The First Click Is in October. The Sale Is Not
Fifty two days separate the first click from the holiday sale it eventually becomes. Most seasonal budgets still switch on in November, which is roughly a month after the hunt began.
By Katie Delaney · 2026-09-07 · 11 min read
Holiday advertising now runs on a fifty two day clock
the average holiday conversion journey measured by Microsoft Advertising
The most useful number published this week is dull, precise and awkward for anyone who plans holiday advertising the way it was planned five years ago. Microsoft Advertising reports that the average holiday conversion journey now takes 52 days.
Holiday advertising has always been a waiting game, but the wait now starts earlier than most media plans admit. Fifty two days from a shopper's first click to the purchase it eventually produces. Count backwards from Black Friday and the scent picks up in early October. Count backwards from a mid December delivery deadline and it starts in late October. Either way, a campaign that switches on in November is joining a hunt already in progress.
Anyone buying holiday advertising on a November flight is therefore paying to re-enter a conversation that began without them, which is the most expensive way to arrive anywhere.
The same research puts numbers on where that first click lands. October is the first click for 56 per cent of November conversions and 35 per cent of December conversions, and 63 per cent of United States shoppers start before Halloween.
The methodology, because it is the only trail worth following
Take that seriously partly because the sample is disclosed. Microsoft names its source as the GMI 2026 Holiday Survey and its own first party consumer trends research, fielded in July 2026, with 1,024 respondents in the United States, 519 in the United Kingdom and 310 in Australia. That is a real panel with real limits, and it is stated rather than implied.
Compare that with the forecasts flying around every autumn. Salesforce predicts that 20 per cent of holiday ecommerce traffic will come from artificial intelligence chat agents this year. It is an interesting expectation and folkfox is treating it as exactly that: the page cites an internal shopping index without disclosing sample size, scope, or whether the underlying data is survey or behavioural. An expectation is not a measurement, and christmas advertising budgets should not be moved on one.
This is not pedantry. Holiday advertising decisions made on undisclosed samples are how a category ends up moving in a herd, and a herd is easy to outfox. The brand that reads the methodology gets a quieter, cheaper October.
Even the specialists undershot the scent of the new channel
Here is the part that should unsettle any planner who thinks last year's assumptions still hold. In October 2025, Adobe forecast a $253.4 billion United States online season from 1 November to 31 December, growing 5.3 per cent, and expected artificial intelligence traffic to rise 520 per cent.
In January, Adobe reported what actually happened: $257.8 billion, up 6.8 per cent, with traffic to retail sites from generative artificial intelligence tools up 693.4 per cent. Forbes noted the season beat the forecast.
| Measure | Adobe forecast, Oct 2025 | Adobe measured, Jan 2026 |
|---|---|---|
| United States online spend, 1 Nov to 31 Dec | $253.4bn | $257.8bn |
| Year on year growth | 5.3% | 6.8% |
| Generative AI traffic to retail sites | up 520% | up 693.4% |
A specialist forecaster with a trillion site visits behind it undershot the artificial intelligence channel by roughly a third. That is not a criticism of Adobe, whose method is published and whose overall number was close. It is a warning about the confidence of anyone whose holiday marketing campaigns assume this channel is behaving the way it did last season.
The lesson for holiday advertising is not that forecasts are useless. It is that the newest channel is the one the forecast will miss by the widest margin, every time, because there is least history under it.
Adobe's method is stated plainly: analysis covering over one trillion visits to United States retail sites, 100 million product listings and 18 categories. That is behavioural data rather than a survey panel, which makes it a genuinely independent second reading beside the Microsoft numbers rather than an echo of them.
Two more measured facts sit alongside it. Twenty five separate days cleared $4 billion in a single day, against 18 such days the previous year, so the peak is flattening into a plateau, and holiday advertising built around two big days is chasing a quarry that has already scattered across the hedgerow. And flexible payments contributed $20 billion, up 9.8 per cent, which is $1.8 billion more than the previous season.
The artificial intelligence channel is real, and it does not convert yet
The temptation, reading a 693 per cent traffic rise, is to shift budget towards the new channel immediately. The undergrowth here is thicker than the headline suggests, and a practitioner voice is worth more than a projection.
160 ChatGPT Ads clicks in Spain, EUR 37.61 spend, 0 purchases. The bigger concern is traffic quality: many recorded sessions are only around 1 to 10 seconds.
That is one advertiser, one market and a small spend, so it settles nothing on its own. It is still the sort of detail that never appears in a trend deck: clicks arriving, sessions lasting seconds, nothing converting. Traffic growth and revenue growth are different measurements, and the gap between them is where a quarter's budget can quietly disappear.
Adobe's own survey work found more than a third of shoppers had used an artificial intelligence service for online shopping, with the leading uses being research at 53 per cent, product recommendations at 40, finding deals at 36 and gift inspiration at 30. Read that list again. Every one of those is an upper funnel behaviour. None of them is a purchase.
That distinction should shape how holiday advertising money is split. Research surfaces earn budget when they demonstrably feed the closing channel, not when their traffic chart looks dramatic.
