Year-over-Year in Q4: Black Friday Moves on the Calendar

The calendar moves the sale days

Black Friday has no fixed date. It follows Thanksgiving, and US federal law places that holiday on the fourth Thursday in November (5 U.S.C. § 6103). In 2025 that was November 27; in 2026 it is November 26 (OPM). The German Retail Federation named November 28 as Black Friday and December 1 as Cyber Monday for 2025 (HDE). For 2026 this gives Friday, November 27, as Black Friday and Monday, November 30, as Cyber Monday.

Exactly 364 days, or 52 full weeks, lie between the two Black Fridays. Seen by weekday, everything lines up: Friday meets Friday, Monday meets Monday, and both sale days fall into the same calendar week in both years, Black Friday in week 48 and Cyber Monday in week 49. By calendar date, nothing lines up. Comparing November 27, 2026 with November 27, 2025 sets the sale day against the Thursday before, which may still have been an ordinary trading day in your own shop. The following table shows the five days around the sale in both readings.

At day level this produces four swings that have nothing to do with the business. On November 27, 2026 you see phantom growth, because the sale day competes with a day without a sale. On November 28 you see a phantom decline, because an ordinary Saturday competes with the previous year's Black Friday. On November 30 and December 1 the same pattern repeats with Cyber Monday. Anyone who copies this daily series into a report unchecked hands management the question of why Cyber Monday collapsed, when it merely fell one day earlier.

At week level the shift is trickier, because it only occurs sometimes. Comparing by calendar week works in 2026, because week 48 contains Black Friday in both years. That is a coincidence of numbering: 2026 has 53 calendar weeks, and in 2027 Black Friday falls on November 26 and therefore in week 47. Comparing a fixed date range such as November 20 to 30 with the same date range of the previous year, on the other hand, yields two sale days in 2026 against one. None of the three readings is miscalculated, but only one compares like with like.

Monthly reports: Cyber Monday changes months

The shift is most visible in the monthly report. In November 2025 only Black Friday fell into the month; Cyber Monday counted towards December. In November 2026 both sale days fall into the same month. A November-over-November comparison therefore counts one whole sale day more in 2026, and the December-over-December comparison one less. Both monthly figures are arithmetically correct and misleading in substance, and they err in opposite directions.

The weight of that single day is not small. For 42 percent of respondents, Cyber Monday is an occasion to shop for the holidays, according to a survey commissioned by the retail federation (HDE). Holiday purchases booked on December 1 in 2025 land on November 30 in 2026, and therefore in the previous month. The size of the sale business and its consequences for capacity planning are covered in preparing for Q4 traffic peaks; this article is about reading the numbers correctly afterwards.

It is worth noting how retail itself does the maths. The retail federation compares the sale days not date against date but event against event: since data collection began in 2016, spending on Black Friday and Cyber Monday has risen continuously, at times by more than twenty percent compared with the respective previous year (HDE). Your own shop report needs exactly this logic.

Then there is the make-up of the month. November 2026 has five Mondays and four Saturdays; November 2025 had four Mondays and five Saturdays. One of the extra Mondays is Cyber Monday. A shop that sells strongly on Saturdays and weakly on Mondays sees a shift even without any sale, and it comes from the calendar alone.

How official statistics remove the calendar

The German Federal Statistical Office publishes two kinds of year-over-year comparisons for every month, and November 2025 shows how far apart they can be. Calendar and seasonally adjusted, retail sales rose by 1.1% in real terms and 1.9% in nominal terms compared with November 2024 (Destatis). The table in the same press release, calculated from the original values without calendar and seasonal adjustment, shows −1.8% in real terms and −1.0% in nominal terms for the same month (Destatis). The same month and the same sector produce a plus once and a minus once.

Internet and mail-order retail shows the same gap between the two readings: adjusted, +5.9% in real terms compared with the prior-year month; from the original values, +2.9% in real terms (Destatis). The press release does not say why the gap is so large. The calendar offers an obvious candidate: November 2025 had five Saturdays and five Sundays, November 2024 five Fridays and five Saturdays. Counted from Monday to Saturday, the month had one trading day fewer in 2025.

These are exactly the effects the procedures are meant to remove. According to Destatis, calendar and seasonal adjustment procedures such as the Berlin procedure (BV 4.1) or X13JDemetra+ serve to remove effects such as the number of trading days and public holidays from the index figures (Destatis). Calendar adjustment targets fluctuations in the number of working days, trading days or particular weekdays (Destatis). The office states explicitly what the adjusted year-over-year comparison is for:

The calendar-adjusted year-over-year comparison serves as a longer-term level comparison and is independent of seasonal fluctuations and calendar effects.

