How dynamic price change algorithms work on Amazon and AliExpress


A familiar schedule?

Maybe you noticed. Prices for some products on online platforms are jumping quite seriously. One week item, from a leather notebook to a wicker basket, can cost $ 5. Next week – already $ 15. And then – again $ 5, or even $ 3. Especially now masks and sanitizers are subject to this law. But prices for many other products with the advent of Covid-19 also began to jump stronger than usual. Even food and expensive gadgets are subject to the general trend.

The virus has made the public more aware of price manipulation for things like toilet paper, cleaning products, vitamins, and various foods. But now this is happening with almost all goods. And the reason is not only in their limited number. Price fluctuations occur on Amazon, and on AliExpress, and Walmart with eBay. The reason is the special algorithms used by sellers to maximize profits. It is almost like a forex or stock market, where everyone, as they can, is trying to predict the best price and outsmart competitors.

How do dynamic prices work (and why are they needed)?

Dynamic (or algorithmic) pricing – allows you to change the cost of goods on the site every hour or even every minute. At the same time, hundreds of positions. Algorithms synthesize a ton of information – about what is happening on the site, on similar sites, for similar products. Even the estimated solvency of the user may be taken into account.

Dynamic prices are no longer a novelty for hotel reservations and airplane tickets. As soon as the system felt the growth in popularity of the room, its prices go up. But for goods, a different, more stable scheme was usually used: the cost of the product from the manufacturer + logistics + seller’s own wrapping. In addition, many buyers have the opinion that “Apple gadgets will always cost that much.”

To prevent sellers from raising prices, Amazon has a system Buy box. Each popular product that you choose on the site is usually held by several sellers. When you click on the “Buy Now” or “Add to Cart” button, the seller with the best price (for a new product) is placed by default. Then you can look at the offers of others – and choose them if, say, you like delivery time more, their goods are in Amazon stock, you like their rating or you want to take a thing that was already in use. But the vast majority of buyers do not reach this step: why, if the first product is the most profitable?

According to Channeladvisor, 82% of buyers do not change the seller they proposed. Therefore, to win the first position in the list is an order of magnitude more important than everything else. This is the difference between a dull existence and a stream of sales.

To determine exactly who “wins” the Buy Box, Amazon uses its own algorithm, similar to the ranking of links in Google. Of course, they never reveal it to anyone. They say it includes not just the price. They look at the seller’s rating, order processing speed, number of reviews, number of returns, previous history. In total, more than a dozen factors are taken into account. But to anyone who was on the site, it is obvious that the main weight still has a price.

Therefore, independent companies like Feedvisor and Sellery are creating pricing algorithms for such sellers. By given one of the studies, the use of such software is especially effective for those who sell goods for $ 250 thousand – $ 2 million per year, and gives an increase of about 15%. This is tens or hundreds of thousands of dollars. Without any improvements on the part of the seller, simply on price manipulations. It is not surprising that companies agree to change the cost of goods dozens, or even hundreds of times a day.


Price chart for PC case (tracking via CamelCamelCamel). The price difference is 2, and sometimes 4 times!

The sites do not prohibit this! They even have own (albeit rather primitive) tools for automatic price control of goods. In Amazon, for example, you can create rules by which the current value of the goods is determined – depending on the price of the Buy Box, date (what if it’s a weekend or a holiday?), And sales.

Moreover: Amazon itself is the best at a dynamic price. Company Algorithms are changing prices millions of times a day, monitoring supply and demand. The cost of goods can change even after the buyer clicked on “Add to Cart” (though usually usually downward). The price is reset even more if the algorithms find some other retailer whose product is cheaper. Counted a lot factors, but in the end, Amazon’s prices of products, according to research, are on average 3% lower than the rest.

Offline, Walmart tried to compete with this, setting a “flat” low price, but for the whole day. Just like they used to in their supermarkets. But after a couple of months the company had to adapt, and now it is also forced to set dynamic prices (so far only online), so that buyers comparing two sites do not leave disappointed every time.

Can price change algorithms harm buyers?

Of course! After all, prices are not necessarily “twisted” into a decrease. If the seller sees that he has no competitors for a certain product, or the algorithm notices that the competitors just sold out their stocks, nothing prevents him from inflating the price by 500% and 2500%. This is what has recently happened with masks and sanitizers. Careless (or desperate) buyers paid $ 250 for goods that cost $ 15 a week earlier.