Microsoft's own reading points the same way from the other side: in Europe, the Middle East and Africa, shoppers who use artificial intelligence tools are 12 points more likely to use search and 9 per cent more likely to convert within 24 hours. The tool feeds the search, and the search closes. That is an argument for funding both, not for moving money from one to the other.
Holiday advertising in regulated dens runs a tighter calendar
Every planner faces the 52 day problem. Advertisers in awkward categories face it with a policy clock running underneath, and this year the two collide.
Google's personalised advertising policy update takes effect on 30 October 2026, opening alcohol to personalised advertising on some inventory, with named exclusions in Egypt, India, Indonesia and Poland. That date lands after the first click for most November conversions has already happened.
Regulated brands buying holiday advertising also carry an approvals queue that consumer brands do not, and legal review does not accelerate because a season is short. Build the creative against the earlier date and hold it.
A regulated advertiser therefore has two dates, not one. The commercial date is early October, when the black friday advertising trail actually opens. The compliance date is 30 October, when a channel becomes available. Building the creative, the landing pages and the approvals in the gap between them is the whole job.

For scale, Deloitte forecast the previous season at $1.61 trillion to $1.62 trillion across November to January, with ecommerce growing 7 to 9 per cent. Note the window: Deloitte counts three months, Adobe counts two, so the two figures are not interchangeable and should never be set against each other as though they were.
What a planner does on Monday
Strip out the forecasts and the week leaves four instructions, each of them boring and each of them cheap.
Move the budget start date to the first week of October, not the first week of November. Re-set attribution windows to at least 52 days, because a 30 day window is structurally incapable of seeing the journey Microsoft measured. Fund the artificial intelligence surface as research and measure it on assisted conversions. And separate the compliance date from the commercial date so neither is discovered late.
Each of those four is a scheduling change, not a spending one, which is why holiday advertising rewards planning far more than it rewards budget. The den is dug in October or it is not dug at all.
None of that requires a bigger budget. It requires the same budget to arrive earlier and be measured over a longer window, which is a scheduling decision rather than a spending one. Holiday advertising rewards the patient prowl, and punishes the brand that waits for the obvious moment.
A thirty day attribution window cannot see a fifty two day journey. It is not measuring conservatively. It is measuring the wrong thing.
Holiday shopping trends move every season, and this one moved hardest on a channel nobody could measure two years ago. The dens that hold are the ones built on stated methodology: a sample size, a date, a definition. Everything else is a nice slide.
Read the numbers folkfox tracks each morning, the paid search work on the PPC page, the paid social side on paid social, and the answer engine side on SEO and GEO. Rates sit on the pricing page and the method sits with content marketing. Christmas advertising is won in October, quietly, by whoever moved first. Holiday advertising is not a November discipline, whatever the calendar on the wall says.
Frequently asked questions
When should holiday advertising actually start?
Early October for most retailers. Microsoft Advertising measures the average holiday conversion journey at 52 days, and finds October carries the first click for 56 per cent of November conversions and 35 per cent of December ones. A campaign that starts in November is entering a journey already a month old.
What is the best holiday marketing campaign approach this year?
Start earlier, measure longer and treat artificial intelligence surfaces as research rather than closing channels. The measured evidence supports an October start, an attribution window of at least 52 days, and funding the AI channel in proportion to its revenue rather than its traffic growth.
Is AI traffic worth budget this holiday season?
It is worth measurement before budget. Traffic from generative AI tools to retail sites rose 693.4 per cent last season, well above the 520 per cent forecast, but the leading uses are research, recommendations and deal finding, all upper funnel. Fund it as a research surface and watch assisted conversions.
Why do holiday advertising keywords show falling search trends in September?
Because the measurement is taken before the season. Seasonal terms show steep negative year on year trend figures when sampled in early autumn and climb sharply from October. Treat an out of season trend line as a timing artefact, not as evidence of declining demand.
How long should the attribution window be for christmas advertising?
At least 52 days, and longer if your considered purchase is expensive. A 30 day window cannot observe a 52 day journey, so it will systematically credit the last touch and understate everything that started the hunt in October.
Do regulated advertisers face a different holiday timetable?
Yes. Alongside the commercial calendar there is a policy calendar. Google's personalised advertising policy update takes effect on 30 October 2026, after the first click for most November conversions has already occurred, so creative and approvals have to be ready before the channel opens.
Three Studies Measured the Same Ad Load and Disagreed by Thirteen Times.
When three measurements of the same thing disagree, the denominator is usually the problem.
Read the pieceThe Owner Stays . The Auction Does Not
The auction mechanics behind your Q4 buy are being rewritten by a court, not a product team.
Read the pieceMachines Read More of It. People Trust Less of It
The other half of the AI traffic story, and what happens to trust when machines are the readers.
Read the pieceThe Price Is Real. The Bill Is Somewhere Else
If you are budgeting an AI line for 2027, the rate you are using probably expires in December.
Read the piecePlanning a Q4 that has to start in October?
folkfox builds seasonal plans on measured journeys and stated methodology, not on forecasts. We set the dates, the windows and the compliance calendar in one document.
Want folkfox in your Google results and AI answers? Set folkfox as a preferred source.

