Halloween 2026 in numbers: 47% of shoppers used AI, and online search hit a 10-year low for costume ideas

Halloween is the first test of the holiday shopping season, and this year it came with a new question. The National Retail Federation and Prosper Insights & Analytics asked 7,889 consumers how they planned to shop for the holiday, and for the first time the headline findings included AI.

Nearly half, 47%, said they are using AI platforms and tools to support their Halloween shopping. At the same time, online search fell to its lowest level in a decade as a source of costume ideas.

Those two numbers come from the same survey, fielded September 1 to 9 with a margin of error of plus or minus 1.1 percentage points. We do not think they prove that assistants took share from search engines. They do show that a seasonal shopping trip now often starts with a conversation, and that matters to the brands that sell the one thing almost every Halloween shopper buys.

The headline numbers

NRF expects Halloween spending to reach $13.5 billion in 2026, keeping pace with last year's record of $13.1 billion. Average spending is expected to be $115.14 per person, in line with $114.45 in 2025. About 74% of consumers plan to celebrate, compared with 73% last year and 68% in 2019.

Table: NRF Halloween survey, key figures for 2026

Measure	2026	Comparison
Total spending	$13.5 billion	$13.1 billion in 2025 (record)
Spending per person	$115.14	$114.45 in 2025
Plan to celebrate	74%	73% in 2025, 68% in 2019
Start shopping in September or earlier	49%	34% in 2016
Using AI tools for Halloween shopping	47%	First reported this year
Online search as top costume idea source	32%	37% in 2025

The early start is notable for brands. Half of consumers said they would begin shopping in September or earlier, up from about a third in 2016. NRF says 42% of those early shoppers do it because they look forward to fall, and 23% to spread out their budgets.

Where the money goes

Costumes, decorations and candy each account for roughly a third of spending. Candy is the smallest of the three in dollars but by far the broadest: 96% of Halloween shoppers plan to buy it.

Show the data

Category	Spending, $ billions
Costumes	4.4
Decorations	4.3
Candy	4.1
Greeting cards	0.8

Source: NRF and Prosper Insights & Analytics, September 22, 2026. Share of shoppers buying each: candy 96%, decorations 78%, costumes 71%, greeting cards 42%.

Greeting cards are the small riser, with 42% planning to buy them, up from 38%. Within costumes, NRF expects $2 billion on adult costumes, $1.5 billion on children's costumes and $0.92 billion on pet costumes.

Where people shop

Discount stores remain the top destination at 39%, followed by specialty Halloween and costume stores at 32% and online at 27%. Local and small businesses reached 13%, almost double the 7% of 2016.

Show the data

Destination	Value
Discount stores	39%
Specialty Halloween stores	32%
Online	27%
Local or small businesses	13%

Source: NRF and Prosper Insights & Analytics, September 22, 2026. Consumers could choose more than one destination.

For packaged goods, this mix is a reminder that most seasonal purchases still happen in physical stores. An AI assistant does not need to complete a purchase to shape it. If a shopper asks for nut-free treats for a classroom party and then drives to a discount store, the answer still decided which brand went on the list.

The AI numbers, read carefully

NRF's wording is that 47% of consumers "are using AI platforms and tools to support their Halloween shopping this year." The two top uses were brainstorming and coming up with ideas while planning, at 34%, and asking for recommendations or ideas during product research, at 32%.

Prosper's Phil Rist described the pattern this way:

As consumers prepare for Halloween, many are using these tools to brainstorm costumes and decorating themes, explore production options and find inspiration that help bring their celebrations to life.

Two cautions. First, "AI platforms and tools" is broad. It can include a chatbot, an AI feature inside a retailer's app or AI summaries in a search engine. The survey release does not split them out. Second, 47% using AI at some point is not 47% of purchases influenced by AI. The release does not measure that.

Even so, 32% asking for recommendations during product research is a large share of shoppers. On an occasion where 96% buy candy, a meaningful number of those recommendation requests will be about treats: allergy-friendly options, full-size versus fun-size, how much to buy for a given number of trick-or-treaters.

What candy shoppers might ask

The survey does not break AI use out by category, so we cannot say how many of the 47% asked about candy specifically. But the shape of the holiday suggests the kinds of questions that come up. Candy is bought by almost everyone, often in bulk, often for other people's children. That raises practical questions: which treats are free of the top allergens, how many pieces come in a bag, how much to buy for a street with a lot of trick-or-treaters, which options are individually wrapped. Each of those has a factual answer that lives on a package and, ideally, on a product page.

We have written before about why assistants struggle when that information only exists in images. A count per bag, an allergen statement and a serving size are exactly the details a shopper asks an assistant to compare, and exactly the details many listings leave in a photo of the back panel. For a seasonal product sold for a few weeks a year, there is little time to correct a wrong answer once the season starts.

Search slipped, but did not disappear

Online search is still the leading source of costume inspiration at 32%. That is down from 37% last year and, NRF says, the lowest level in a decade. Retail stores and costume shops came second for costumes at 26% and first for decorations at 31%.

We would not draw a straight line from the AI number to the search number. The survey does not ask whether people replaced search with an assistant. But the direction fits what other data shows this fall. Adobe reported on October 29 that AI-referred traffic to US retail sites grew 71% year over year in September and converted 90% better than other traffic. We break that report down in (post:adobe-ai-traffic-converts-90-percent-better-holiday-2026).

What Halloween tells CPG brands about the next eight weeks

Halloween is a dress rehearsal. The questions are seasonal, specific and time-bound, which is the kind of question people bring to an assistant. Thanksgiving and the December holidays bring more of them, about recipes, hosting, gifts and stocking up.

Here is what we would take from the survey.

Seasonal questions need seasonal answers on your pages. An assistant asked for "candy without peanuts for a school party" can only recommend your product for that need if a page it can read says so plainly. Tip (tip:20) covers writing for specific needs. If your allergen information lives only in a package photo, it is not there for the assistant.

Starting early is now normal. With half of shoppers starting in September, seasonal product pages and retailer listings should be live and accurate well before the occasion. Assistants also favor fresh, dated information, which (tip:23) explains.

Value questions are coming. NRF found 31% will comparison shop if prices are higher than expected and 26% will look for coupons or sales. Comparison is exactly what assistants are good at. Make sure your count, weight and price per unit are easy to read on every retailer listing, so a comparison works in your favor rather than skipping you for missing data.

Check the answer, not just the ranking. Ask an assistant the questions your shoppers would ask for the next occasion, such as Thanksgiving baking or holiday hosting, and read what comes back. Some teams track these answers with tools such as SearchPanel from XENA Intelligence; a spreadsheet of ten questions checked weekly works too.

The takeaway

Halloween spending held at a record level, and the shape of the shopping trip is changing underneath it. Nearly half of shoppers used AI tools to plan, and the long-time default for ideas, online search, fell to a 10-year low.

For brands, the lesson from today's holiday is simple. Seasonal shoppers ask specific questions early. Whoever has the clearest, most current answer on a page an assistant can read has a better chance of being the brand that ends up in the cart. If you want a fuller checklist for the weeks ahead, our (post:holiday-2026-ai-shopping-playbook-cpg) covers it.