Blogpost Doug Ross
BLOGGPOST
ARTIFICIAL INTELLIGENCE

An Opinionated Guide to Fixing Site Search Using AI and ML

SogetiLabs member Doug Ross blogging about how easily site search can be improved significantly using a combination of Artificial Intelligence (AI), Master Data, and Natural Language Heuristics (NLH).

Is it just me or do most website search functions seem weak?

If my observations are accurate, it’s hard to understand, because site search is such a useful and natural way for consumers to find products or services that might want to buy.

Yet, when I use retail-focused and B2B site search functions, I often find myself getting frustrated. And I don’t think I’m alone.

Just to confirm my suspicions, I visited a few websites of popular brands. The following table depicts some of the problems I noticed.

The ramifications of a poor search experience are manifold:

  • Bounce rates (i.e., shoppers who drop from the site altogether) increase as buyers fail to find useful results
  • Conversion rates decline, directly impacting topline revenue
  • Customer loyalty and retention drop, potentially affecting both the topline and margins

In short, site search has become an important lever for influencing buyer behavior, yet it sometimes appears to be a neglected aspect of e-commerce websites.

What can be done to fix site search?

How can we make a brand’s products and services more accessible to site visitors?

Here at Sogeti — and our parent organization Capgemini — we have discovered that site search can be improved significantly using a combination of Artificial Intelligence (AI), Master Data, and Natural Language Heuristics (NLH).

In short, we have united several technologies to address the fundamental challenge of finding, not just searching.

Here’s a real world example. For one of the world’s largest automotive remarketers, we created a natural language search engine that “understands” the language of vehicles.

Type in 19 ford fusion black 20k AWD and this next-generation search understands the buyer is seeking a 2019 All-Wheel-Drive Ford Fusion with less than 20,000 miles and an exterior color of black.

How easy is that?

Here’s another example:

As humans, we can look at this search phrase and understand that the user is looking for any 2014-2018 Hyundai Elantra with less than 80,000 miles. Our next-gen search offers these (accurate) results:

This site search engine recognizes dozens of attributes such as seller, location, make, model, trim, color, drivetrain, body style, engine, transmission, etc.

So, great, Doug. What are the impacts of this next-gen search?

Well, the business results for this global automotive remarketer are clear:

  • A significant lift in conversions (sales) as users can find more vehicles that match their preferences
  • An increase in customer satisfaction, as users can simply type in their search criteria over dozens of product attributes versus filling out a complex form to filter results
  • A lift in customer retention, as the search function allows buyers to beat the competition for in-demand vehicles
  • An entree into the growing voice search space (e.g., Alexa or Siri)

While some may not yet have realized it, site search is the “gift that keeps on giving”. It has the potential to drive top-line growth, reduce margin pressure, and improve customer experience.

Any firm with an e-commerce presence would do well to consider enhancing their site search to drive real world business results. Feel free to connect with me at any time if you have questions or just want to discuss the content of this blogpost.

Hat tip: Clint Gibler for the article’s headline.

 

todo todo
CONTACT
  • Doug Ross
    Doug Ross
    VP, National Solutions Architect - AI and RPA | USA
    +1 937 291 8100