Contextual Search

"Contextual search is a form of optimizing web-based search results based on context provided by the user and the device being used to enter the query" - this is the definition provided by Wikipedia. But what does that mean? What determines the context for each individual user? Context can be defined by different factors and the challenge is in how to obtain them.

1. Historical factors:
   These factors give information of what the user has done in the past: what products did he/she order, what products did he/she put on his favourite lists or in an abandoned basket?
   
   
2. Demographical factors:
   Things like age, gender and occupation can provide clues to possible interests.
   
   
3. Behavioural factors
   In which topics did the user express explicit and implicit interest? Which products did he/she view or what articles did he/she read? Did the user provide some interest preferences on his/her account? And what to think about the state of mind? Any expression of mood whether positive or negative reflected in a status update or post could impact the (order of) content/product presented.
   And then you can also have a look at the collaborative aspects, what did other comparable users do or buy?
   
   
4. Geographical factors:
   Particularly with the increase of mobile usage the geographical location of where the user is, can be used to offer results that really help the user. The country, region, or town the user is in and the proximity to points of interest including shops can be included to increase the relevance of the search result.
   And more online related: Where did they come from (which referral site)? And what type of persons visit these sites? From what article/topic were they coming to your shop? All context that can influence what you will present to that single user and in what order making it a personal experience.
   
   
5. Environmental factors:
   When talking about the environmental factors you should think about the surroundings that influence the context. Think about what device the user is using. And in case of mobile phones what apps are installed, and which contacts does he have on that device? And what Day of the week is it, what time of year, and are major events such as Christmas, Easter or Summer Holidays close by? Also, the time of day can have an impact on what is more or less relevant, like coffee in the morning and beer in the late afternoon or evening. Similarly, the local whether can also impact searches with ice cream or BBQ related products becoming more prominent on warm sunny days.

All in all, quite some factors that can define a personal context which, If you can manage to obtain that data, can be incorporated into the query logic.

As mentioned, the legacy systems are not (yet) fit enough, so what technologies would be required to implement contextual search?

• The Semantics
  Putting the keywords entered by the user in a context alone is not enough. You will need to understand the customer intent with the query he is asking. The semantic technology will do just that. This technique aims to determine the intent and contextual meaning instead of simply matching keywords to pages/products thus enabling you to deliver a personalized and relevant search experience to your user.

• Natural Language Processing (NLP)
  Natural language processing or NLP is the technology that focusses on the design and analysis of algorithms for processing the human language. In other words, it helps the computer to understand our language therefore being able to deliver better and more relevant search results.

• Machine Learning (ML)
  The array of techniques that Machine Learning is providing allows you to monitor the search behaviour of each individual user and detect patterns, which in some cases would have been overlooked or which would only come to light after a longer period of time. These patterns are then used to draw accurate conclusions on what products certain users are looking for and as such which products to offer next or which products to highlight in a search result by changing their relevancy (collaborative filtering).

In the era of digitalization organizations are slowly sitting on a data gold mine, which only will deliver value if you can unleash its potential. Contextual search amongst others will help organizations to find the most relevant information from unstructured data and gain value from it with higher conversions and larger average order value per order (a Salesforce study from several years ago showed improvements of AOV: +10.3%, Add2Cart: +24%, Conversion: +4.6 times higher).

In return the end user will be receiving personalized and relevant results with significantly less effort. It can even lead to providing the right products before the user was starting to look for it (predictive search).