Comet ads, brutally reviewed.
Comet is running 13 live ads on LinkedIn and Google. Their average fit to their own ideal buyer is 7.5 out of 10, and 2 ads need fixing. Each ad below is one of Comet’s real live creatives, read straight from the ad libraries.
Comet’s live ads, scored against their buyer
- The open-source eval engineering platform, scored 7/10 against Comet’s ideal buyer. Sharp ICP hook and product screenshot land well, but no proof or CTA to convert.
- Open-source LLM evals you can run on OpenShift, scored 8/10 against Comet’s ideal buyer. Tight segment hook on OpenShift plus on-prem data-residency addresses a real enterprise blocker precisely.
- Help improve Opikāand earn $200, scored 6/10 against Comet’s ideal buyer. Incentive is clear but the ask is high effort for existing non-users, limiting qualified response rate.
- Send your OpenTelemetry traces to Opik, scored 8/10 against Comet’s ideal buyer. Meets engineers exactly at their existing OpenTelemetry workflow, removing adoption friction with a concrete integration path.
- Tune your eval until it agrees with your reviewers, scored 8/10 against Comet’s ideal buyer. Specific before/after metric (87% to 97%) makes the optimizer's value concrete and credible for eval engineers.
- AI FinOps for coding agents, scored 8/10 against Comet’s ideal buyer. Specific cost-breakdown UI hits the token-spend pain point with real dollar figures and named tools.
- A Langfuse alternative that tunes your prompts, scored 8/10 against Comet’s ideal buyer. Named rival plus six concrete algorithms and a live metric lift make a compelling switch case.
- Build your evals directly on LiteLLM, scored 8/10 against Comet’s ideal buyer. One-line integration promise with real config snippet removes the biggest adoption objection for LiteLLM users.
- Don't import evals. Engineer them., scored 7/10 against Comet’s ideal buyer. Punchy contrarian hook and versioning workflow speak directly to engineers who distrust black-box eval libraries.
- Find what's burning your tokens, scored 8/10 against Comet’s ideal buyer. Pain-first hook with granular spend breakdown by MCP server, skill and tool call lands squarely on cost-overrun anxiety.
- LLM judges that run on models you've already approved, scored 8/10 against Comet’s ideal buyer. Enterprise compliance angle is precise and rare, approved-model constraint is a real blocker this ad directly removes.
- Since TypeSafe AI released Jev in September, there has been a lot of talk about , scored 6/10 against Comet’s ideal buyer. Solid product bridge but the decision-model hook is too niche and unexplained to land cold.
- Replace Galileo and LangSmith with one open-source tool, scored 8/10 against Comet’s ideal buyer. Named rivals, open-source angle, and a product screenshot make a sharp bottom-funnel case for switching.
See Comet’s full live-ad board on AdRoast. Every creative scored against their ICP, with the exact fix for each.
Learn the method behind these scores: the AdRoast guides cover how to audit your ads against your ICP and why clicks do not turn into pipeline.