Overview of On-Site Search Technologies from 1990’s to Now

outdated technology.jpg

When it comes to e-commerce websites, almost nothing is more important for IT, sales and marketing professionals than an understanding of how buyers want to search for and find products.

A review of how on-site search began, how the technologies evolved, and where they are today can help e-commerce companies make the best decisions regarding their platforms, in-house programming options and vendor solutions.

Major site search technologies and companies

There are many companies dealing with search technologies, some doing core technologies, and some building services on top of such core technologies. To keep this article shorter, only a few systems will be covered here:

  • Apache Lucene
  • Apache Solr
  • Elasticsearch
  • Swiftype
  • Algolia
  • AWS CloudSearch
  • SLI Systems
  • Google Site Search

Released in 1999, this advanced search engine “building block” provides analysis, tokenization (breaking up text into sentences and words), indexing, spellchecking, highlighting, etc.

Lucerne, without any middle-ware, is very difficult to use for site search. It is mostly unreasonable for an e-commerce company to try to do so.

2. Apache Solr search technology

Five years after it debuted Lucene, Apache released Solr. Based on Lucene, this server application provides an API that makes Solr much easier to integrate with all types of applications, including site search.

Many developers believe that Solr is a nightmare to learn and use. Although Solr is somewhat DevOps-friendly (server admin), there’s a steep learning curve, and even a strong team of developers will need months, if not a year, to build a good internal site search using this software.

3. Elasticsearch

In 2010, software developer Shay Banon created Elasticsearch, a server application based on Lucene, as an open source program. Elasticsearch provides an API that is easier to use than Solr.

While the application is difficult to learn, most developers love it once they master it. I’m one of them. It takes just as long to build a real e-commerce search with Elasticsearch as it does with Solr, but it is more DevOps friendly, and one can start playing with it almost immediately.

Solr and Elasticsearch have become the standards in the site search space, with a number of search-as-a-service providers building their applications on top of these systems.

4. Swiftype on-site search service

Swiftype was created in 2012 as a “search as a service” application based on Elasticsearch. It took two more years of development before it was ready for widespread release and use.

Swiftype works as a plug-in for popular e-commerce systems and can be integrated into any site with multiple pages.

It’s difficult to adjust how Swiftype ingests data, making it difficult (but not impossible) to tune it for B2B e-commerce sites. Once a developer learns the application, it’s easy to make it look and feel like the pages are a part of your website.

5. Algolia on-site search service

This application has a similar history to Swiftype. The company was founded in 2012 and its “search as a service” application was released in 2014.

Algolia is like Swiftype in almost every way except that it’s most likely not based on Elasticsearch, has a nicer API for pushing data into the system, but it’s harder to make the integration seamless.

6. AWS CloudSearch on-site search service

Based on Solr, CloudSearch is another “search as a service” application. Originally based on A9, Amazon’s proprietary technology, this application wasn’t popular with developers, but has since morphed into something very similar to Algolia.

AWS CloudSearch is harder to set up and work with as documentation/help is lacking in completeness.

7. SLI Systems on-site search service

Founded 2001, this application was made to integrate easily with almost any e-commerce website. I think it has somewhat poor front-end customization options (because it integrates with anything), making it impossible to create a seamless customer experience.

It’s great for marketing teams because it provides many ways to customize results (e.g., boost products, synonyms, etc.) and, after initial setup, it doesn’t require additional work from website developers. SLI provides plenty of support and handholding during setup, but feels and works like an outdated system.

GSS is easy to quickly integrate into a website, but has limited customization options. This makes it difficult to generate pages that look like a part of your website.

9. Other site search services

There are multiple other “search as a service” providers, including companies like SearchSpring, BloomReach, and others that can also be worth considering for your e-commerce site search.

Site search as a service providers normally charge $500-$5,000 per month to use their technology, and offer support for your team in integrating their search services into your infrastructure.

Other information to consider when upgrading e-commerce systems

Steersman can help you get an all-in-one business automation system with a built-in advanced e-commerce front end to run your operations and production, report and analyze financials, manage all product information, and serve your catalog to clients, allowing them to purchase online.

Steersman Search comes as a component of Steersman Business Suite, an all-in-one business automation system

Work with Steersman directly or through your web agency to setup Steersman Business Suite and website SaaS. Front end design, layouts, and coding can be done by Steersman or capable design and implementation partner firms.

~ by Art Shipulin, CTO at Steersman