume
September5%
October8%
November31%
December56%

The same basket’s November and December total was 3,823,119 in 2019 and 3,864,448 in 2025. Within 2019 to 2025, the busiest season was about 17% above the quietest. The series supports planning for recurring demand in this basket; it does not establish that total retail spending or all gift searches stayed flat.

24-query basket: November and December estimated US searches
MeasureEstimated searches
20163,387,075
20173,746,887
20183,545,902
20193,823,119
20204,272,632
20214,457,292
20223,901,564
20233,893,797
20243,823,101
20253,864,448

In the 24-search basket, the busiest season from 2019 to 2025 was only 17% above the quietest. Plan the questions a year ahead, the way you plan stock.

Are gift questions becoming more specific?#

In a fixed cohort of 232 phrases already tracked in 2019, phrases with five or more words grew 52.1% in November and December estimated volume by 2025. Three-word phrases grew 11.9%, and four-word phrases grew 14.5%. Long-phrase growth paused in 2025, so the six-year result should not be presented as uninterrupted annual acceleration. [C1]

Fixed 2019 cohort: November and December growth to 2025
MeasureChange in estimated searches
Three words11.9%
Four words14.5%
Five or more words52.1%

Wide table: swipe horizontally or focus the table and use the arrow keys to see every column.

Growth values: word length and a separate topical cut of the 232-phrase cohort
GroupPhrasesNov + Dec 2019Nov + Dec 2025Change
Three words414,242,3824,747,415+11.9%
Four words1042,285,7372,617,300+14.5%
Five or more words87999,7251,520,624+52.1%
Situation or feeling9131,791226,159+71.6%
Passion or role26501,586652,040+30.0%
Age or life stage871,638,2042,310,674+41.0%
Relationship593,150,6273,103,283-1.5%

In the same 232-phrase cohort, exact-age phrases supplied 73% of long-phrase volume. A separate cut of the same cohort shows situation or feeling searches growing 71.6%, age or life-stage searches 41.0%, passion or role searches 30.0%, and relationship searches declining 1.5%. These topical groups are not additional word-length buckets and should not be added to the first three rows.

Next year’s questions will be more specific versions of this year’s. Write the facts that stay true: exact age range, activity, use, material, fit.

What lasts when aesthetic trend language fades?#

In six selected comparisons, aesthetic terms declined substantially from their strongest period, while several related plain product terms grew. This is a comparison of chosen pairs, not a forecast for every trend. [C1]

Wide table: swipe horizontally or focus the table and use the arrow keys to see every column.

Selected term pairs: aesthetic peak window versus the 12 months to October 2026
Aesthetic termChangePlain product termChange
Coastal grandmother-93%Linen pants+97%
Glazed donut nails-91%Nail polish+59%
Barbiecore-86%Pink dress-2%
Mob wife aesthetic-83%Faux fur coat+26%
Dark academia-65%Tweed blazer+63%
Quiet luxury-32%Cashmere sweater+47%

For each pair, both terms use the same two windows: the aesthetic term’s best 12 months and the 12 months to October 2026. The source history runs from January 2020 to October 2026 and was pulled on 20 October, so the final month was incomplete.

The 59% nail-polish increase includes a jump in Ahrefs estimates from April 2026. Its calendar-2025 gain alone was 12%. Pink dress also declined, by 2%, so the table does not show all plain product terms rising.

A trend word belongs in a campaign with an end date. The product fact beneath it belongs in the catalog, where it keeps answering after the word fades.

Can stated product facts change an AI recommendation?#

Controlled research shows that listing changes can affect selections in particular experimental settings. It does not establish a universal lift for every retailer or model. The two independent studies below measured different outcomes with different denominators.

Wide table: swipe horizontally or focus the table and use the arrow keys to see every column.

Independent studies: keep outcomes and units separate
Study and outcomeListing or subgroupReported change
Mock storefront: GPT-5.1 selection shareToilet paper: ply, septic compatibility, scent+15.5 percentage points
Mock storefront: GPT-5.1 selection shareStapler: included quantity of staples+9.5 percentage points
Retail field experiment: salesAI-written product descriptions overall+2.1%
English-language experiment: salesListings with under 50 words of original text+6.5%
English-language experiment: salesDetailed listingsNo statistically significant lift

In the mock-storefront study, adding ply, septic compatibility and scent information changed GPT-5.1’s toilet-paper selection share by 15.5 percentage points. Front-loading the included quantity of staples changed stapler selection share by 9.5 percentage points. There were 200 trials per category. These are changes in selection share, not percentage sales gains. [C3]

The separate randomized retail field experiment reported a 2.1% overall sales increase from AI-written product descriptions. In its English-language experiment, products with under 50 words of original text gained 6.5%; detailed listings showed no statistically significant lift. The length subgroup results must remain scoped to that experiment. “No significant lift” is not proof of exactly zero effect. [C4]

AI-written descriptions lifted sales for listings with under 50 words of original text; detailed listings showed no significant lift.

