Black Friday moved from November 23 last year to November 29 this year, so same-date comparisons compare different days. Compare Black Friday with Black Friday, weeks by their distance from it, and write down what else changed.

Last Saturday, a year ago

Last Saturday was November 23. A year earlier, November 23 was Black Friday.

Anyone who compared last week's figures with the same dates last year set an ordinary week in late November against the biggest shopping weekend of 2018. The result looked like a collapse.

It was not. US Thanksgiving falls on the fourth Thursday of November, Black Friday follows it, and this year that puts it on November 29, six days later than last year.

Reports that line up dates will mislead in both directions for the next few weeks, unless the events are lined up instead.

Where the calendar misleads

The third catches careful people. Matching the day of the week compares each day with the one fifty-two weeks earlier, but this year's Black Friday comes fifty-three weeks after last year's.

Compare the event with the event

Black Friday with Black Friday: November 23 last year against November 29 this year. Cyber Monday with Cyber Monday: November 26 against December 2.

The weekend between them works the same way, and so do the days before. The week leading up to Black Friday is compared with the week leading up to last year's, whatever the dates.

Use the same kind of week in both years. Monday to Sunday is the usual choice, which makes the week containing Black Friday the event week.

Put that way, every week of the season has a name rather than a date: two weeks before, the week before, the event, the first week after. Compare names with names.

Day of the week still matters

Outside the event, matching the day of the week is the right default. A Saturday compared with a Thursday mostly tells you about Saturdays.

Universal Analytics has a match day of week option when comparing date ranges, and for ordinary weeks it should be ticked.

For the event itself, set the comparison dates by hand. The automatic matching is a week out this year, and nothing in the report says so.

Christmas needs the same care next month. It fell on a Tuesday last year and falls on a Wednesday this year, so compare the days around it by their distance from the twenty-fifth.

Monthly totals this year

November last year contained Thanksgiving, Black Friday, the weekend and Cyber Monday. November this year contains Thanksgiving, Black Friday and the Saturday after; the Sunday and Cyber Monday fall in December.

So November may well look weaker than last year and December stronger, without anything having changed in the business.

For the season as a whole, compare the whole season: November and December together, or the weeks counted from Black Friday. Month against month is the wrong unit this year.

A worked example

A running store with an online shop opened its weekly report yesterday morning. Visits and online orders for November 18 to 24 were well below the same dates last year.

The owner's first thought was to bring the Black Friday sale forward and raise the advertising budget.

The same dates last year had contained US Thanksgiving, Black Friday and the Saturday after. This year they were the quiet week before.

Compared with last year's week before Black Friday, November 12 to 18, the figures were close. The difference was the calendar.

The sale stayed on its planned dates. The store now keeps a short table of each year's event dates, and its weekly report compares weeks by their distance from Black Friday.

Write down what else changed

Even with the weeks lined up, the two years differ in ways the report cannot see.

The sale may be deeper or start earlier. The email may go out on a different day, to a larger list. Ads may have started a week sooner, with a different budget.

The site may have changed: a new checkout, a redesign, a different shipping offer.

So may the measuring. Safari now limits how long analytics can recognise a returning visitor, which makes new and returning counts hard to compare with last year.

A filter added or removed in Analytics shifts every figure from the day it changed, and so does tracking that broke during a redesign.

Note each of these in a line, with the date, in Analytics annotations or a plain document. Next year's comparison will need them.

Count orders, not only visits

Visits are the easiest number to compare and the least useful. A weekend with fewer visits and more orders was a good weekend.

Take orders, revenue and average order value from the store system itself, which counts every order whether or not analytics saw the visitor.

For a business that takes enquiries rather than orders, the same applies to enquiries and the work they became.

A table for next year

The lasting fix is a small table kept from year to year: the dates of US Thanksgiving, Black Friday and Cyber Monday, and the day of the week Christmas falls on.

Next year Black Friday moves again, to November 27. With the table, the comparison takes minutes. Without it, someone repeats this week's confusion.

The counter-case

A business untouched by Black Friday, such as a clinic or a trades company, can keep comparing months. A six-day shift does little to its figures.

A very small store has a different problem: at a few orders a day, any single day is mostly noise, and only the whole season says much.

And over a full year the shifts even out. The distortion is concentrated in late November and December.

For any business whose busiest weeks follow the shopping calendar, compare event with event, and week with matching week.

Before Friday

  1. Write down both years' event dates.
  2. Name each week by its distance from Black Friday.
  3. Set comparison dates by hand for the event.
  4. Match the day of week for ordinary weeks.
  5. Compare the whole season, not November alone.
  6. Note what else changed this year.
  7. Take order figures from the store system.

Step three is the one the reports will not do for you. Every default comparison in the tools lines up dates or days of the week, and this year neither lines up Black Friday.

Counting what those orders and enquiries actually became is covered in measuring what happens after the form is sent.

Frequently asked questions

Why do this year's November figures look low?

Because Black Friday moved from November 23 to November 29, and Cyber Monday into December. Same-date comparisons set ordinary days against last year's sale days.

Is matching the day of week enough?

Not for the event. It compares each day with the one fifty-two weeks earlier, but this year Black Friday comes fifty-three weeks after last year's.

How should I compare Black Friday weekends?

Event with event: Black Friday with Black Friday, Cyber Monday with Cyber Monday, and each week by its distance from the event, with comparison dates set by hand.

Should I compare November with November?

Not this year. Cyber Monday falls in December, so compare November and December together, or the weeks counted from Black Friday.

What else should I write down?

The sale's depth and start date, email send days, ad start dates and budgets, changes to the site, and changes to measurement such as Safari's shorter cookies.

When does Black Friday fall next year?

On November 27, 2020. A small table of each year's event dates makes next year's comparison quick instead of confusing.

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.

This year's report showing a collapse?

Send it to us. We will line up the weeks by event and tell you what actually changed.

Start a Conversation