For many recommendations, you may want to set fallbacks, so that if there’s not enough relevant data, your recommendations won’t be blank or filled with completely arbitrary items or content. Doing so in Personyze is a relatively simple step at the end of the algorithm portion of any recommendation wizard.

Setting Fallback Recommendations

At the end of any given recommendation’s algorithm page, you’ll see a little plus symbol next to the Fallbacks area, next to the filters.

When you click that, you’ll be presented with an additional and identical recommendation algorithm menu, which will allow you to choose any algorithm to fall back on. I would suggest something like “most popular from category” or maybe even “Any products” with some filters.

Now, your recommendation will have a fallback if no statistical significance is available for the data for your main algorithm, so that you never end up with completely arbitrary or empty recommendations.