Recommendation engine

With prevailing use of e-commerce sites any buyers would come across a set of recommendations for them and behind that is good recommendation engine which is understanding, predicting the customer behavior and needs with certain precision and errors.

I bought shoes from amazon the other day because it was shown to me when there was a marathon event in my city. After developing a recommendation engine for a product that generates a multitude of offers for international travelers the key lies in

Identifying the persona

Learn from experiences to establish contexts

Build offers using the context

Explain to the customer why was the offer made

Persona defines people on the basis of different factors like spending powers and what they looking for as per their age group or what they expect as part of their lifestyle and we look for similar behavior and motivations. All recommendation engines establish contexts by learning from past and present experiences and good ones predict the future. It consists of

Long-term experiences- For e.g. people buying during black friday sale or millennials buying a new car every 4th year.

Short-term experiences - For e.g. number of people traveling to Japan for leisure or increase in the number of cars sold by Audi.

Future predictions - It is done by drawing inspirations from short and long term experiences. For e.g. who will travel in the next 30 days, where will they travel and enrich their predictions with predicted preferences.

Current contexts - It monitors current moments in real time to tweak the offers it prepared.

Building the offer on the basis of the above mentioned recommendation and then matching it with current persona and needs of the customer is then presented to the customer. Now explaining the offer to the customer is the hard part and the correlations that machines will make out of it can build good accuracy.