Does dynamic price optimization work for my e-commerce business's pricing? (5 typical use cases)

Does dynamic price optimization work for my e-commerce business’s pricing? (5 typical use cases)

“We don’t have enough historical transaction data.”

“We don’t have big enough sales volumes.”

“We don’t want to be the cheapest.”

“We can’t change prices often enough.”

Many e-commerce and retail companies (falsely) assume that dynamic price optimization is not relevant for them – and these are just some of the common misconceptions that we keep hearing.

The truth is, AI-powered dynamic pricing it’s not just for extremely time-sensitive businesses like hotel bookings and flight tickets.

Dynamic price optimization works for many types of businesses and products. 

That said, if you’re not optimizing your prices proactively yet, you’re most likely leaving easy money on the table. (It’s not just a myth that a 1% change in price can translate into a 10-20% increase in profits.)

A quick disclaimer before jumping into the topic: sometimes you hear people referring to rule-based pricing as “dynamic pricing,” but in this article, we talk about automatic price optimization based on machine learning – and more specifically, reinforcement learning.

5 types of products that almost all retailers carry with a perfect use case for dynamic price optimization

So, how do you know if price and profit optimization works for your business? 

For the sake of simplicity, let’s take a hardware store as an example, and see what are the 5 different types of products they could use dynamic price optimization‘ for:

  1. Premium-priced products with added value, e.g., Weber barbeque grills
  2. Products with no direct competition, e.g., Torx screw drive vs. slot head
  3. Products with high volumes and low prices, e.g., screws, wall plugs, and such 
  4. Category B products, e.g., two-by-fours (category A) vs. nails (category B)
  5. “Forgotten products,” e.g., single screwdrivers for professionals vs. screwdriver sets for laypeople


1. Premium-priced products with added value

Suppose you’re selling products that carry a tangible (or sometimes imaginary) surplus value (anything from a sustainable and transparent production chain to higher-quality materials or a brand that reflects a particular lifestyle). 

The odds are that you’re using price signalling and positioning yourself in the market with a premium pricing strategy.

However, the truth is that most premium pricing is not done as systematically as it should: most brands simply take costs, add a needed margin and some extra on top of it, align the price point with their direct competitors, and then simply stick with that price.

This is where dynamic price optimization comes in. 

With dynamic price optimization, you can subtly test your price point in the market without damaging your brand image (or alternatively in some cases, use data about the relation of price and demand to inform your pricing strategy at a higher level).

What often happens with premium-priced products is that you find out that you could increase your prices quite a bit before it affects your sales volumes at all.

Hardware store equivalent: High-end barbeque grills like Weber. Some people are willing to pay extra just to get a Weber in their backyard. They won’t even consider a cheaper option because they’re so convinced of Weber’s betterness – regardless of if it’s true and if a cheaper option can get the same job done.


2. Products with no direct competition

Some products have no direct competition in the market, which means the single thing that moves the needle is the price you sell that specific product at.

In a very competitive market (read: if all your competitors sold exactly the same product), you wouldn’t necessarily have much room for even the smallest price changes. 

In a situation where there is no direct competition, people are not going to go elsewhere – simply because they can’t. This is why a controlled increase in product price can make all the difference in the world in your profit margins. 

Hardware store equivalent: Torx screw drive vs. slot head. If you need to use durable Torx head screws that resist cam-out better than slot heads, your only option is to buy a Torx head screw drive.


3. Products with high volumes and low prices

Low prices, low margins, right? Not necessarily, if you can outsource your pricing to artificial intelligence.

If you’re selling a lot of products for a low price, it’s tempting to think it doesn’t pay off to test several price points. 

No one could afford to do that manually, but with a dynamic pricing tool, you can automate this process.

By doing so, you can get better profit margins from low-priced products with high sales volumes by adjusting the price based on the real willingness to pay and in real-time. And with high sales volumes, that 20-cent difference in price can make a huge difference.

Hardware store equivalent: Screws, wall plugs, and such. 


4. Category B products

You’re surely pricing your loss leader products and bestsellers actively but might not be optimizing prices for other products that bring in the majority of profits.

Category B products might not bring you as much money as your precious As. With proper automated optimization, they could help you increase your overall profit margins by heaps.

If you need a few two-by-fours (category A), you’re going to need some nails (category B), too. But you’re not going to go that extra mile and visit another hardware store for cheaper nails. In fact, you’re most likely no to even know which store sells them cheaper. 