Independent studies, not Lily AI results.

What did the 2026 surveys and forecasts add?#

They described stated shopping intentions and expected AI referral traffic. They did not measure catalog coverage or completed 2026 holiday sales. Retain that distinction when citing the figures.

Wide table: swipe horizontally or focus the table and use the arrow keys to see every column.

2026 holiday survey intentions: different questions and samples
SurveyStated intentionReported share
Deloitte, US n=4,000Generative AI as primary tool for at least one holiday shopping task39%
Deloitte, same surveyWould allow AI to complete a purchase16%
Bain, US 2026 n=1,105Expect to start holiday shopping on GenAI24%, versus 17% in 2025 (n=950)
Adobe September 2026 forecast: year-on-year AI-referred retail traffic growth
MeasureForecast growth
Season130%
Thanksgiving159%
Black Friday95%
Cyber Monday88%

Deloitte surveyed 4,000 US consumers online from 26 August to 10 September 2026. Bain’s US online-shopping samples were 1,105 in 2026 and 950 in 2025, with multiple choices permitted. Their questions and populations differ, so their percentages are not successive stages in one shopping funnel. [C5] [C6]

Adobe’s 28 September 2026 forecast modeled AI-referred traffic to US retail sites, drawing on more than one trillion visits. Referrals count visits from AI services, not all AI interactions or purchases. The forecast is dated context, not a holiday result. [C7]

Who should own the work before, during and after peak?#

Use one recurring-question list and assign accountable owners for prioritization, product evidence and performance measurement. This is Lily AI’s recommended operating plan, not an experimentally validated rollout sequence.

Wide table: swipe horizontally or focus the table and use the arrow keys to see every column.

Recommended ownership before, during and after peak
OwnerBefore the freezeDuring peakJanuary review
Ecommerce leadershipShare the top 100 relevant gift queries; assign category questions.Approve permitted changes; monitor coverage and commercial results.Review answered and missed demand; fund durable attribute work.
Merchandising and catalogVerify age, activity, use and gift facts.Add highlights and details through a Merchant Center supplemental data source, which updates existing products without a site release. Check the live record.Remove expired seasonal copy; keep useful facts.
Performance marketingMap last November and December search terms to the question list.Route missed requests; track impressions, clicks and AI referrals.Report by question and product; inform next season’s brief.

Before the freeze, category leads can map their ten most important recurring questions to products and written evidence. Choose the list from relevant category demand rather than assuming every phrase in this report applies to every retailer. Record the product, question, evidence source, publishing route, owner and verification date.

Before the freeze, agree which feed changes stay allowed. Product highlights, product details and other supplemental data change listings without a site release. Name one owner to check that each change went live.

During peak, log missed questions alongside impressions, clicks, AI referral sessions and downstream outcomes. After peak, remove expired seasonal claims while retaining verified product facts. Use a control where feasible when assessing the effect of enrichment; seasonality and unrelated changes can confound simple before-and-after comparisons.

How was the research conducted?#

Source and measurement

Lily AI’s original research is dated 26 October 2026. All search data is from Ahrefs Keywords Explorer, US, pulled 20 October 2026. Monthly volumes are modeled estimates, not Google’s counts, unique people, purchases or revenue. October 2026 is incomplete, so the seasonal and December comparisons end in December 2025. The trend pairs run to October 2026.

Different cohorts

The 24-query basket describes seasonality. The 598 gift-intent queries describe December 2025 constraints. The filtered 558-query cohort compares consecutive Decembers. The 232 “gifts for” phrases tracked in 2019 measure longer-run growth. The six-year ranking comparison uses 231 phrases. The six illustrative December series below are examples, not a replacement for the full basket.

Exclusive classification

Each query receives one primary constraint in this priority order: situation or feeling; passion or role; relationship; exact age; teenager; other age band; gender; budget; occasion. Coverage maps that constraint to the reviewed specification. It does not measure whether an individual retailer populated a field, whether an agent read it or whether a product was recommended.