— Federal Statistical Office (Destatis), press release No. 004 of 7 January 2026, translated from German

Eurostat describes the calendar effect as the variation caused by the changing number of particular weekdays or holidays in different months, quarters or years (Eurostat). For an online shop, Black Friday is exactly such a moving day, only with far more weight than an ordinary weekday. A shop does not need a statistical adjustment procedure for this. It is enough to choose comparison periods so that sale day meets sale day and weekday meets weekday.

Structural breaks and revisions: what Destatis discloses openly

Besides the calendar, official statistics explicitly name a second source of error: the structural break. With the reporting month of October 2025, Destatis replaced the sample of reporting retail businesses, and the new sample takes effect retroactively from the reporting month of January 2024 (Destatis). To keep results comparable despite different samples, the index figures are chained forward (Destatis). In addition, the affected time series have shown a structural increase since the reporting month of August 2024 because a larger company in internet and mail-order retail was restructured (Destatis). How this special effect feeds into planning for the holiday season is described in Q4 readiness for online shops.

The second lesson is revision. For October 2025, Destatis initially reported −0.3% in real terms compared with the previous month; after revision of the provisional results the figure is an increase of 0.3% in real terms (Destatis). Provisional results are usually published 30 days after the end of the reporting month, and 45 days after the end of the month the data in the GENESIS-Online database are updated (Destatis). The direction of a monthly figure can therefore flip after the fact.

A shop has the same two problems on a smaller scale. The revenue of a sale day is not final the day after: cancellations, returns and refunds arrive days or weeks later. If you report refunds through tracking afterwards, you only see the sale day at its final level then. And every change to the measurement itself is a structural break, even if nobody calls it that. Typical cases in your own analytics:

• Relaunch or new theme: different page structure, different checkout steps, often a different measurement setup too.
• New payment method: shifts the checkout completion rate and sometimes the basket value.
• Campaign start: a newsletter or an ad before the sale day pulls demand forward; clean campaign parameters keep this apart, see campaign tagging hygiene.
• Filters for automated traffic: tightening the bot filter lowers the visitor count without a single customer fewer, see filtering bot traffic in analytics.
• Consent and measurement method: a new consent banner or a switch to cookieless analytics changes which visits are counted at all.
• Browser data: frozen user agent strings change device and version reports, see frozen user agent strings and device reports.
• Price and shipping changes: a new free-shipping threshold shifts the basket value without any change in demand.

Setting up a weekday-aligned comparison in XICTRAQ

In XICTRAQ Analytics, the comparison sits in the overview. The comparison control offers three modes: previous period, previous year and a custom range from–to. The selected comparison applies to every card at once; KPI cards show the delta to the comparison period, breakdown cards a comparison pill in the header. Important for Q4: the previous year mode compares by calendar date. For Black Friday this means: for weekday-aligned sale weeks, set your own comparison period. The details are in the analytics documentation.

1. Set the main range to the 2026 sale week: Monday, November 23, to Sunday, November 29, 2026.
2. In the comparison control, choose "Custom" and enter the previous year's sale week: Monday, November 24, to Sunday, November 30, 2025.
3. Compare Cyber Monday separately: November 30, 2026 against December 1, 2025. In both years it falls directly after the Monday-to-Sunday week.
4. Compare the sale weekend from Friday to Monday as its own block: November 27 to 30, 2026 against November 28 to December 1, 2025.
5. Read the days after the sale by weekday as well: Tuesday, December 1, 2026, against Tuesday, December 2, 2025.

The rule of thumb behind this: a weekday-aligned prior-year period lies 364 days back, not one calendar year. This makes every comparison period quick to work out, including for sales that are not tied to Thanksgiving, such as an anniversary weekend that always falls on the last weekend of a month. Conversely, days tied to a date, such as Christmas Eve or New Year's Eve, are compared by date. There, the previous year mode is exactly right.

These metrics belong in the same comparison as visitors and page views because they explain each other. A revenue increase with a flat order count says something different from an increase driven by more orders with smaller baskets. Which other metrics a shop should watch is covered in e-commerce monitoring for online shops; how orders, revenue and basket value are recorded without tracking cookies is shown on the page analytics for online shops.

