When a shopper opens Fit Quiz on your store, they are asked a few quick, easy-to-answer questions about their size preferences, body type and fit expectations. Within seconds, they receive an accurate size recommendation.
But what actually happens behind the scenes?
Unlike traditional sizing tools that rely on measurement charts, Fit Quiz uses a dynamic, data-driven recommendation engine. Here is a breakdown of how our algorithm calculates the perfect fit for your customers.
Why we don't use size charts or garment measurements
For many e-commerce brands, creating and maintaining size charts and garment measurements are a major challenge.
They are often inaccurate: products rarely match initial manufacturing specs. It means, that two identical products in the same size might have different measurements.
They are too generic: brands often only offer one standard size chart for an entire category or collection. This doesn't account for differences between fits, fabric stretch, drape, or how an item is supposed to look on a real body.
They are a maintenance nightmare: updating spreadsheets for every new collection, style, and category takes hours of manual work.
Fit Quiz completely eliminates the need for size charts. Instead, our algorithm bypasses static spreadsheets entirely by looking at three dynamic data layers to predict the perfect fit.
The three pillars of the Fit Quiz algorithm
Our algorithm cross-references your products against millions of data points across three core pillars. It automatically determines which one of these should be used, based on the quality and quantity of available data. In means, that in some cases all three are applied, while in others a combination of these is leveraged with one or the other having a bigger weight in the model.
1. Global shopper network data
Fit Quiz utilizes an extensive, anonymous network consisting of the fashion, sizing, and fit preferences of more than 18 million shoppers globally.
When your customer answers the quiz, the algorithm instantly analyzes this database to find:
Fashion doubles: it maps the user against other shoppers in our global network who have similar body profiles, fit expectations, and product preferences.
Product characteristics: it scans the item's category, type, and sizing range to match it with similar products in our database.
Contextual nuances: it factors in geographic variations (as sizing standards shift by region) and the overall fashion styles.
2. Your store’s order and return data
The algorithm learns specifically from your business by analyzing your historical and ongoing sales and returns. It dives deep into which products are returned, how often it happens and what reasons shoppers leave to better understand how your unique items are fitting real shoppers.
This is possible, when your return management app is connected to Fit Quiz. You can choose from one of the 15 return solutions that are already integrated with Fit Quiz or request a custom integration.
⚠️ A note on data reliability (statistical significance)
We take the accuracy of your recommendations very seriously. The algorithm only incorporates return data when it hits a threshold of "statistical significance", meaning there is enough data to prove a true pattern. If your store, category or product don't have a lot of orders or returns yet, the algorithm will ignore this layer entirely to prevent a few random returns from skewing your store's overall sizing results.
3. Standard or custom Size Models
Our algorithm also incorporates your products unique design. Size Models are your way of directly telling the algorithm exactly how your clothes are meant to fit.
Size Models can be viewed, edited or created directly in the Fit Quiz app within your Shopify admin. You can choose a standard Size Model, like true to size or half size down, or set up a custom model.
These models are made to reflect your specific design in the final recommendation. Whether a jacket is meant to be oversized, or a pair of jeans has an ultra-skinny fit, the algorithm calculates the size using your input in Size Models.