The same goes for more expensive safety shoes and cheaper protective gloves, or paint and mineral turpentine.

Hardware store equivalent: Two-by-fours (category A) vs. nails (category B)

5. “Forgotten products”

Some products are selling steadily year over year, no matter what. 

Take classic lampshades, for instance. These are not necessarily the cash cows of your business but they’re still doing relatively well, which is why many retailers end up not prioritizing finding the highest possible price point that maximizes their profits.

A very similar situation occurs when a new mobile phone model is released to the market. 

To avoid having to put old models on sale (which typically means losing money), you might want to consider lowering the price step-by-step and getting an estimate of when that product will run out of stock at a particular price point. 

Getting back to our hardware store example: if you’re a construction professional, you’re more likely to buy a single screwdriver for specific use (the pricing of which retailers often tend to ignore), while the rest of us just want to get the cheapest screwdriver set (the pricing of which is retailers definitely don’t ignore).

Hardware store equivalent: Single screwdrivers for professionals vs. screwdriver sets for laypeople

Read our e-book "AI in pricing" to learn more about dynamic price optimization

When dynamic price optimization is not the best choice?

All that being said, we want to be completely honest: there are a few scenarios when dynamic price optimization with reinforcement learning is not the optimal choice.


Scenario 1: You can’t change your prices at all

You don’t need to change your prices every day or even every week to benefit from dynamic price optimization.

But if you can’t change your prices at all for a reason or another, you won’t get any data on how price changes affect your conversions and demand. In that case, our algorithm won’t be able to give you any pricing recommendations.

This can be a problem: 

  • in heavily regulated industries,
  • for retailers with fixed resale prices, and
  • for some omnichannel retailers with no electronic shelf labels and a pricing strategy that prevents from having different prices online and offline.

However, in omnichannel, you can still benefit from price recommendations. Instead of automatic price optimization, you can use the algorithm to inform your overall pricing strategy at a high level and to help you understand market changes and how your pricing affects the level of demand.


Scenario 2: Low sales volumes and short sales lifecycle

Slow sales cycles or small sales volumes alone are not problems for our algorithm: it will give you just as reliable recommendations; it’s just learning a bit slower than it otherwise would, but eventually, you will get the same benefits.

But if your sales volumes are low AND your sales lifecycle short (e.g., just 30 days in total), our algorithm won’t have enough opportunities to learn from your transaction data. 

In some cases, it’s possible to get around this by optimizing a group of similar products instead of single SKUs (e.g. if you’re selling the same print in 50 different colors).


Scenario 3: Unique items that are sold just once and come with a high price tag

Let’s say you want to sell your house. In this case, there’s just one house for sale and you sell it just once, so the reinforcement learning algorithm has no transaction data to learn from.

In this case, it makes more sense to use a neural network that’s provided with the right data set in advance (which is obviously more expensive, but the high price of the house makes up for it). The same goes for any unique, premium-priced items like antique jewelry or expensive art.


Scenario 4: You’ve already found your optimal price point (for now)

Disclaimer: This is very unlikely to happen. This is simply because most businesses carry so many SKUs in their inventory that it’s simply impossible to find out the optimal price point for even your key items — let alone for all of them.

Just for the sake of it, let’s imagine you have already found your optimal price point for your bestselling product.

In this case, our algorithm would make small price changes – just to find out after a while that you’ve already found the point where you can reach your maximum profit margins.

What’s good to keep in mind, though, is that your optimal price point also changes over time.

So even if you knew your optimal price point right now, you’d still need to continue testing how it changes over time to avoid missing any opportunities.

Watch a webinar replay: "Dynamic Pricing —Truth behind the buzzword'


As we saw above, the beauty of a machine learning based price optimization tool is that it’s able to increase profit margins equally well for high-end e-commerce brands, cut-price retailers with high sales volumes, and anything and everything in between. You also don’t need a large product portfolio or massive amounts of historical data to benefit from dynamic price optimization. 

But the best part? 

Our dynamic price optimization tool can produce 10% better profit margins than traditional methods of calculating price elasticity (and a LOT more compared to not optimizing your prices at all). 

Actually, we’re so confident it works for just about everyone that we even promise a 5% increase in profit margins during the first 90 days of use only (T&Cs apply).

Still not sure if dynamic price optimization could work for your business?