Recording sale days and special effects as annotations

An adjusted comparison explains the number; an annotation explains the curve. In single-website mode, XICTRAQ lets you add trend annotations, event markers placed directly on the trend chart, for example for a relaunch, a campaign start or a price change. For Q4, at least three kinds of markers belong there: the sale days themselves, the special effects in your own shop and the changes to the measurement. An annotation plan for one website can look like this:

# Example: trend annotations of one website for the Q4 comparison
2025-11-28  Black Friday, sale Friday to Monday
2025-12-01  Cyber Monday
2026-09-14  New payment method in checkout
2026-10-19  Relaunch of category pages
2026-11-23  Campaign start sale week, newsletter
2026-11-27  Black Friday, sale Friday to Monday
2026-11-30  Cyber Monday

Two rules make annotations useful. First, the prior-year period needs markers too, otherwise the comparison curve shows swings without explanation. Second, the date belongs in the time zone in which the sale runs. A sale that starts at midnight German time begins at 23:00 UTC on the previous day in November; if you evaluate day boundaries in UTC, you push an hour of sale revenue into the wrong day. The background is covered in daylight saving time and time zones in monitoring.

Report adjusted comparisons only

A comparison only belongs in front of management or clients once it accounts for the calendar. The Federal Statistical Office shows what that looks like: its business cycle observation focuses on comparing calendar and seasonally adjusted values with the previous month or quarter, and the original values still sit next to them in the table (Destatis). A shop report can adopt the same form: the adjusted comparison first, the raw value next to it, the method in one sentence. What belongs in a Q4 report:

• The method in one sentence, for example: sale week November 23 to 29, 2026 against November 24 to 30, 2025, Cyber Monday separately.
• The weekday-aligned comparison for revenue, orders and basket value; the raw value by calendar date at most as a second column.
• The special effects of both years with dates: relaunch, new payment method, campaign start, changes to filters and measurement.
• A "provisional" note as long as cancellations and refunds are still coming in.
• For monthly figures, the note that Cyber Monday falls in November in 2026 and fell in December in 2025, or a combined value for November and December.

XICTRAQ supports this in two ways. The annotations report section takes timeline markers such as deploys and campaigns into the report as a storyline, so sale days and special effects appear in it with their dates. The comparison section of the automatic report calculates against the previous period; you therefore set up the weekday-aligned sale comparison in the analytics overview and carry it into the summary. How reports are put together and sent is described in the reports documentation, and an overview is on the page reporting and status pages.

If you send reports to clients regularly, define the rule once and apply it every year. How to set up recurring client reports is covered in automating weekly client reports. Agencies that look after several shops keep access and reports separate per client, see XICTRAQ for agencies. A shared standard pays off there in particular: when every client receives the same method line in December, nobody asks why Cyber Monday shows up in November for one client and in December for another.

Easter, holidays, clock changes: the same discipline all year

Black Friday is only the most conspicuous case. Eurostat cites the timing of certain public holidays as an example of the calendar effect: Easter can fall in March or in April, depending on the year (Eurostat). In 2026 Easter Sunday fell on April 5; in 2027 it falls on March 28. A quarterly comparison of Q1 2027 against Q1 2026 therefore contains Easter trading on one side only, and the April comparison tips the other way. On top of that come public holidays that apply only in individual German states, bridge days, and the clock change, which creates a 23-hour day in spring and a 25-hour day in autumn.

The rule is the same in every case: before comparing, clarify whether the days in question are tied to a date, to a weekday or to a movable feast, and choose the comparison period accordingly. Date-bound days are compared by date, weekday-bound days by weekday, movable feasts feast against feast. A comparison that does not answer this question measures the calendar as well.

Checklist for the Q4 year-over-year comparison

• Note the sale dates of both years: Black Friday on November 28, 2025 and November 27, 2026, Cyber Monday on December 1, 2025 and November 30, 2026.
• Set up weekday-aligned comparison periods, 364 days apart.
• Compare Cyber Monday and the sale weekend separately.
• Add a note to monthly reports for November and December or combine them into one value.
• Annotate sale days, relaunch, new payment method and campaign start in both years.
• Mark changes to measurement, consent and bot filter as structural breaks.
• Read revenue, orders and basket value in the same comparison.
• Label sale figures as provisional until cancellations and refunds have been booked.
• Before the sale week, check that checkout and the withdrawal path work, see checking the withdrawal button during peak season.