Experts express concern over dynamic prices also because they can lead to collusion of sellers. Moreover, this can happen even by accident, without such plans on the part of man. One study last year even revealedthat price-setting AIs learn collusion very quickly, and can even do so bypassing their owners. Moreover, it can be difficult to notice.

There is a reverse case. Algorithms can tune in to enmity, and constantly lower prices, interrupting each other. This makes the sale of goods unprofitable. Especially often this happens at such competitive sites as Amazon and Ali. Actually, that’s why Amazon does not prohibit the use of dynamic prices. It may seem to the seller that at the lowest cost he will squeeze out competitors (and this happens regularly), but then the next seller appears with a similar strategy, which will squeeze him in the same way. It turns out to be a classic prisoner dilemma. Researchers say this also dangerous: as a result, the necessary goods will either not be on sale, or a monopoly will be formed on them. To achieve the desired diversity with such methods is unlikely to succeed.

It turns out even funnier when sellers, suspecting that with the lowest prices they won’t get a big profit, adjust the algorithms to a different strategy: setting prices is always slightly higher than their competitors. Like, maybe the buyer will think that he has better quality, or like the rating, and the margin will be higher! In one study, two such algorithms, fighting each other, brought Amazon book value up to $ 24 million


The saga of one book in four parts. From $ 3.25 to $ 23,698,655.93

Of course, this is an extreme, and it can be easily avoided by setting the maximum allowable prices. But even with this limitation, algorithms can drive the price up. Professor Chris Wilson, in his study of dynamic prices on large Internet sites, says that this is Pandora’s box, and the logic of developed AI systems can no longer be understood. “Price management can take place in ways that are not obvious to a person.”

Can algorithms be beneficial to buyers?

Yes of course! Firstly, this happens directly. Still, more often algorithms try to play at the lowest price, and if you don’t fall for anomalies, you can save a lot. Secondly, that area is still not well understood, and, possibly, dynamic prices, on the contrary, contribute to the development of competition, helping to reflect the current situation in the market.


Informed.co machine learning platform is one of those that offers to “conquer” a Buy Box, as well as increase sales on eBay and Walmart

In one research In 2016, it turned out that dynamic price change algorithms can be indispensable for non-profit organizations, such as symphony orchestras, allowing you to increase revenues and ensure that the halls are always full. Prices may decrease depending on the place, program, day of the week, time, indicators of similar concerts earlier and so on.

The same scientists then discovered the same unused potential in sporting events. If the team does not show itself to the full or plays against a not very popular opponent, dynamic prices will help to attract enough fans, while not losing too much profit. For sports fans, this is also beneficial: tickets will cost so much cheaper.

Peter Fader, a professor at the University of Pennsylvania who was the research director, explains:

Dynamic pricing, not just price inflation, is a natural way in all industries. This should be the rule rather than the exception.

American airlines started to use dynamic prices in the 1980s to compete with low-cost airlines. John Morgan, a market pricing professor at UC Berkeley, said this had a huge impact on the entire airline industry:

There is no doubt that consumers have benefited from this. It began with this: when the system sees that the plane will not be full, prices drop sharply. These new, lower prices have opened the possibility of air travel for those who previously could not afford them.

Is it possible to make sure that you are paying the normal price?

There are systems for regular automatic change in the value of goods for Walmart, and for eBay, and for AliExpress. If you suddenly become interested in something, it is likely that demand arose not only from you. And the cost of goods in this regard can dramatically overestimate. So now almost all sellers are addicted.

Those who buy regularly should definitely use browser extensions. Without this, nowhere is it possible, otherwise you can get into tens, or even hundreds of dollars, especially if the product is unfamiliar to you. Can recommend Camelcamelcamel and Keepa for amazon or AliTools for Aliexpress. Tracking prices, checking to see if you accidentally hit the peak. Set limits that will notify you if the price drops to an acceptable value for you. And do not buy if you see that the price has only recently risen sharply. It’s easier to wait a few days, or choose something else, similar. Fortunately, there is enough choice on online marketplaces.

Similar Posts

Leave a Reply