How to set up Advanced Apparel Recommender?

What is Advanced Apparel Recommender?

With this recommender, user enters basic information (i.e. age, weight, height) and the app estimates the user's body size which is then used to make apparel size recommendation. It is a low lower friction than simple generic table recommender which asks for user's body measurements directly.

How do I create one?

First, on the sizing editor, click on "Size Recommender" tab, then you will see the following block. Click on "Use this" for the"Advaned Apparel Recommender".

Creation flow

1. Select source table

Now the recommender editor will popup. If your size chart only has 1 table, we will automatically select it and you will see a screen like this:

If you have more than 1 size chart, then we will firstshow a table selector screen. Choose the table you want to use for the recommender.

2. Set up product information

Proceed to the second tab to set up product info for this size chart.

Click on "Set up product info" button to select the clothing category most appropriate for this size chart

After saving, the section should look something like this:

Other configuration options will show up as well.

3. Select measurement fields

Now you should see a block right under for "Select fields for the Recommender"

Select the fields you want to use for the recommender. For tops, chest is required and for bottoms, the waist is required. Click on the circular text directly. Green means the measurement is highlighted.

You may notice an error message saying canonical measurement linking is not set up yet. Don't worry, we will set that up in the next section.

4. Field Canonical measurement linking

After selecting a measurement, the Measurement fields settings section will show up. 

First set up the canonical measurement linking. This is important for the recommender to know exactly what the measurement field is. Since the table cell can have any text or language, there was no way for us to know whether a measurement definitely represent chest measurement or not. This is why we need the explicit linking. 

Clicking on the button will show up the selector. 

Be very careful when deciding betweenProduct measurement vsBody measurement. If the measurements you have are measured directly from the clothing, then they are product measurements. When product measurement is selected, the recommender will assume certain level of easing depending on the product category for each measurement field. Otherwise, for body measurement the recommender will not add any easing.

5. Field recommend settings

Once the canonical measurement linking is setup. You will see more options show up like the following.

Depending on the type of the field, there would be different options

Cover types:

"cover" types are like chest, waist, and hip.

• Stretch factor: how stretch is the clothing. If you are unsure, just leave at 10%.
• Easing: This option only shows up for product measurements. This is the preferred gap between the clothing and the body. Depending on the product type, we would have a default easing. This can be customized as well.

Length Types:

"Length" types are like shirt length, sleeve, inseam, etc.

• Ideal length reference point: provide us on a guidance on how long the specific measurement should fall on the body.
• acceptable range from reference point: A range from the given reference point for "best length".

That's it! Now a advanced recommender is set up. There are more settings option in the settings tab for you to further customize things like recommender language, text, flow.

Common FAQ

• I input my own info and the recommender shows no sizes are recommended. What's wrong?
  • A common problem is not setting up the canonical measurements correctly. For example, selecting half product circumference for chest when the numbers in the table are for full chest circumference.
• My recommender result is not good. How can I customize it to fit better for my product?
  • See step 5 in the guide above. You'll want  to customize things like easing. For example, if you want the recommender to recommend bigger sizes, you can increase the easing, and vice versa. 
• My pants/Jean recommendation is off! I normally wear 32 but recommender is suggesting 34.
  • So it is important to input the actual pants waist measurement instead of the label number. The thing is most of the brands have vanity sizing issues where a label 32 jean is actually 34 or even 36 inches in waist.