Ranking Attributes Video Tutorial

 

Ranking cocktails are also known as Ranking attributes.

Video tutorial transcript:

Hello, and welcome to this short video introducing ranking cocktails.

This video demonstration will familiarise you with ranking cocktails and how they are set up in Fredhopper’s Merchandising Studio.

Please be aware that test data is used throughout this demo, and you should refer to your own data sets when applying this functionality.

What are ranking cocktails?

Rankingcocktails allow you to execute simple or more complex ecommerce strategies.

Applying ranking cocktails will enable you to order products using a combination of metrics at the same time, rather than pure sort alone.

These ranking cocktails can then be applied to category and search pages, product recommendations campaigns and result modification groups.

Let’s think about ranking cocktails as the recipe for ranking items in your catalogue. As with any recipe, you need to know the ingredients and the quantities required.

The ingredients are the attributes, and the quantities are the weights.

Once you have defined your ingredients and your quantities, you can create your ranking cocktail.

• Attributes are taken from your existing data model. These can be, for example, match rate, inventory, price, margin, discount percentage, colour, unit sales, etc..
• Weights represent the importance of the ingredients for the recipe, from 0 – 100%

 

Here are 2 ranking cocktail examples:

The first one is a general sale strategy, where we want high inventory products to be sorted first in the list of products. Of those products, we want to sort high price to the top followed by high margin products. These attributes have been given a weight of 70, 20 and 10, respectively, totalling 100.

The second example of a ranking cocktail may be to sort products by newest in stock, so we can use Freshness, then sort those products with the highest product views, followed by products that have sold in the last 7 days, then of those products show the products with the highest inventory. These attributes have been given a weight of 50, 25, 15 and 10 respectively, totalling 100.

The weight of each attribute must be a positive number between 0 and 100. The higher the weight number, the higher the importance it has in the ranking attribute. 

The weight of any single attribute cannot exceed 100%.

The sum of all attributes can exceed 100%, but for simplicity keep it to 100% as a total.

This is a very useful way of sorting products to match your business strategy.

For more information on weights please refer to the ‘Getting Started with ranking cocktails’ article.

How to demo

So how do we set this up in the Merchandising Studio?

Let’s use a different strategy and implement this as a ranking cocktail. So we want to push products with a low discount percentage and high unit sales, and high inventory to the top.

Using your credentials, open your Merchandising Studio. This will open on the campaigns tab by default.

From the navigation bar, click on the Rankings tab, then choose Ranking attributes.

Next, click on New.

Whenever you create a ranking cocktail it is best practice to apply a consistent naming convention, so you can quickly identify the ranking cocktails that you may want to apply to a rule or campaign. We suggest that you use the prefix rc_ followed by the name for example; rc_newin_strategy

Type in a unique name for this Ranking Cocktail.

For this scenario, we will use rc_lowdiscount_highstock.

If you have read and write authorisation to a number of Scopes, you will need to choose which one you want this cocktail to be applied to. If you only have one, then it will automatically be selected by default.

Choose your scope.

Next, start to add the properties to your Ranking Cocktail by adding Attributes:

From the Attributes dropdown, choose the relevant option. Remember, this is a demo environment so your list of attributes may differ, as shown here.

Please note Attributes selected from the dropdown can only be used once in each Ranking Cocktail.

As we want to push the lowest discounted products, I’ll select discount percentage as my first attribute. Followed by units sold last week, so I need to click on Add Attribute and choose units sold last week. And lastly, add the inventory attribute:

Now, let's configure each of the attributes. Firstly we need to select how we want to normalise the values of the attributes.

Normalisation determines how Fredhopper transforms the values of the attributes so that they are represented on a standardised scale. This ensures that each attribute in the recipe is evaluated objectively.

Logarithmic normalisation is typically used where attributes have extreme outliers in their data set. These may include sales, add’s to basket, user reviews, and days since first stocking item.

Linear normalisation is used where attributes have fewer outliers, for example, product ratings, availability or even price when you have a homogenous catalogue.

 For more information on Normalisation, please refer to the ‘Getting Started with ranking cocktails' article.

So, let’s set Discount percentage to linear, units sold to logarithmic and Inventory also to logarithmic.

Note that if you know that your discount percentage ranges between only a few percents say 20– 30% then set this to Linear as there will not be any extreme values to skew the normalisation.

Conversely, if you have a varied discount range anything from 5% - 90% for example and the majority of discounts are between 20 – 50%, but some fall outside this discount range you may want to choose logarithmic. Therefore it is important you understand your products and discount ranges in this example.

Now we need to decide if we want to push high values or low values to the top, and lastly, choose a weight for each attribute.

So let’s start with Discount percentage. We want products with low discount percentages to appear at the top, and we will give this a weight of 70.

Next, unit sales we’ll set to 20.  Lastly, the Inventory attribute. Let’s give this a weight of 10:

Make a final check that the attributes have been configured according to your scenario. You can also check that the summary below meets the scenario or strategy you want to achieve.

Once you are happy with your settings, click Save.

If you are ready to set this Ranking Cocktail live, send the Ranking Cocktail for Publication.

Note; follow your business process workflow to ensure the Ranking Cocktail is published.

Once published, this ranking cocktail can be applied in ranking rules, result modifications and campaigns across your site.

A few things to remember:

• Use the naming convention rc_ prefix followed by the name, for example, rc_sales_seasonal
• Test the behaviour of each attribute using Preview
• Add at least one attribute for which data is always available
• Align the ranking attributes with your ecommerce strategy and your shopper's behaviour
• Choose the best normalisation option based on the vales and the value distribution of the attribute
• Pay attention to the pushing values you have chosen
• Clean up/manage ranking attributes regularly to ensure they are still fit for purpose

Thank you for watching this demo video. We hope you have enjoyed this quick introduction to ranking cocktails.