Axomaly learns the patterns in your SaaS product and business metrics, then flags meaningful changes without brittle static thresholds. Send values through a simple REST API and get explainable results.
curl -X POST https://api.axomaly.com/metrics/ad5a7883-100d-4726-9999-7e42102a9898/detect \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "timestamp": "2026-08-25T05:44:05.719Z", "value": 1842, "metadata": { "metric": "daily_active_users", "plan": "pro" } }'SaaS metrics change with growth, releases, campaigns, seasonality, and customer behaviour. Fixed limits require constant maintenance and often create noise instead of useful insight.
Axomaly learns the usual patterns of each metric and identifies meaningful deviations from them.
Detection follows evolving metric behaviour instead of relying on permanently fixed limits.
Each result includes context to help you understand why the behaviour was flagged as unusual.
Monitor changing SaaS metrics without constantly retuning static rules.
You should not need to choose anomaly-detection algorithms or tune statistical parameters manually. Describe what you are monitoring and Axomaly prepares a detection configuration for your review.
Smart Configuration
Describe what matters
Start with the metric in your own words.
Payment success rate for our checkout flow, measured every five minutes.
Try an example
Configuration prepared
Ready for review
Review and confirm before creating the metric.
Send timestamped values and optional context through the REST API. Axomaly builds the behavioural history needed for detection.
Simple REST API
Receive an anomaly decision with context showing what made the latest behaviour unusual.
Explainable Output
Adaptive detection for the business, operational, and application metrics behind your SaaS.
Customer & Revenue
See when conversion, retention, and payment metrics move outside their usual behaviour.
Example signal
Payment success rate
Usual range
97.2% to 98.7%
Latest
93.8%
Product Operations
Catch unusual changes in the jobs, queues, webhooks, and integrations that keep your product running.
Typical
Unusual increase
Typical
Product Experience
Detect meaningful shifts in the latency, errors, and throughput signals that shape your customer experience.
Example readout
P99 API latency
842ms
Last 5 minutes
Compared with the 220 to 310 ms baseline
API errors
0.8%
Throughput
12.4k/min
Other Time-Series Metrics
Monitoring something else? Send timestamped values and optional context through the same REST API.
{
"timestamp": "2026-08-13T09:45:23Z",
"value": 93.8,
"metadata": {
"plan": "pro"
}
}
See what changed, how confident the result is, and which signals made the behaviour unusual.
Incoming observations
Every 5 minutes
Business metric
metricId: ad5a7883-100d-4726-9999-7e42102a9898
Detection response
Illustrative example
93.8%
Latest observation
Confidence
Direction
Below expected
Expected range
97.2% to 98.7%
Evaluated
3 detectors
Detection explanation
Payment success dropped to 93.8%, below its learned range of 97.2% to 98.7%. The size and speed of the decrease were both statistically unusual.
{
"id": "019c6a8e-3ef9-7000-a3b2-8f429a2074d1",
"isAnomaly": true,
"direction": "BelowExpected",
"score": 0.86,
"confidence": 0.93,
"hints": {
"Detection": [
"Value 93.80 is 3.21σ below mean 98.01",
"Value 93.80 below lower IQR bound [97.20, 98.70]",
"Sudden rate change: -4.6% (threshold: 2.1%)"
],
"Statistical": [
"Z-score threshold: 2.74σ (base: 2.5σ)",
"Q1=97.74, Q3=98.29, IQR=0.55"
],
"Diagnostic": [
"Detector agreement: 3 of 3 evaluated detectors"
]
},
"detectedAt": "2026-08-13T09:45:23Z",
"detectorScores": {
"ZScore": 0.84,
"IQR": 0.91,
"RateOfChange": 0.78
},
"processingTime": "00:00:00.0018200",
"completedDetectors": 3,
"evaluatedDetectors": 3,
"totalDetectors": 3,
"isPartialResult": false,
"terminationReason": "All configured detectors completed",
"expectedRange": {
"lowerBound": 97.2,
"upperBound": 98.7,
"mean": 98.01
}
}Join the Axomaly Alpha, start with a metric that matters to your SaaS, and help shape the product through real-world feedback.
Early Access
We'll help you create one business, operational, or application metric, then evaluate the results together.
