North-Star and Guardrail Metrics for Google Cloud Storage

You are the product manager for Google Cloud Storage, the object storage service in Google Cloud where customers keep files and other unstructured data in buckets. How would you measure whether the product is succeeding? Define what success means for this product, pick the metric you would treat as the primary measure, and lay out the supporting and guardrail metrics you would track around it. ```hint Value is more than volume Customers put data in object storage for very different jobs: serving content to an application, feeding analytics and ML pipelines, or keeping backups and archives. Ask what a successful customer looks like for each job, and whether total bytes stored would rise for the right reasons. ``` ### Clarifying Questions - Is the scope only the object storage service, or also neighboring Google storage products such as Drive, Filestore or the disks attached to virtual machines? - Is success judged for the storage line on its own, or also by how much it pulls customers into other Google Cloud services? - Should the metrics favor winning new customers, given that Google Cloud trails larger cloud rivals, or growing and keeping existing accounts? - Which customers matter most: startups, mid-size companies or large enterprises, and cloud-native applications or hybrid setups that copy on-premises data to the cloud? ### What a Strong Answer Covers - A short description of the product, its main workloads and storage classes, and a definition of success tied to what customers and Google Cloud each need from it - Customer segments and workloads, and how a healthy account looks different in each - A primary metric that reflects value delivered rather than raw volume or vanity counts, with the reasoning for choosing it - Supporting metrics across the customer lifecycle (acquisition, activation, engagement, revenue, retention) and service-quality measures such as availability, latency and durability - Guardrails and interpretation: unexpected bills, security misconfigurations, and metrics that can rise for the wrong reasons ### Follow-up Questions - Total stored bytes grow quickly while the number of active paying customers stays flat. Is that success, and what would you check first? - Lifecycle rules that move data to colder, cheaper classes lower revenue per byte. Should the team encourage them, and how would your metrics show whether that was the right call? - Which early signal would warn you that a large customer is about to move its data to another cloud?