How to Make Search Smarter with Dynamic Boosts

How to Make Search Smarter with Dynamic Boosts

Not too long ago we introduced relevance settings, a no-code feature that enables our customers to adjust search results based on any number of factors, such as making textual or ranking adjustments.

Now we’re bringing those same intuitive settings to dynamic boosts. The power of dynamic boosting is the ability to amplify better types of results to drive higher conversions. No other search platform gives users the ability to configure these kinds of changes so easily. 


Dynamic boosting can push higher performing results to the top.



Dynamic boosting massively increases search performance by automating the ordering of products to maximise business outcomes such as conversions, revenue, or profitability. We’ve seen businesses quickly lift revenue by over 10% using only this feature! 

It works by collecting and processing signals related to search queries (clicks, cart adds, purchases) to build models that understand how to optimize each individual query. 

Dynamic boosting is highly flexible and can be used in many different ways, including personalization. But before we dive deeper into the details, some background on why dynamic boosting is necessary.

Understanding search intent

Customers expect to find relevant products when they search in an online store. During winter they are unlikely to look for a summer jacket. When searching for a “TV”, they are probably not looking for a “TV wall mount”. 

But on many sites with standard keyword search, that’s exactly what you get. Most search engines only look at the index without understanding the context. To compensate for the poor results, hundreds of manually created search rules are created. Manual rules are hard to maintain and quickly start to interfere with each other, making it extremely difficult to optimise your search experience and, as a result, your conversion rate. 

Available now, dynamic boosting reads your customers' minds! Well, technically it analyses their behavior (searches, clicks, cart-adds, purchases…) and builds a model that determines, in addition to the textual and ranking relevance mentioned above, what search results are the most relevant for a given search query. Because this model constantly evolves, it can take into account seasonality, other factors, and trends that may influence your customers’ purchase behavior. 

Easy set-up

Although the dynamic boost system is quite complicated under the hood, we made it as easy as possible to set-up. No need to hire a data scientist or pay for extremely expensive consulting hours. You simply input what is important and Sajari will do the rest. 



Dynamic boosts have two different modes:

  1. Records (each individual result is compared independently)
  2. Category (aggregated associations to a specific field are used, such as a categories, brand, gender)

Record mode

As the name suggests, record mode boosts individual records if they are popular for specific search terms. 

Take for example a fashion store. A search for “dresses” brings up dozens of results. But Meghan Markle was seen wearing a Knot Dress last week and it turned out to become the latest trend. The machine learning in Sajari will recognise these behavioural patterns and boost one or more individual products dynamically. 



Category mode

Category mode does work in a similar way, but instead of boosting the product, a category attribute (brand, material, or actual categories) can be boosted. 

Let’s take the TV example from above. If 90% of purchases made after a customer searched for “TV” were actual TV sets, Sajari’s Dynamic boosting will boost products in the “Television” category for future searches and de-prioritise items that might be in the “Television Accessories” category. 

This is extremely useful for marketplaces or stores that sell a variety of different products. 

Search personalization with dynamic boosting

All the examples above are aimed at optimizing the result order for a specific query input. But it can also be used to improve search personalization where we may want to dynamically boost results to match characteristics of the searchers themselves. 

Take the example where we can geolocate the customers that are searching your site. This geo attribute is added to each query where a dynamic boost can create a relationship between the state or latitude the person is in and the brand of the products they purchase. 

It turns out people in California are more likely to buy the brand “Patagonia” compared to those in Massachusetts who are more likely to buy “North Face”. Dynamic boosting automatically discovers this relationship and dynamically changes the result ranking accordingly. 

Better relevance = better conversion rate

Dynamic boosting learns what is relevant to your customers and intelligently adjusts the ranking of results according to those preferences. Better search leads to an increase in conversion rate for your store. Oh, and your buyers will be happier too!

To see it for yourself, signup for a free 14-day trial or schedule a demo.

Try Sajari completely free for 14-days.
Start today