Ranking Strategies

You may already be familiar with boosting, our way of letting you determine which pages rank higher than others in your content search results. With the Site Search 360 ecommerce search, you also have access to our powerful ranking strategy tool to determine how your products should be ordered in your search results.

This is available in the Site Search 360 plans with the ecommerce feature only

You can set up ranking strategies using any data from your product feed, such as click data, sales data, margins, preferred brands, sale items, new items, and much more.

In addition, Site Search 360 uses AI-based machine-learning algorithms that can be set up to automatically optimize your ranking strategy based on your Site Search 360 analytics (i.e., clicks, products added to cart, and products sold).

How do I make a ranking strategy?

Creating a ranking strategy is easy. Simply navigate to Result Manager > Ranking Strategy and select + New Ranking Strategy.

Give your ranking strategy a name and description. Then click + Add Attribute to start defining which attribute(s) should be used to rank products. The percentages you set for each attribute determine how important that attribute is.

In my example, I made a ranking strategy to prioritize a specific brand as well as products that have the "value" Fair Trade. Since Brand is set to 25% and Values to 75%, products from this brand that are Fair Trade will rank higher than Fair Trade products from other brands. Products from this brand that are not Fair Trade, however, will show up after all Fair Trade products. With this ranking strategy, the result set for "dress" would contain all the dresses the shop sells, ranked in the following order:

Poplinen and Fair Trade
Other brands and Fair Trade
Poplinen, not Fair Trade
Other brands, not Fair Trade

What if I do not have any useful data for a ranking strategy?

If you do not have your own data for a ranking strategy, Site Search 360 has your back. As mentioned earlier, we use our powerful search analytics to provide valuable ranking insights to Site Search 360 ecommerce customers.

When you create your ranking strategy, you will find 3 machine-learning attributes: sxAddedToCart, sxLearnedImportance, and sxSold.

Here are these scores explained:

sxAddedtoCart: how many times a product was added to cart (all time)
sxSold: how many times a product was sold (all time)
sxLearnedImportance: a score based on a product's click-through rate in the last 100 days (the older the click, the lower the value) as well as where the product was clicked (a product clicked on position 9 is more significant than a product clicked on position 1, so the product clicked on position 9 gets a boost in sxLearnedImportance).

You can use one or all three attributes in your ranking strategy. These values will continuously change based on your visitors' search behavior. Thus, your ranking strategy will continuously update and improve too!

What happens if I do not have a ranking strategy?

If you do not have your own custom ranking strategy active, our default ranking strategy is applied. The default ranking strategy uses information about the detected category (if any category is detected using our ontology) and the similarity between query and product name. Category matches are more important than name matches and will be ranked higher.

For example, if the query is "smartphone", products matching the category, such as the iPhone 13 and the Samsung Galaxy S23 will be ranked higher than products with names such as "smartphone cover" or "smartphone charging cable". This is because our semantic search understands that a customer who types in "smartphone" is more likely to be looking for actual phones and not just products with the keyword "smartphone".

Can I have multiple ranking strategies?

Yes, you can create multiple ranking strategies. You have the option to toggle them on and off, so you can make ranking strategies for special promotions, times of year, etc.

You can even use multiple ranking strategies at once if you select "chessboard" under Ranking Pattern.

With this chessboard pattern, the top results from each ranking strategy will be shown alternatingly.

Can I use ranking strategies for a specific query?

Yes, if you want to fine tune how products are ranked for a specific query, you can use the Result Manager to apply a special ranking strategy just to those result sets. For example, you might create a ranking strategy called DVDs with ranking criteria specific to DVDs (e.g., Rotten Tomatoes score, release date). Then you can use the Result Manager to apply this ranking strategy to queries like "movie", "DVD", etc.