Contents
Article by Gabriel Tassé, published on gabrieltasse.com · https://gabrieltasse.com/en/resources/economic-crises-amazon-discounts
Tariffs, war in Ukraine, conflict in the Middle East, inflation: for six years, every economic shock has come with the same prediction. Prices will climb, and discounts will disappear.
In our Black Friday study, we looked at prices once a year. Here, I change the scale: the same database, read month by month, from January 2020 to October 2026.
For each product and each day, we compare the price with the product’s usual price, meaning its median price over the previous twelve months. Two measures come out of it:
- On discount: the product sells at least 10% below its usual price.
- Pricier than usual: it sells at least 5% above.
Here is what those two curves look like, with the six events people mention to me most often.
- 1Pandemic March 2020
- 2War in Ukraine and first rate hike February and March 2022
- 3Canada Post strike, then GST holiday November and December 2024
- 4U.S. tariffs and Canadian counter-tariffs March 2025
- 5Israel–Iran conflict June 2025
- 6Most counter-tariffs withdrawn September 2025
Share of selling days, all products combined. A product’s usual price is its median price over the previous twelve months. The November peaks are Black Friday.
Source: Keepa price history, Amazon.ca, January 2020 to October 2026.
The pandemic: the one shock you cannot miss
At the start of 2020, about 19% of selling days were on discount. In the spring, that share fell below 14%. At the same time, the share of products sold for more than usual jumped from 25% to 37%.
It is the largest gap of the whole period, for both measures, and it holds when we change the way we calculate: depending on the variant, the share of products pricier than usual rises by 11 to 13 points, and the share on discount falls by 5 to 6.5 points. Discounts did not return to their earlier level until the end of 2020.
No surprise there: factories at a standstill, empty shelves and everyone buying online at once. When a product sells itself, nobody needs to mark it down.
2022: a trace, but not the one we expected
The war in Ukraine began on February 24, 2022. Six days later, the Bank of Canada raised its policy rate for the first time since 2018. The two events fall in practically the same month: our data cannot tell them apart.
What we see is inflation. Throughout 2022, about 30% of selling days were at a higher price than usual, against 25% in 2021. On Amazon’s own prices, without third-party sellers, that share reached 40% in the summer.
Discounts, for their part, did not disappear. Their frequency even rose in the spring of 2022, from 19% to 23% of days. That rebound is fragile, though: it sits at the edge of ordinary variation, its size ranges from 3 to 6 points depending on the calculation, and it does not show up on Amazon’s own prices. I note it without drawing a conclusion.
Canada Post, tariffs, Iran: almost nothing
For the four most recent events, I compared the three months that follow with the three months that precede, once the seasonal effect is removed. Then I repeated the same calculation for every other month of the period, to learn what an ordinary variation looks like.
1Pandemic
−6.5 pts 19% → 14%
2War in Ukraine and first rate hike
+4.9 pts 19% → 23%
3Canada Post strike, then GST holiday
−1.1 pts 21% → 20%
4U.S. tariffs and Canadian counter-tariffs
+1.5 pts 20% → 21%
5Israel–Iran conflict
−0.6 pts 21% → 20%
6Most counter-tariffs withdrawn
+0.7 pts 20% → 21%
Change in the share of days on discount: the three months after the event compared with the three months before, once the seasonal effect is removed. The two percentages are the levels observed before and after.
Source: Keepa price history, Amazon.ca, January 2020 to October 2026.
The Canada Post strike and the GST holiday, at the end of 2024? A gap of about one point. The Canadian counter-tariffs of March 2025? A point and a half, in the direction of slightly more discounts. The conflict between Israel and Iran, in June 2025? Less than one point. The withdrawal of most counter-tariffs, in September 2025? Less than one point as well.
In normal times, the frequency of discounts already moves by a point and a half from one quarter to the next. None of these four events stands out from that background noise.
One nuance on tariffs: discounts stayed just as frequent, but they became slightly shallower, by one to two points depending on the calculation. The signal is thin, and when you examine this many comparisons at once, you should expect one or two to appear by chance alone.
Out of six events, only one clearly changed how often products are discounted on Amazon.ca: the pandemic.
More frequent, not deeper
Our first article showed that good Black Friday prices are becoming more common. The depth of discounts does not follow: when a product is on discount, the average reduction was about 25% in 2020 and 22% in 2025.
In other words, crises do not make discounts disappear, but time does not make them more generous either.