Selection and tracking limitations

Recent leading terms favor survivors, especially in the six-year comparison. The consecutive-year ranking excludes year-stamped phrases, zero volume in either December and movements above threefold. “Gifts for kids” is excluded from rank comparisons because of an apparent tracking change, but remains in the main constraint totals with a sensitivity check.

A blind second classification matched all 50 sampled queries. The analysis removed 59 months that fell below 30% of both neighboring months as apparent glitches; none was a holiday November or December. Ahrefs fills some recent months by repeating the previous year, so hostess gifts are excluded from year-on-year comparisons. These checks reduce specific risks without making the selected dataset representative of every shopper.

Data availability

The tables show aggregate values from Lily AI’s analysis of Ahrefs US estimates, pulled 20 October 2026. Raw Ahrefs exports and query-level classification records are not published here. Shares are rounded; the 60.4% is computed from unrounded volumes. The external experiments, surveys and forecast remain separate evidence sets with their own dates and limitations.

Which queries and annual values support the seasonal analysis?#

The 24-query basket contains: gifts for dad who wants nothing; gifts for mom who wants nothing; gifts for someone who has everything; gifts for people who are hard to buy for; white elephant gifts; stocking stuffers; secret santa gifts; hostess gifts; thanksgiving hostess gift; gifts for coworkers; gifts for teachers; gifts for grandparents; gifts for teenage boys; gifts for teenage girls; gifts for boyfriend; gifts for girlfriend; gifts for him; gifts for her; gifts for mom; gifts for dad; christmas gift ideas; gifts under 25; gifts under 50; friendsgiving.

The six series below are December estimates. “Gifts for people always cold” is short for “gifts for people who are always cold”. December values differ from the November and December totals in the seasonality table above. The query “white elephant gifts” here is distinct from “white elephant gift ideas” in the leading-phrases table. [C1]

Wide table: swipe horizontally or focus the table and use the arrow keys to see every column.

Six example phrases: December estimated US searches, 2016 to 2025
Phrase2016201720182019202020212022202320242025
Gifts for dad431,220443,760401,627478,438519,174435,977427,149444,066440,261463,044
White elephant gifts163,531184,673116,960171,35665,293162,297151,292171,704179,993171,779
Gifts for coworkers31,83937,14539,32049,46138,66248,75853,17950,54945,73744,617
Gifts for runners19,46620,47017,82523,20020,45518,17022,04624,88725,45224,618
Gifts for teenage boys20,90920,53019,08019,05620,94318,80018,96321,62821,62521,603
Gifts for people always cold4,2294,2344,6673,7613,6313,6344,0923,9513,3765,177

Sources#

  1. C1. Lily AI Research: search data

    Original research dated 26 October 2026. Lily AI analysis of Ahrefs Keywords Explorer US estimates, pulled 20 October 2026: 24 holiday searches, November and December 2016 to 2025; 598 gift searches, December 2024 and 2025; 232 “gifts for” phrases, 2019 to 2025; six aesthetic and product word pairs, January 2020 to October 2026. Raw exports are not published here.

  2. C2. Google Merchant Center: product data specification

    Google Merchant Center Help: the product data specification and the age_group, product_highlight, product_detail and question_and_answer pages. Checked 29 October 2026.

  3. C3. Allouah, Besbes, Figueroa, Kanoria and Kumar: What Is Your AI Agent Buying?

    arXiv 2508.02630v3, revised 17 December 2025. Independent controlled study; 200 trials per category. Selection-share changes are percentage points.

  4. C4. Fang, Yuan, Zhang, Donati and Sarvary: Generative AI and Sales Productivity

    Field Experiments in Online Retail. arXiv 2510.12049v6, 29 June 2026. The description-length results come from its English-language experiment.

  5. C5. Deloitte: 2026 Holiday Retail Survey

    4,000 US consumers, online, 26 August to 10 September 2026. Figures are stated intentions.

  6. C6. Bain: 2026 holiday shopping outlook

    US online shoppers: 2026 n=1,105; 2025 n=950. Multiple choices permitted.

  7. C7. Adobe: US holiday shopping forecast

    28 September 2026. AI-referred traffic growth is a forecast, not an observed 2026 holiday result. Model based on more than one trillion US retail-site visits.

Where Lily AI fits#

Lily Max is the agentic product intelligence engine for AI commerce. It enriches retail product catalogs and measures results against a control across Google Ads, Meta Ads, AI shopping assistants and onsite search.

A free feed audit looks at your Google Merchant Center feed and shows where richer product data can lift Shopping and AI-discovery performance.

Get a free feed audit

The complete report is freely readable. You do not need the feed audit to read any part of this report.