fbpx

TRADITIONAL PRICING        MACHINE-LEARNING  PRICING  

Retail pricing strategies: machine learning vs. traditional methods

Well thought-out pricing in e-commerce is essential in today’s online retail, to maximize profits and operate successfully on the market in the long term. This article explains the advantages and disadvantages of different pricing strategies and how machine learning-based pricing can help.

Table of Contents

What is rule-based pricing?

Rule-based pricing is the most widely used and conventional type of ecommerce pricing today. It uses static pricing rules or “if then” formulas to dynamically adjust prices and respond to changing influencers. Frequently, the company’s own prices are linked to the competitive price: for example, the company’s own price is always 10% lower than the direct competitor’s price. According to this pricing strategy, the majority of prices in online retailing are still established and adjusted. In addition to competition-based pricing, cost-based and value-based pricing are also used.

Popular rule-based pricing strategies

Cost-based pricing

  • Cost-based pricing uses production or material costs as the basis for the pricing strategy.
  • To determine the product price, a certain profit level or minimum margin must be added.
  • Based on the costs, a price floor and a price ceiling are set. 
  • If the price of competitors is below their own price lower limit, retailers try to reduce their costs in order to lower their own lower limit.
Pros Cons
  • Simple calculation

  • Security in the contribution margin via fixed margins

  • The basis of the pricing strategy is exclusively based on costs

  • Takes neither customer behavior nor competitor prices and offerings into account

  • Neglects customers' willingness to pay

  • No maximization of total profits possible

Value-based pricing

  • Valued-basedt pricing is based on the benefit or value proposition that the item offers the end customer.
  • In other words, it is about finding the price that the customer is willing to pay.
  • The value of the product and the customer’s perception of value are key to pricing – not production costs.
  • Prices and relative value to competitors are usually determined by customer surveys. 
  • This method is often used for initial pricing (e.g., determining the recommended retail price).  It  is unsuitable for continuous optimization of discounts. 
  • This form of pricing can be applied to individual customer segments.
Pros Cons
  • Price supports the product image

  • Takes into account the customer's willingness to pay 

  • Very good method for initial product pricing

  • Potentially higher profits possible

  • Requires high manual effort and human analysis

  • Deep understanding of customer segmentation required

  • Often based on surveys, not real transaction data

  • Neglects costs, which can lead to unprofitable prices

Competition-based pricing

  • E-commerce pricing for a product or service which is based on the prices of the competition.
  • Typically, this type of pricing is applied by companies where competitors sell similar or the same products.
  • Prices are set based on pricing rules that establish a fixed margin (e.g. always 5% cheaper) or a specific ranking (e.g. among the top 3 cheapest) relative to the competition.
  • This pricing strategy requires a deep understanding of the relevance of different competitors and their price changes on their own sales.
Pros Cons
  • Existing and emerging competitors always in sight                                          

  • Aggressive growth strategy with a focus on revenue thanks to competitive pricing

  • Easy to implement when competitor prices are available

  • Potentially higher profits possible

  • Neglects the customer’s willingness to pay                                                        

  • Neglects production costs

  • Relies on competition to act intelligently

  • Can have a negative impact on profitability 

  • Can lead to downward price spiral in the marketplace

What are the disadvantages of rules-based pricing?

Omnichannel retailers today operate in a very dynamic market environment. There is a plethora of influencing factors that affect pricing decisions, which are subject to constant change (e.g. competitor prices, own stock range, marketing activities, and also seasonality).

Managing static pricing formulas in such an environment requires a lot of manual effort. Regular checks must be made to ensure that the rules set are delivering the desired results. Retailers have to make pricing decisions for many hundreds or even thousands of items while keeping an eye on competitors and other influencing factors such as inventory levels or marketing activities – a tremendous effort that ties up time and resources.

Rule-based e-commerce pricing is often based on a few parameters. While these are easy to measure, they say nothing about customers’ willingness to pay or their buying behavior. Valuable profit and sales potential is thus wasted. Particularly with a large product range, discounts cannot be sufficiently differentiated with rule-based pricing at the product level. The defined rules then apply across product categories, leaving even more profit potential untapped. Another disadvantage is the orientation to competitors’ prices, which can lead to a downward price spiral in which retailers undercut each other.

Last but not least, conventional pricing systems cannot learn from new data. Since they do not adapt, they react sluggishly and too slowly for today’s market. This is an additional factor that contributes to profit potentials not being fully exploited.

Study

According to a recent global survey of more than 1,700 business leaders by Bain, 85% of B2B management teams believe their pricing decisions need improvement, but only 15% have effective tools and dashboards to set and monitor prices .

What is e-commerce pricing based on machine learning?

Rule-based pricing only considers selected factors of pricing without applying a systematic measurement of price willingness. The more products in the assortment, the more complex the rule system becomes and the higher the effort to maintain it.  

Pricing methods that use machine learning algorithms can take all relevant influencing factors into account. First, price elasticities are measured. Based on this, the effect of price changes on profits and sales is predicted for each product. In the optimization process, the price points that lead to the achievement of the previously set goals are then selected. 

For the user, the pricing process is turned upside down: instead of first defining rules and then performing analyses to achieve KPIs, ML-based pricing optimizes in a targeted manner. The pricing manager specifies the objective and the algorithm issues the appropriate prices. Whether the target is reached is already known before the prices go live.

What are the advantages of machine learning-based pricing?

Leading online retailers are leading the way: they use advanced pricing solutions to set optimal prices for each product. These pricing solutions are based on AI, as described earlier. Machine learning-based algorithms measure customers’ willingness to pay via price elasticity. This information is combined with forecasting algorithms. The software also uses all relevant data – from historical transactions to competitors’ prices and the weather. Depending on predefined KPIs (business objectives), the optimal prices for each product can thus be calculated down to the second.  

Advanced pricing software can also be used to simulate different strategies and pricing scenarios for selected categories or product groups. Revenue, sales and profit results for specific targets can be forecast and ultimately implemented with just a few clicks. The administrative effort is low, leaving more time for strategic issues.

Benefits of all contemporary pricing strategy based on machine learning:

  • Leads to 5-15% higher profitability through a better understanding of willingness to pay
  • Achieves a high level of automation by eliminating the need to manage pricing rules
  • Takes into account all relevant influencing factors
  • Continually learns and improves
  • Understands in advance how customers will react to pricing and price changes, optimally meeting business objectives
  • Additionally provides a forecasting function for sales, revenue and profit
Study

According to McKinsey, AI-based pricing and promotions have the potential to add up to $500 billion in value globally.

Conclusion

Compared to rules-based pricing, machine learning-based dynamic pricing offers key competitive advantages. These include automation of the pricing process taking into account all important influencing factors; simple price control via targets; and a significant increase in sales and profits. In addition, pricing is continuously improved by self-learning algorithms. Prices are no longer determined on the basis of static pricing rules but with the help of precise measurements of price willingness. In this way, the full potential of existing data is exploited.

At 7Learnings, we have more than 10 years of experience with machine learning technology. Leading European online retailers use our technology to optimize their prices. On average, we have been able to increase our clients' profits by more than 10% and prove it in A/B tests. If you want to learn more about the potential of next-generation dynamic pricing, feel free to contact us to arrange a product demo.

Subscribe to our newsletter

Be first to read our blog posts, case studies and knowledge pages