Reduce unexpected costs with the new AI-powered Cost Anomaly Detection

Controlling runaway spend and minimizing unexpected costs is a priority for every business. Imagine a scenario where faulty development or rogue code results in a usage spike over the weekend, unbeknownst to you. If not caught in time, this kind of usage can result in cost spikes that can exhaust your budgets and put a strain on finances. 

At Google Cloud, we provide customers with a comprehensive set of cost management tools and controls to help prevent surprises. Now, we’re expanding our FinOps capabilities with AI technology that further simplifies cost management and helps ensure spend predictability. At Google Cloud Next ’24, we announced Cost Anomaly Detection and today, it's available to all customers in public preview. Cost Anomaly Detection helps identify anomalies in real or near-real-time and enables timely alerts so that you can avoid surprises, take swift action and control runaway costs. 

Getting to know Cost Anomaly Detection

Google Cloud’s Cost Anomaly Detection can help you identify unusual spikes in cloud spending, across all products and services, by automatically monitoring your cloud projects and displaying any spikes in your billing console. This product does not require any setup and is available at no cost for all customers. Important components include:

1. Detection

Using AI, Cost Anomaly Detection identifies your spend patterns based on historical and seasonal trends and forecasts an expected rate of daily spend specific to your project. It continuously monitors your actual spend every hour and detects any deviation. These deviations are then identified as spikes or anomalies — a.k.a. ‘cost impact’ within the Cost Anomaly Detection dashboard. Since Cost Anomaly Detection monitors your spend on an hourly basis, it can identify any unexpected upward spikes within 24 hours, for most services, detecting anomalies in near real-time.

2. Investigation

Once an anomaly is detected, you want to understand its root cause. For each anomaly it identifies, Cost Anomaly Detection provides a detailed, easy-to-understand root-cause analysis that lists the top contributors to the spend. This allows you to narrow your investigation on the exact project, service, region or SKU that needs corrective action, thereby enabling quicker remediation. 

3. Alerts

Once you know of an anomaly and its root cause, the appropriate owners need to be alerted of the impact to their respective projects, so they can cap or turn off usage. Today, anomaly notifications are sent through email and Pub/Sub, allowing for a wide range of personas to be notified, from the FinOps team to engineering. Cost Anomaly Detection also lets you easily set up customizable alert preferences that notify a set of desired recipients of an anomaly as soon as it is detected, while Pub/Sub alerts help with integration with your internal workflow management tools.

Cost Anomaly Detection also lets you tailor your alerting threshold, based on cost impact, so that only significant anomalies are displayed and alerted. We recommend monitoring anomalies for at least one month before defining a threshold that applies across all your projects. 

Additionally, Cost Anomaly Detection is continuously learning about your spend patterns, helping to reduce the possibility of false positives and increase sensitivity to not only monthly and seasonal trends, but also inter-day and inter-week fluctuations. To that end, for every identified anomaly, you can provide feedback on whether it was truly unexpected or a false positive due to, for example, a planned migration. This feedback helps the Cost Anomaly Detection AI models adapt in real-time, to your usage and take planned usage into consideration when evaluating future spikes.

Enhanced cost observability

With Cost Anomaly Detection, you have another way of optimizing your spend: controlling unintended cost. This, when coupled with existing tools such as Budgets, allows for a more robust and flexible cost-control governance. The product requires no setup, detects same-day anomalies, and enables focused action through detailed root-cause analysis and near-real-time alerts. If you’re already using your own anomaly detection solution, we encourage you to try Cost Anomaly Detection for free, to compare and contrast the results and the customizable controls available. 

Head over to the Google Cloud billing console to access this experience and start elevating your FinOps game! For more details on this product, read the documentation here.