Complex technologies for complex pricing issues

To help retailers, Competera is using an integrated approach that combines proven pricing theories and best econometric practices. We then automate all of it with the help of bespoke deep learning algorithms capable of making accurate pricing decisions without any additional help.

Despite the fact that Competera allows users to set the right price for each product across each category, we don’t consider these products in isolation. We focus on the performance of the entire portfolio, taking into account such factors as product cross-dependencies, global targets and metrics that require protection.

What’s under our hood

Competera uses an efficient price optimization algorithm based on machine learning and cloud computations

Our system can be depicted as a neural network that learns how to predict sales based on historical price changes, creates a map of cross-product sales interactions and brings non-pricing factors into the context.

Competera can also be seen as a solver that generates a myriad of pricing configurations, sees how they perform and presents the best one to the user's dashboard.

analytics

To meet the optimization target, our users set two metrics – a global and a protective metric (the one that increases revenue while protecting gross profit margin, for example). Next, they can add some business constraints like price thresholds, MAP, etc.

Having the specified parameters, the solver examines thousands of price configurations in a minute and, based on accurate mathematics, makes the best decision, unlike the usual approach where a person can analyze only a dozen of configurations putting a lot of effort on it.

At the same time, with each new pricing set, an algorithm trains itself using the same technique that is used to train Deep Learning models.

analytics

To address specific industry challenges, we use a variety of approaches. Discover them below:

Case 1

First price recommendation

The ability to predict the attractiveness of a future price is the cornerstone of this case. To do this, we use an AI algorithm capable of processing tabular, textual and visual information about products in order to allocate them to a group of similar products that were sold in the past. Then, basing the decision on the data, an algorithm offers its recommendation for the optimal first price.

First price recommendation

Data input

  • 2 years historical sales
  • Historical price changes
  • Product cost
  • Product descriptions, features

Approach

  • Similar products classification
  • Processing of factors affecting price
  • Search for the best selling price
Case 2

Market-driven pricing

To complete this case, we combine machine learning with rule-based pricing. An auxiliary ML algorithm using Graph Theory and causal reasoning, identifies retailer’s competitors and potential KVIs. All the repricing processes are done with a rule-based engine that can apply rules of any complexity, even customizable decision-making trees.

Market-driven pricing

Data input

  • Historical sales data (min 2 years)
  • Historical competitive data
  • Any third-party data

Approach

  • True competitors and KVIs identification for optimal price positioning
  • Pricing logic built on any business constraints and variables
  • Tailor-crafted pricing decision trees
Case 3

Demand-based pricing

Competera's demand-driven pricing is powered by Neural networks measuring product elasticity and its cross-elasticity to ensure that goals on both the category and portfolio level are achieved. This engine can be run on independent assortment groups, which allows parallelization and scalability. The accuracy of every recommendation is gained through an analysis of context-dependent product environments.

Demand-based pricing

Data input

  • Historical sales (min 2 years)
  • Historical promo (min 2 years)
  • Historical price changes
  • Product stock availability
  • Product description

Approach

  • Measuring the response of demand to price changes
  • Processing historical data to evaluate the impact of pricing and non-pricing factors
  • Selecting optimal prices for the whole shelf or product category
  • Aligning goals on category and portfolio levels

Trust & Security

Competera uses a single-tenant approach. We create separate, fully isolated cloud projects (using leading cloud providers like AWS or GCP) to perform all computing, data warehousing, and processing tasks.

We ensure that customer’s data never leaves the secure perimeter of their own cloud project as well as never intersects with data warehouses of other clients. On top of that, only authorized Competera engineers working on your project have access to the data warehouse.

analytics

Feel empowered with next-generation technologies, enhance your pricing using advanced software provided by Competera

Best Analytics / BI Solutions

E-commerce Germany award

Price Optimization Solutions

Constellation

#1 eCommerce Solutions Software 2019

Crozdesk

Top 3 startups at the AI Summit

London Tech Week

Great User Experience

Finance Online

High Performer 2020

G2 Crowd