The future of search in eCommerce

The future of discovery in online shopping is continuously evolving. More and more data is collected from the users, but search for many is stuck on keyword search. With the availability of sophisticated search technologies through Google and other search engines, the keyword based search technologies are becoming less relevant as the behaviour of consumers change.

With the new technologies that enable site search to do natural language processing to answer queries that are no longer keyword based. This means that people will be increasingly relying on searching for products on eCommerce sites as they do on search engines. Instead of doing keyword search, people are searching with queries they might ask from a sales person in a physical store.

One of the clearest trends in search is that the length of search queries is growing. Instead of doing a search for "red cotton pants", many will soon be doing more abstract queries like "pants that matches the color of my red shirt and are good for summer". If you search is powered by indexing searches like Lucene based Solr or Elastic this kind of queries will likely not result in any results.

This type of searches require sophisticated language processing technologies and means that many shops will likely need to get access to these from a dedicated search vendor or alternatively tap to the Google Knowledge Graph (the technology that powers Google's semantic search) to provide this type of capabilities. So far there are some efforts in the Open Source community, but these technologies will require quite a bit of work to get started with compared to a Semantic search SaaS. Search is one of the key features when moving into a serverless architecture for eCommerce and Content Management.

In addition the search engine will require a sophisticated content storage which rich workflow features, which in turn will lead to a fully fledged content management framework instead of a simple shopping catalog. This is one space where advanced content management systems with support for chunky content and a rich content model with natural relations between content objects.