they say you should ask customers what to build. they didn't say how. so we asked our support desk.
evidently, people want to know which Fomo settings work best for their website.
that's why this month, we're proud to release fomo insights.
powered by machine learning, fomo insights is designed to increase conversions by suggesting optimal settings for your website, in real time.
how insights works
we're due for an in-depth engineering post here, but in summary:
- impressions and clicks on Fomo client websites are collected via POST requests to a secure endpoint
- Redis aggregates cohorts of event data, then batch inserts to a database we host using Google BigQuery
- BigML, our backend provider of machine learning algorithms, helps process the data
- A proprietary, standalone Fomo Stats service interprets the data and forms raw metrics into human-readable suggestions
- Each of our apps, from the Shopify plugin to our core product, retrieves Insights via private GET endpoints on the Stats Server
- Insights are polled every several minutes, as well as whenever a Fomo user visits their analytics dashboard
putting it all together
at the time of this writing, Fomo serves around 500,000,000 notifications per month.
each notification includes a browser fingerprint, along with some other parameters about the Fomo client's website, totaling to ~20 columns of data per notification.
Google BigQuery and BigML work together with our Stats server to analyze 10,000,000,000 data points per month, and thus recommend the highest converting settings to our users, in real-time.
to learn more about Fomo Insights, or to contribute to its development, shoot us an email.