1 of 14

ECOMBEE

Recommendations Beyond Expectations

2 of 14

Our approach

Recommender Engine

Analytics on the top of Recommender data

Global Recommendation Service

3 of 14

Recommender Engine

Software as a Service

Simple and intuitive API

Rich query language

Special functions (filtering, boosting)

Real time machine learning

Excellent scalability

Sophisticated algorithms

Artificial Intelligence at your service

4 of 14

Our E-commerce recommendations

5 of 14

Analytics on the top of recommender data

6 of 14

Data products based on our Recommender

7 of 14

Our motivation

  • Vast majority of E-commerce data lay idle
    • important behavioral data
    • only large companies can afford data science team

​

  • The cost of inaction?
    • 10-20% of current revenues are lost,
    • losing market to big players in longer term

​

  • Recombee
    • we have a solution easy to deploy and customize
    • affordable to small/medium sized companies
    • advanced AI automation

8 of 14

Our competitive advantages I

Scalability:

Complex monitoring

Distributed

Big Data solution

9 of 14

Our competitive advantages II

Sophisticated recommendation algorithms:

Nearest neighbors

Association Rules

Matrix factorization

...

  • Starting with simple cold-start models (Reminder, Bestseller) up to complex ensembles (e.g. ensemble KNN + token attribute similarity model)
  • Model parametrization and regularization
  • Wrappers such as periodic or diversifying model
  • Online retraining in real-time
  • Precomputed models for instant response

Very accurate algorithms - increase CR up to 20%

Fast algorithms - great cost/performance ratio

10 of 14

Our competitive advantages III

Artificial intelligence:

AI powered multivariate testing

​

11 of 14

Our competitive advantages IV

  • Universality, transfer learning – we generalize solutions over multiple domains. Solution in one domain improves recommendations in others (e.g. song periodicity -> retail product periodicity)
  • We are data mining / AI experts – we go beyond recommendation, improve it with other approaches
  • Scalable R&D team – close collaboration with universities, research lab, talented students
  • Vision – we design our products for the present and future. We often build market for our innovative solutions.

12 of 14

integration

  • Enables easy integration – application transfers data to our cloud and produces recommendations
  • Multiple applications – you can evaluate potential of recommendation, generate personalized campaigns, etc.
  • Insights – upon request, we can deliver reports on customers and product insights
  • Learn more – https://git.recombee.net/keboola/recombee-app-description

13 of 14

References

14 of 14

Contact us

Ing. Pavel Kordík, Ph.D.

pavel.kordik@recombee.com

+420 604 499 078

​

​

​

Recombee, s.r.o.

Rybná 716/24, Staré Město, 110 00 Praha 1

www.recombee.com

​

Ing. Tomáš Řehořek

tomas.rehorek@recombee.com

+420 728 085 130

​

​