How Competera Pricing Platform Works?
Artificial intelligence and machine learning changed the retail industry once and for all. As a leading second-generation AI pricing solution, Competera has become one of the technology revolution pioneers. Let’s uncover the unique technology blend under the hood of Competera.
The Evolution of Pricing Software
Just like any tool used today, pricing software has gone through an immense evolution before it got to the point where it is now. Point a cursor on each stage of the scheme below to explore the evolution in detail.
Dependency on calculation input
- Simple rules often triggered by basic market factors' changes
- Regular repricing are possible
- Constant manual checks are required
- Dependence on market data
Rule-Based + consulting:
- Pricing based on more factors, e.g. elasticity calculated manually by consultants
- Human supervision and regular checks and recalculations are required
- Often market-share focused
- Limited flexibility and scalability
Rules + Elasticity Software:
- Typically a spinoff from consultancies
- Elasticity is calculated using a mathematical approach
- Elasticity considered as a constant, i.e. one of the factors for rules
- Limited dynamic capabilities
ML adjusted pricing:
- Next evolutionary stage of the mathematical and static rules
- Elasticity is considered as a static feature for rules
- Elasticity is recalculated each repricing using basic AI, e.g. Bayesian inference and regression modeling
- One of the downsizes is that elasticity calculations must rely on exceptionally accurate data on every granularity level
Deep Learning Pricing:
- Recurrent neural networks self-train on large and diverse datasets
- Huge diversity of information is considered and correlated with each other to reveal relationships and impacts
- Less guidance and supervision is required to create meaningful recommendations
- Decreased vulnerability to sparse data due to multiple factors consideration
- Advanced defensibility capabilities
Why AI pricing solutions of the second-generation are a revolution?
G1 – Elasticity Correlation
How do my sales react to price changes?
G2 – Deep learning
What impacts the purchase decision of my shoppers?
The first generation of algorithmic pricing solutions automated the process of demand elasticity’s calculation. Eventually, the risk of human factor was mitigated, yet it was still about a single dependency calculation, i.e. an individual SKU’s elasticity calculated as constant.
Put it simply, the G1 solutions took over the work previously done by the pricing consultancies, which itself was a huge step ahead. However, it is not only one factor, price, that affects the sales in retail. The human brain is much more complex and while making purchases, customers either consciously or unconsciously are impacted by dozens of factors, not one or two.
When G1 solutions were deployed, they were also significantly restricted by the technical and cloud computing limitations of the software market of that time. As a result, G1 solutions still could not consider billions of interrelations within the portfolio along with the other factors impacting purchasing decisions besides elasticity. Things changed with the introduction of the second-generation AI/ML pricing solution.
Competera sustains optimal prices across portfolio and all sales channels in real-time
Instead of just measuring how a price change impacted sales, Deep Learning unlocked capabilities to investigate each factor that have impacted sales historically and predict how they would affect shoppers decisions in the future. With the large volumes of data available, it became possible to identify these impacts with high accuracy even without having the historical price change data for a particular retail store!
Competera's algorithms (™) are capable of continuously recalculating billions of possible price combinations across all the stores, categories, and sales channels based on 20+ pricing and non-pricing factors.
You can hardly find a real-life analogy among automobiles, but if you imagine one, it is like changing a car which consumes 10 gallons per 100 miles right now to the one consuming 0.1 gallons of petrol every 100 000 miles. Just to give you an idea of how tremendously higher are the quantities, complexity, and accuracy of calculations done by G2 solutions compared to G1 approach.
The most effective pricing engine is used for each product role and strategy
Why is Competera’s technology blend one of a kind?
Context-dependent own price elasticity (e.g. depends of % of price change, season, price positioning to competitors...)
Linking of similar products, substitutes and complements to each other prior to optimization based on SotA ML
One-to-many and many-to-one cross-elasticities taken into account
Model re-trains at each pricing cycle (daily or weekly)
Data lake - all customers contribute to the single data model while each client's data remains unrevealed and protected
Find answers to some of the most common questions people have regarding the use of Competera.
Which factors are considered by Competera?
- Product characteristics (belongs to the category, brand)
- Own regular price elasticity
- Cross-product elasticity
- Own price sales cannibalization
- Other product price sales cannibalization
- Cannibalization through regular pricing
- Cannibalization through promo
- Inventory levels & Sell-out
- Out of stock
- Pricing tiers/store clusters
- Media reach and efficiency
- Competitors’ prices
- Competitors’ stock
- Competitors’ promo
- Competitors’ markdowns activity
- Currency exchange rates
- Life cycle
- Retail internal events
- External events
- Similar products sales
Which size of data set do retailers need to start using Competera?
Ideally, a retailer should have at least two years of historical sales data. However, we can still start with just six months of data and use simpler pricing rules. As soon as the model becomes capable of predicting impact with high accuracy, we can use ML optimization.
How do you ensure the quality of recommendations?
We have an ongoing SLA and monitor over 10 metrics, including model confidence level, elasticity distribution etc. to guarantee the quality of recommendations. All of these metrics are available in the user interface and can be easily accessed by the customer.
Get consultation from Competera’s pricing experts
Let's find the best pricing approach
covering your business needs.
Please fill out the form to discover
main features of Competera
Best Analytics / BI Solutions
E-commerce Germany award
Price Optimization Solutions
Trusted Vendor 2022
Top 3 startups at the AI Summit
London Tech Week
Now Tech: Pricing and Promotion
G2 High Performer 2022