Every company tracks—or should be tracking—metrics relating to finances (e.g., revenue), customer (e.g. retention), operations (e.g. inventory), employees (e.g. engagement), and a variety of industry-specific (e.g., cost per patient, MRR, etc.).
Staying on top of your metrics isn’t just nice to have—it’s mission-critical.
But all too often, people’s experiences tracking metrics looks like this:
- “A decrease in user activity! Time to spend the next three hours digging through spreadsheets to figure out what’s actually going on.”
- “Something’s up with our Q3 revenue, but I can’t put my finger on it. Looks like another late night crunching numbers in Excel.”
What if you could set up proactive alerts that not only notify you of changes but pinpoint exactly what changed? For example:
- Your Android users in the 18-24 demographic dropped off by 15% in the last 24 hours.
- Your eco-friendly product line is outperforming projections by 22% in the Midwest.
Our newly improved Tellius Feed metrics tracker unlocks significant time-savings and transforms companies from relying on reactive analytics to leveraging proactive analytics. Let’s look at a couple scenarios where the improved Tellius Feed could make a difference.
Written by: Ajay Khanna,
Founder and CEO at Tellius
In This Post
Scenario 1: Multidimensional dive
Anne, our seasoned data analyst, is trying to figure out why her company’s app engagement suddenly decreased. She needs to pinpoint exactly where and why users are dropping off before the C-suite starts asking questions.
Anne needs a tool that can:
- Monitor user engagement across multiple dimensions simultaneously
- Alert her to subtle shifts in user behavior before they become trends
- Provide granular insights without requiring her to set up dozens of separate tracking metrics
How Feed delivers for Anne
Anne can now set up a multi-layered monitoring system where she drills down by device type, app version, and age group, setting thresholds for unexpected drops. When an anomaly is detected, she gets an instant alert, allowing her to investigate and address issues before they impact user experience.
Setting up Feed
- Primary metric: Daily Active Users (DAU)
- Drill-down dimensions: User Age Group, Device Type, App Version
- Resolution: Daily
- Anomaly detection method: Threshold-based trigger (Any segment showing <20% deviation)
How Feed works
- Feed isn’t just tracking one big DAU number. It’s crunching data for every possible combination of age group, device type, and app version.
- Feed doesn’t just say “DAU down.” It says “DAU down 15% for iPhone users in the 18-24 age group using Android v2.1.”
- One Feed setup does the work of dozens of individual metric trackers, saving Anne hours of setup and monitoring time.
Scenario 2: Identifying patterns that matter
Bart, our business strategist, is struggling to identify genuine anomalies in regional profit amidst seasonal fluctuations, leading to wasted time investigating normal patterns and underprepared executive meetings.
Bart needs a solution that can:
- Break down revenue across multiple product lines and regions
- Identify underperforming segments even when overall numbers look good
- Provide actionable insights without requiring a data science degree
How Feed delivers for Bart
Bart can create a Feed that monitors revenue across various product categories, states, and customer segments. With custom thresholds set, he receives timely alerts about potential issues (e.g., declining performance in specific states). Now, instead of waiting for quarterly reports, he gets real-time insights that let him pivot strategies on the fly.
Setting up Feed
- Primary metric: Revenue
- Drill-down dimensions: Product Category, States, Customer Segment
- Resolution: Monthly
- Anomaly detection method: ARIMA, which considers historical data without a set threshold
How Feed works
- If the 20% drop in eco-friendly products in the Midwest is actually normal for that time of year (maybe due to seasonal buying patterns), ARIMA won’t raise an unnecessary alert.
- This saves Bart from chasing every fluctuation, focusing his attention on truly anomalous trends.
- Instead of a data dump, Bart gets a concise, insight-rich report that lets him walk into Monday meetings armed with strategic talking points.
The Feed difference
Our revamped Feed is not just about getting alerts—it’s about getting the right alerts, at the right time, with the right level of detail. And that’s exactly what it delivers in both scenarios.
By combining granular metric selection, smart anomaly detection, and customizable alerts, it empowers you to stay ahead of trends and make data-driven decisions. Whether you’re debugging a complex system, optimizing a sales strategy, or identifying emerging market trends, Feed offers the granularity, flexibility, and immediacy you need to stay ahead for both technical and business users.
What's new with the Feed in 5.1?
With version 5.1, we’ve brought in the following changes to the Feed feature that go beyond the simplistic “if this, then do this” kind of alerts:
- Don’t just track metrics—dissect them. Break down your measures across multiple dimensions to uncover the hidden drivers of change.
- Pick your preferred method of anomaly detection from ARIMA and Threshold-based.
- ARIMA is perfect for users who want the system to automatically analyze historical data, including seasonal patterns, to identify if the latest value is abnormal, without setting exact thresholds.
- Threshold-based methods are best for users who have specific, predefined values they want to monitor, regardless of historical trends or seasonality.
- Choosing between ARIMA and threshold-based methods depends on whether you need adaptive, pattern-based anomaly detection (ARIMA) or strict, user-defined limit monitoring (threshold-based).
- Why wait for dataset refreshes? Set your own schedule for anomaly checks.
Check out other improvements we’ve made in Tellius 5.1 on our Release Hub page or release notes. Or sign up for our webinar to watch the new features live in action!
