Cloud cost anomalies are uniquely painful: you don't find out about them until the end of the month, by which point the spending has already happened and the cause is often difficult to trace. A misconfigured Lambda, an accidental NAT gateway, a forgotten data transfer — these can add thousands of dollars to a bill before anyone notices.
Applicare correlates your AWS spend with application events — deploys, traffic surges, configuration changes — so you know why costs are changing before the bill arrives.
Applicare ingests AWS Cost Explorer data at hourly granularity and builds per-service spend baselines — the same per-entity baseline approach used for performance metrics. When spend deviates from the baseline for a given service or resource type, IntelliSense flags it immediately rather than waiting for a budget alert threshold.
More importantly, Applicare correlates cost changes with the entity graph — so when spend spikes, it can tell you which deploy, which configuration change, or which traffic event caused it.
1. Data transfer from a new deployment
A service update that adds inter-region API calls can quietly add hundreds of dollars per day in data transfer charges. Applicare detects the new connection pattern in the entity graph and correlates it with the spend increase within hours of deployment.
2. Lambda cold start cascade
A Lambda function configured with insufficient provisioned concurrency will cold-start under load, dramatically increasing both execution time and cost. Applicare correlates Lambda execution duration spikes with cost increases and identifies the functions responsible.
3. Forgotten development resources
RDS instances, NAT gateways, and load balancers left running after a dev project ends are a persistent source of wasted spend. Applicare's entity graph identifies resources with zero traffic that continue to generate charges.
One of our customers discovered a NAT gateway processing 2TB per day of traffic from a misconfigured logging agent — $180/day in charges that had been running for 6 weeks. Applicare flagged it within 4 hours of the agent deployment.
Beyond anomaly detection, Applicare continuously analyses resource utilisation against actual workload patterns to generate rightsizing recommendations. Unlike AWS's built-in rightsizing tools, Applicare's recommendations are correlated with application performance data — so it won't suggest downsizing an instance that's running at 30% CPU but serving latency-sensitive traffic that would degrade on a smaller instance.
Across customers who implemented Applicare's rightsizing recommendations and resolved the anomalies it detected in their first 90 days, the median AWS cost reduction was 40%. The biggest contributors: rightsizing over-provisioned RDS instances (avg 35% of savings), eliminating data transfer anomalies (avg 28%), and right-sizing EC2 instances based on actual utilisation (avg 22%).