Learn about Search

Introduction to Search

The search function on an e-commerce website allows users to search for specific products or items they are interested in purchasing. It typically involves a search bar where users can enter keywords, phrases, or product names, and the website returns relevant and in Constructor’s case attractive results based on their search query.

The search function can be a vital component of an e-commerce website, as it allows users to quickly and easily find the products they are looking for without having to browse through multiple pages or categories. It can also provide a personalized shopping experience, as it may suggest related or complementary products based on the user's search history or preferences.

From a business perspective, the search function can help e-commerce websites increase sales and customer satisfaction by providing a more efficient and streamlined shopping experience. It can also provide valuable data and insights into customer search behavior, which can be used to optimize the website's product offerings and marketing strategies.

The undersurface of the Constructor’s search function includes several key components, including:

Natural Language Processing (NLP)

NLP involves analyzing and understanding natural language text, including queries entered by users in the search bar. Constructor uses a variety of NLP techniques to improve search accuracy. Additionally, our query expansion includes techniques like spell corrections, auto-generated synonyms, acronyms, and more to construct a firm understanding of what are buyers’ intentions so that the most attractive results are returned

Query retrieval

Query retrieval involves searching the product catalog or database to retrieve all products that match a user's search query. Constructor’s primary matching mechanism is clickstream. We utilize both semantic analysis and clickstream-based AI to identify candidate results that could be attractive for a given query. These are then ranked using algorithms within the same families, focusing on learning from user behavior to determine which results will be most appealing for specific queries and users. As a backup, when sufficient clickstream data is unavailable, the system can default to vector and keyword matching approaches, ensuring performance that is at least comparable to legacy systems.

Indexing

Indexing involves creating and maintaining an index of all products available on the e-commerce website, organized in a way that enables efficient search and retrieval.

Ranking

Ranking involves sorting and prioritizing search results based on the user's search query. One of Constructor’s ranking advantages is that we factor attractiveness into this prioritization, not only relevance. This gives us more opportunities to increase click-through rates and ultimately boost sales. Constructor uses a two-stage ranking system that provides significant advantages in terms of efficiency, relevance, personalization, user experience, and conversion rates. We also give opportunities for merchandisers to tune the results based on their goals using various rules and options in the dashboard.

One of the advantages of our ranking algorithm is that we give you the opportunity to pick the metric which matters for your business and set up the algorithm to optimize for it.

Real-time personalization is also factored into ranking, based on the user's search & browsing history, purchase history and preferences.

Filters

Filters enable users to refine their search results based on various criteria such as price range, brand, category, size, color, etc. Constructor supports dynamic ordering of filters so users can always experience the filters that drive better results first.

Autocomplete

Autocomplete (sometimes referred to as "autosuggest") suggests search terms to users as they type in the search bar, based on the most popular or relevant search queries which leads to a better conversion rate. To minimize effort, Constructor anticipates the user's intention to return attractive results.


What’s Next