Outgrowing the tool that held all the answers
Contentful is the composable content platform behind the digital experiences of more than 4,800 organizations. Every month, it delivers content through more than 180 billion API calls, helping customers power websites, mobile apps, and digital experiences around the world.
By 2024, Contentful’s observability stack had grown into a patchwork of more than 11 tools, each chosen by whichever team adopted it first. In order to continue delivering reliable service to its customers, Contentful is actively reducing engineering overhead and operational costs caused by fragmented tooling.
Engineers ship continuously, and when something breaks in production, logs are usually the first place they look. As an investigation progresses, Contentful’s on-call engineers might work across several tools at once, with multiple windows open, to diagnose a single production issue. For years, logs lived in Splunk, isolated from other observability context. However, after successfully migrating its infrastructure and application performance monitoring to Datadog, Contentful wanted to establish end-to-end correlation between its logs, metrics, and traces for better visibility and troubleshooting.
When logs became the debugger of last resort
Contentful had no continuous profiling for its Node.js services, so engineers filled the gap the only way available to them: they logged more. “People would put more and more application-level logs into Splunk and use that as a debugging tool, which led to a lot of bloat and lengthy queries,” says Michael Pearce, Senior Engineering Manager at Contentful.
Indexes kept expanding. Queries dragged. And because SPL rewarded specialized expertise, access remained constrained to a limited set of technical specialists across engineering and customer experience. With its Splunk contract approaching renewal, Contentful decided it was time to migrate logs to Datadog alongside the rest of its observability platform.
A pilot the teams demanded to keep
Contentful tested the move rather than announcing it. Engineers dual-shipped logs from the largest production services into both Splunk and Datadog in order to validate capability without risking visibility. In Datadog, they routed the majority of logs to the Flex tier for efficient yet readily queryable log indexing. Then, they handed Datadog access to the people who used Splunk every day.
The reaction surprised them. “People were coming back saying, this is so much quicker to query the data I’m looking for. The query speed was just mind-blowing, for logs in Standard and Flex indexing alike,” says Pearce.
Because Datadog structures logs into searchable fields before ingestion, teams could easily narrow results by selecting facet values in the Log Explorer instead of composing complex SPL by hand. “We were expecting big pushback, and it was quite the opposite,” adds Pearce. “Our engineers find Datadog Log Management to be the faster, better experience, even enabling us to explore new use cases.”
Senior leaders who had spent years running a saved Splunk query each month and pasting the output into a spreadsheet now build a dashboard once, share it, and watch the data over time. Access to log-based queries, notebooks, dashboards, and alerts widened from roughly 200 engineers to more than 500 people across leadership, engineering, product, customer experience, security, and IT.
“People were coming back saying, this is so much quicker to query the data I'm looking for. The query speed was just mind-blowing.”
When it was time to migrate, the team successfully moved over 9 TB of daily log telemetry in six weeks. This was achieved through a structured approach that integrated AI-assisted migration workflows. Contentful ranked its Splunk indexes by size and worked downward, using Splunk’s own usage telemetry to decide which alerts and dashboards earned a migration.
For the long tail of AWS Lambda logging changes, the team built an agent using the Datadog MCP server, opened pull requests and classified each as green, amber, or red based on confidence, so teams reviewed changes instead of writing them. “We’re not the experts in everyone’s code. But [the Datadog MCP server] helped us speed up that migration,” says Pearce.
Building on clarity
After a rapid migration, Contentful retired Splunk for log management and onboarded its teams to Datadog. Pearce prepared enablement sessions and even built a migration assistant around Datadog documentation, but few people needed either. The migration proceeded smoothly, and teams quickly became comfortable using the platform. “They’re not coming to us with queries, and that’s usually a good sign. If there’s something wrong with the platform, our support channel will be full of queries,” says Pearce.
“We're not the experts in everyone's code. But [the Datadog MCP server] helped us speed up that migration.”
Contentful’s availability was already strong before the migration. The goal was to maintain that level of reliability as the business continued to scale. With logs, metrics, traces, and profiles unified in Datadog, engineers can investigate issues without switching tools, while SLO dashboards and DORA metrics help the team measure reliability alongside software delivery performance. The additional visibility is already paying off. During the proof of concept, enabling Continuous Profiler on Contentful’s Node.js services uncovered a months-old performance issue within 15 minutes, and Watchdog identified anomalous metrics that no one had thought to alert on.
Together, these capabilities help the team identify potential issues earlier and maintain confidence as they ship continuously. “Our availability metrics are excellent, but we feel that better observability is the way to maintain that and even improve on it if possible,” says Pearce. “We want to ensure that we catch any issues before they ever reach our customers.”
