How Klevu AI works

Klevu AI encompasses linguistics and continuous machine learning to improve online shopping experiences. Klevu shines in a crowded product discovery market because of the comprehensiveness of the AI offering, and how it impacts user experience.

How Klevu AI worksHow Klevu AI works

The product feed is imported then enriched and indexed in real-time, the shopper queries are processed and understood, the right products are displayed in the right order on the ecommerce store, our ML algorithms are continuously updated using data from shopper searches, clicks, purchases, and product reviews.

What makes Klevu AI special?

Data Enrichment

Data Enrichment

Klevu enriches and normalizes product catalog data automatically and saves the enriched data when products are imported, rather than relying on customer experience failures and manual fixing.

NLP: Semantic Query Processing

NLP: Semantic Query Processing

Klevu processes and understands what the shopper means when they search, considering the context of the store, to improve the quality of the results.

Read how Somm Cellars supercharged search conversion rate by 460% using NLP >

Price Parsing

Price Parsing

Klevu comprehends price relativity from a search query. For example, when a user enters ‘gifts under $400’, the price parsing module understands that the user is looking for gifts between $0 and $400.

Decompounding

Decompounding

Klevu can break complex words into their simplest forms to better understand shoppers’ queries. For example, a board that’s black can be expressed as ‘blackboard’.

Emerging Tech
Emerging Tech Brew

Anatomy of an online shopping
AI system

Klevu helps over 3,000 brands use data and machine learning to improve product search.

Dynamic search suggestions

Dynamic search suggestions

Klevu helps shoppers think and refine their search by enabling contextual expansion of queries.

CMS and Product Search

CMS and Product Search

With Klevu, shoppers can use search to be able to find answers to common questions, not only product retrieval, but content and pages.

Dimensional Transformation and Enrichment

Dimensional Transformation and Enrichment

Klevu automatically normalizes different weights and measures and enriches the product catalog, making it easy for your shoppers to make very specific searches.

Deep Neural Networks for Semantic Search

Deep Neural Networks for Semantic Search

Klevu AI uses deep learning techniques to train the AI, which helps the AI learn as a human would, instead of being trained by a human. Deep learning is proven to be faster and more accurate than other ML models.

Learn more about deep learning >

Collaborative Filtering and Personalized Shopping Experiences

Collaborative Filtering and Personalized Shopping Experiences

Klevu AI uses a similar model to Spotify in order to offer personalized product recommendations using a click profile, rather than demographic profiles. This method allows for relevant personalization to even new or anonymous visitors, whilst maintaining privacy.

Continuous Machine Learning

Continuous Machine Learning

The Klevu AI can learn from your customers’ purchases, clicks, product ratings and continuously improves and reacts to trends in shopper behavior.

From the Klevu AI Lab

The Klevu AI Lab is an experimental, multidisciplinary team of scientists and engineers that test, learn and iterate on artificial intelligence in ecommerce, until a new Klevu AI breakthrough is customer-ready. Listed below is a selection of current experiments. Schedule a demo to see these in action.

Image Labeling

Image Labeling

For a product image, Klevu AI generates labels and captions that explains object classification, shape and color. You can use this information for metadata generation and content moderation for your product catalog.

Visually Similar Images

Visually Similar Images

More often than not, shoppers are looking for products that are visually similar to something in the real world. Klevu AI analyzes key properties, such as shape and color, from the uploaded photo, and displays products that look similar to that shape and color.

Intent Parsing

Intent Parsing

Imagine you are a shopper on an ecommerce site and you enter a question or a query relating to products or your order. Klevu AI intent parsing service deconstructs the input and will show the intent behind the question/query along with the level of confidence.

Did you know

Klevu was the only ecommerce search technology included in the Cambridge University Press journal round up in Jan 2022 for making significant use of natural language processing.

Read more >

Gartner

All Gartner Digital Commerce Search ‘basics’ coveredSearch: Catalogue enrichmentSemantics: Query NLPSemantics: Natural Language Entity Extraction and NLU
Algolia    
Klevu    
Attraqt    
Bloomreach    
Fact-Finder    
Kibo    

Klevu AI uniquely offers advanced linguistics in the NLP Enrichment Pipeline, as well as continuous Machine Learning and user behavior analytics, meaning it is constantly learning from every purchase, click and product rating. Retailers that have switched to Klevu AI from simpler indexing engines have seen a major uplift with Klevu:

41%
Increase in revenue from search
31%
increase in ecommerce conversion rate from search
26%
Increase in per session value

If you’re curious about how other product discovery technologies stack up against Klevu AI, download the

Product Discovery Buyer’s Guide >

Want to know more?

Klevu’s Discovery Suite includes On-Site Search, Smart Recommendations, Smart Category Merchandising, and a 121 Personalization Engine – built for ecommerce and powered by Klevu AI.

Schedule a demo