The ChatGPT Shopping Calendar: What 2025 Says About Your Q4 Budget

OpenAI's September 2026 stat pack includes three slides of 2025 US product-search activity inside ChatGPT, indexed against a matched-week baseline. Back-to-school built for five weeks to a 26% average lift in August. Labor Day week reached +24%. Cyber 5 averaged +47%, Black Friday peaked at +94%, and activity was back to baseline by 3 December. It is the first seasonality curve anyone has for the channel, and it is enough to plan a Q4 around.

What the three windows show

Three features of the curves matter more than the headline numbers. Back-to-school is a ramp: the index fell to about 90 in the last week of July and climbed without a pause to about 157 by 1 September, so the demand arrived over five weeks. Cyber 5 is the opposite: activity was already 20 to 33% above baseline in the four days before Black Friday, spiked to roughly 193 on the day, dropped to about 130 on the Sunday, bumped again to about 154 on Cyber Monday and was at 105 by 2 December. And in every window the "share" line, shopping as a share of all ChatGPT activity, moved with the "activity" line, which means shopping grew faster than overall usage. People were using ChatGPT more, and using it to shop disproportionately.

The small print, before the plan

The data is US only, from 2025, and it measures product-search activity inside ChatGPT rather than ad impressions or purchases. The lifts are relative to the preceding four weeks, so the back-to-school +26% sits on top of a July trough. Ads began serving in Europe on 24 August 2026, so there is no European curve, and European Black Friday behaviour is a different shape in most categories anyway. Read the US curve as the planning prior for timing, and treat your own 2026 numbers as the first European baseline. The benchmarks OpenAI published alongside it carry the same qualifier: observational, with the sample stated.

The week-by-week plan for a D2C brand

The plan follows from one fact about the platform: conversion bidding needs conversion history, and OpenAI's own benchmark counted only advertisers with at least five click-attributed conversions on the event. Everything before November exists to have that history in place when the curve turns up.

• Mid-October: pixel and Conversions API installed together, events deduplicated with shared IDs, diagnostics panel checked. The Conversions API alone added 57% to measured conversions in OpenAI's data across more than 1,000 advertisers.

• Last week of October: a click campaign live with 100 reviewed ads across distinct situations, context hints written for the buyer's Q4 questions, and a product feed if the catalogue clears the roughly 1,000-product floor. Feed-active accounts showed a 74% higher click-through rate in OpenAI's sample of more than 10,000 advertisers.

• First two weeks of November: accumulate. Five conversions on the primary event is the minimum, more is better. Clone the click campaign into a Conversion Optimization campaign once the floor is cleared and run both side by side.

• Week before Thanksgiving: budgets loaded. Activity was already above baseline four days out in 2025, and the auction on Black Friday itself is the most expensive hour of the year on every channel.

• 27 November to 1 December: hold the budget across all five days. Black Friday is the peak, Cyber Monday is the second peak, the Sunday between is soft.

• 2 to 3 December: pull back. The 2025 index was at 105 on 2 December and at baseline by the 7th. Money spent in the second week of December buys the January curve, and January is the seasonal trough in every region OpenAI charted.

Which categories to plan for

Personal care led both late-summer windows and appears in the consumer goods and beauty category volumes, which OpenAI puts at over 450 million weekly messages growing 44% since January. Apparel is in all three windows. Computers, electronics and appliances own Cyber 5 at +94%. If you sell in one of those three, the seasonality data is a direct argument for the channel. If you sell in a category that does not appear, the general shape still applies, and the reach calculator below sizes the demand for your own situation. Best Buy, one of the named pilot advertisers in the same pack, reports that the channel uncovered demand across a broader mix of categories than it expected, so treat the slide categories as a starting list.

What this is worth

Early auctions are thin auctions. The US market had more than 10,000 advertisers above $100 of monthly spend by July 2026, according to the eligibility footnote in OpenAI's own benchmark. Europe's auction is weeks old. A brand that has tracking, creative and a first conversion history in place by the end of October enters the most valuable five days of the year with a bidder that knows what to buy. One that starts in November enters it with a click campaign and a hope. The reach calculator sizes your situation, and the Conversion Optimization guide has the switch itself.