FAQ
Have a question? We probably have you covered.
I want a recipe that does [X]. Can Miso help?
Most likely. We want to hear your ideas. Contact us at any time at support@askmiso.com.
Can Miso recommend UX patterns for implementing a recipe?
Yes. We have much experience in the UX patterns that are most effective for personalized search and recommendations. Ask your Miso solutions engineer about Bento, our coming-soon UI Kit of best practices and style guides.
Do I send interaction data in bulk or in real time?
You can bulk upload historical interactions. However, you must stream live interactions in real time at low latency. Miso then personalizes the experience for each user immediately. It reacts as they click, scroll, and search. This matters most for a new user, because Miso tailors the recommendations at cold start. See Integrating Your Data for sample code.
What do Miso's results look like for first-time users?
Miso provides results from all the activity on your site. Miso then learns immediately from the user's first few interactions. For this reason, stream interactions to Miso in real time. You can also use the user_cohort parameter. This parameter tells Miso to personalize the cold start behavior as if the user belonged to a certain cohort on your site. See the API documentation for more details.
Can Miso incorporate negative signals?
Yes. Negative signals include removing an item from the cart, requesting a refund, removing an item from a collection, or pressing a 'dislike' button. A product detail view with a very short duration can also indicate that the item was not relevant. For more information on the different types of interactions that Miso supports, please see Importing Interaction Data.
Can Miso use social engagement (likes, comments) in its model?
Yes. Miso can use social feedback such as liking, commenting, sharing, and rating. These are strong signals of user preference. Send them to Miso if you have them. For more information on the different types of interactions that Miso supports, please see Importing Interaction Data.
Can Miso's engines be trained using limited structured data?
Provide as many details as you can about the products or content in your catalog. More details give more meaningful search and recommendation results. These details include attributes such as the title, description, brand, and any domain-specific tags. For more information on our product schema, see Importing Product Catalog Data. When necessary, Miso can also use unstructured data, such as images and videos, to strengthen the personalization models.
Can your APIs support asynchronous writes?
Yes. To use asynchronous mode, append ?async=1 to the Product or User Upload APIs.
Do I need to send customers' purchase history to Miso?
During the onboarding phase, we typically ask for at least 3 months of clickstream data. This data covers product detail page views and add to cart events. The purchase history of customers is not required. If it is available, Miso can incorporate it into the initial training. After you configure the real-time clickstream data feed, Miso automatically tracks purchase history. Miso then retrains the models on a given cadence.
What is your strategy to minimize latency for customers globally?
Miso runs satellite servers in US East, Asia, and Europe. They serve latency-sensitive requests, such as Search and Recommendation API calls. To keep latency low, some customers send those requests from their front-end JavaScript. Their users' browsers then talk to the Miso API directly. Contact your Miso solutions engineer for more details about this option.
Can Miso be used headless?
Yes. You use Miso mostly through REST APIs in your front-end and back-end clients. You can also use our Node.js and Python SDK, primarily for data upload (content and interactions). You manage the initial engine training and certain merchandising tools, such as synonyms, through Dojo. Dojo is our no-code UI for your Miso instance.
For Answer deployments, we also recommend our front-end JS SDK. It provides ready-to-use UI components, built-in analytics, and customizable HTML templates. These save significant development time.
Can you decouple the Search and Recommendation engines?
Yes. In fact, the Search and Recommendation engines are entirely separate and can be fine-tuned individually.
Can I manually swap out a recommendation or search result?
Yes. You can boost, bury, or exclude products, brands, or categories from your recommendations or search results. These actions use pre-defined conditions. You set the conditions programmatically or in Dojo, our no-code UI. For more information, see our Boosting and Pinning recipe.
