Searchandizing

Introduction to Searchandizing

Constructor Searchandizing provides a powerful toolset to help you augment Constructor's machine learning algorithms to affect the results shown. Searchandizing is commonly used for promotions, sales, and other merchandizing activities.

Searchandizing can be performed on both search and autosuggest queries (a "Query Rule") and browse requests a ("Category Rule").

Types of Searchandizing Rules

There are four different types of searchandizing rules you can apply to your results:

Boost

This applies a weak, moderate, or strong boost to criteria you specify, making the affected items appear in higher in the results. A "boost" is meant to influence the ranking generated by Constructor's algorithms, not to override it. If you'd like results to appear at particular positions, please use the Slotting feature.

Bury

This applies a weak, moderate, or strong bury to criteria you specify, making the affected items appear in lower in the results. As with "boosts", a "bury" is meant to influence product rankings. If you'd like results to be removed altogether, please use the Blacklist feature.

Slotting

Slotting lets you specify precisely where an item should appear in your search, browse, or autosuggest results. In the slotting interface, you can drag and drop items directly where you'd like them to appear in your results.

Blocklist

With the Blocklist feature, you can specify items to not appear in the results at all. Blocklisting items can be done by search term or browse page via the Customer Dashboard, but to globally blocklist an item it must be done through the API.

Note: In our customer dashboard, the blacklist is referenced as a blocklist. In the future, we will update the API to reflect this change (but don't worry, we will preserve backward compatibility when we do so).

Manual vs. Automatic Rules

In addition to manually creating searchandizing rules, Constructor's machine learning algorithms work to optimize your results by creating automatic searchandizing rules based on user behavior. For instance, our algorithms may notice that at a grocery store, a search for "pepper" usually results in the purchase of the pepper spice, while a search for "peppers" usually results in the purchase of bell peppers. In that case, our algorithms will add a "boost" for the spice category in a search for "pepper" and a boost for the vegetable category in a search for "peppers".

Automatic rules are shown next to manual rules on searchandizing pages. They can be manually deleted if necessary.