And 2026? Fewer discounts, mostly among third-party sellers
The current year stands apart from the previous ones. From June to October 2026, about 15% of selling days were on discount, against 19% or 20% in the three previous years. The drop set in during the spring and has not reversed: since May, every month has sat about four to five points below its usual average. At the same time, the share of products sold for more than usual went from 22.5% to 26%.
I first suspected our sample: products tracked for a long time end up ageing, and a model at the end of its life is rarely marked down. So we added about 650 more recent products, then compared products of the same age. The drop remains. Among products tracked for less than two years, 27% of days were on discount in 2025, and 19% in 2026.
Products Amazon sells itself (1,659 products)
Products offered only by third-party sellers (4,260 products)
Share of selling days on discount (at least 10% below the usual price), June to October. All products combined: 20% in 2025, 15% in 2026.
Sample completed in November 2026 with more recent products. Source: Keepa price history, Amazon.ca, January 2020 to October 2026.
The drop does not affect everyone in the same way. For products Amazon sells itself, the share of days on discount goes from 16.7% to 15.8%. For products offered only by third-party sellers, it goes from 20.5% to 14.1%. And if we look only at Amazon’s prices, with no third-party offer at all, the difference is minimal: 17% in both years.
Caution is in order. Measured like the six previous events, over three months before and three months after, the change of spring 2026 stays within ordinary variation. What sets it apart is how long it has lasted: six months in a row below normal. I cannot attribute it to a specific cause either, and it says nothing yet about Black Friday week.
One reassuring sign, in fact: in the days leading up to that week, 20% of the best-selling products were on discount this year, against 20% to 22% every year since 2020. For the most popular products, then, nothing points to a less generous Black Friday. The verdict will come with the week itself, and we will update this article.
What I take away
- Only one event left a clear trace on discounts: the pandemic. Fewer discounts, more products sold above their usual price, for most of 2020.
- The inflation of 2022 shows up in everyday prices, not in discounts disappearing.
- The Canada Post strike, the 2025 tariffs and the Israel–Iran conflict blend into ordinary month-to-month variation.
- In 2026, discounts have been less frequent since the spring, mostly among third-party sellers. The drop is not explained by the ageing of the products we track, but we cannot attribute it to a specific event.
A coincidence in time is not proof. We observe what happens before and after each event; we cannot claim that one caused the other. But the absence of a trace is information in itself: headlines move faster than prices.
About the analysis. 6,851 products sold on Amazon.ca, in 19 categories, from January 2020 to October 2026 (Keepa price history). This is every product tracked for our Black Friday study, not only the 100 best sellers of each year, hence a higher number. The price kept for each day is the best offer for a new product, from Amazon or a third-party seller. A product’s usual price is its median price over the 365 days before each month. The seasonal effect is removed by comparing each month with the same month in other years. For each event, we compare the three months after with the three months before (the month of the event counts among the months after if it happens in the first days of the month, and among the months before otherwise), then we place that gap among those of every other month of the period.
Limits. The 2025 events are close together: their three-month windows partly overlap. A second national strike at Canada Post, from September 25 to October 10, 2025, followed by rotating strikes, falls in the months after the counter-tariffs were withdrawn: we cannot separate the two. We reran the calculations with a usual price set over 6 months and over 24 months, with windows of 2 and 6 months, and on Amazon’s prices only: the four recent events stay within ordinary variation, and the pandemic remains the clearest shock (it cannot be checked over 24 months, for lack of history, and it is less marked on Amazon’s own prices); the findings on 2022 depend more on the calculation choices. For 2026, the sample was completed in November with more recent products; a product enters the calculation only once we have about a year of prices, so 2026 launches are not included yet. The same-age comparison groups products by the number of years since Keepa began tracking them. The results cover Amazon.ca only, and a coincidence in time does not establish cause and effect.
Gabriel Tassé is Vice-President, Strategy, Innovation and Thought Leadership at Tink and teaches at HEC Montréal.
Sources
- Price history: Keepa (opens in a new tab), Amazon.ca, January 2020 to November 2026.
- Event dates: pandemic declared by the World Health Organization (opens in a new tab) on March 11, 2020; Russia’s invasion of Ukraine on February 24, 2022; first increase in the Bank of Canada (opens in a new tab) policy rate on March 2, 2022; Canada Post strike from November 15 to December 17, 2024, then from September 25 to October 10, 2025; GST holiday (opens in a new tab) from December 14, 2024 to February 15, 2025; U.S. tariffs and Canadian counter-tariffs (opens in a new tab) on March 4, 2025, withdrawn on September 1, 2025 (opens in a new tab) except for steel, aluminum and automobiles; conflict between Israel and Iran from June 13 to 24, 2025.