Choose a metric
Start with a business, operational, or application signal that matters.
Describe what matters
Smart Metric Creation prepares a detection recommendation for review.
Share your experience
Tell us what helps, what is missing, and what you need next.
Your experience helps shape Axomaly.
What works, what doesn't, and what you need next will help guide how the product develops.
Free during alpha. No credit card required.
What you'll get
Create your first metric with direct support.
Get help when you need it from the person building Axomaly.
Use Axomaly with data that reflects how your SaaS behaves.
Your experience will help guide how the product develops.
Learn how Axomaly monitors your SaaS metrics, explains unusual behaviour, and fits into your existing workflow.
Axomaly works with timestamped time-series values and optional metadata. That includes business, operational, and application metrics such as payment success, queue depth, webhook delivery, latency, errors, and throughput.
In short
Use the same detection workflow across business, operational, and application metrics.
A static threshold alerts when a value crosses a fixed boundary. Axomaly learns the usual behaviour in a metric's history and evaluates new values against the expected range, reducing the need to maintain brittle fixed rules as patterns change.
In short
Detection based on learned behaviour, not one fixed limit.
The amount of history depends on the metric's cadence and detector configuration. Smart Metric Creation recommends the baseline requirements for your review before the metric is created. Detection begins once the required history is available.
In short
Baseline requirements are shown before you create the metric.
A detection result includes the anomaly decision, confidence, and expected range, along with the direction of change and the detector evidence used to classify the latest value as usual or unusual.
In short
See what changed and which signals shaped the decision.
No. Axomaly complements your existing logs, traces, dashboards, and alerting. Send the selected metric values you want evaluated through the REST API, then use the detection response in the workflows your team already relies on.
In short
Add adaptive detection without replacing your existing tooling.
You will receive guided onboarding and direct support as you begin with one SaaS metric. The alpha is free and does not require a credit card. Your real-world experience and feedback will help guide how Axomaly develops.
In short
Start with one metric and help shape what comes next.
During the Axomaly Alpha, submitted metric values, timestamps, metadata, and detection results are hosted in the EU and retained for up to 12 months. Data is automatically deleted on a rolling basis after that period. Metric data and detector state are isolated by organisation and are never shared between customers. Data is encrypted in transit using HTTPS/TLS.
In short
EU hosting, organisation-isolated data, and a defined retention period.
Axomaly works with timestamped time-series values and optional metadata. That includes business, operational, and application metrics such as payment success, queue depth, webhook delivery, latency, errors, and throughput.
In short
Use the same detection workflow across business, operational, and application metrics.
A static threshold alerts when a value crosses a fixed boundary. Axomaly learns the usual behaviour in a metric's history and evaluates new values against the expected range, reducing the need to maintain brittle fixed rules as patterns change.
In short
Detection based on learned behaviour, not one fixed limit.
The amount of history depends on the metric's cadence and detector configuration. Smart Metric Creation recommends the baseline requirements for your review before the metric is created. Detection begins once the required history is available.
In short
Baseline requirements are shown before you create the metric.
A detection result includes the anomaly decision, confidence, and expected range, along with the direction of change and the detector evidence used to classify the latest value as usual or unusual.
In short
See what changed and which signals shaped the decision.
No. Axomaly complements your existing logs, traces, dashboards, and alerting. Send the selected metric values you want evaluated through the REST API, then use the detection response in the workflows your team already relies on.
In short
Add adaptive detection without replacing your existing tooling.
You will receive guided onboarding and direct support as you begin with one SaaS metric. The alpha is free and does not require a credit card. Your real-world experience and feedback will help guide how Axomaly develops.
In short
Start with one metric and help shape what comes next.
During the Axomaly Alpha, submitted metric values, timestamps, metadata, and detection results are hosted in the EU and retained for up to 12 months. Data is automatically deleted on a rolling basis after that period. Metric data and detector state are isolated by organisation and are never shared between customers. Data is encrypted in transit using HTTPS/TLS.
In short
EU hosting, organisation-isolated data, and a defined retention period.
Join the Axomaly Alpha with a SaaS metric that matters to you. Get guided onboarding, explainable detection, and a direct voice in how the product develops.