Skip to content
Agents tracked: 502 Releases (7d): 613 Agent status: 4 with issuesSince last limit reset: Claude 7d · Codex 5d For AI agents: Markdown Updated Nov 18, 2026
Case study SupermetricsResults as reported by Anthropic

Supermetrics' Claude connector cut an agency's 10-hour reports to 20 minutes

Supermetrics, which moves marketing data for more than 200,000 businesses, built a Claude connector on its MCP server so marketers can ask Claude Cowork about live ad and analytics data and then create or edit campaigns in the same conversation, with human approval. Active users of the connector grew 250% month over month on average since its February launch, and one agency's 10-hour client reports now take 20 minutes, as reported by Anthropic.

The problem

Dashboards only answer the questions their designer anticipated. Anything else meant waiting on a technical colleague or exporting data from several platforms and reconciling it by hand: Layer, a 25-person Norwegian agency, quoted a client 10 hours of billable work for one report that pulled from five different sources. A wrong number in a client report costs real budget, so many agency teams did not trust AI assistants with client-facing work.

How they did it

  1. Add the Supermetrics connector in Claude - in Customize, then Connectors, search for Supermetrics (or use Try in Claude on Supermetrics' Claude page) - and pick the Supermetrics team to use.
  2. Connect the ad and analytics sources you need in Supermetrics first; the connector returns normalized data with the source, timestamp and transformation logic attached to every figure.
  3. Ask questions in Cowork in plain language and name Supermetrics in the prompt if you have several data connectors; Cowork renders the answers as charts inline in the conversation.
  4. Validate the output on your own reports before showing clients - Layer spent weeks checking Claude's numbers against its own data first.
  5. Turn on write actions only with guardrails - campaigns are created in a paused state, approval levels are set per account, and an approval page lists each change for stakeholders to approve or reject.
  6. Move routine checks to scheduled answers, such as daily pacing checks in Slack and anomaly alerts, instead of a monthly reporting scramble.

Results as reported by Anthropic

  • Active users of Supermetrics' Claude connector grew 250% month over month on average since its February launch, as reported by Anthropic.
  • At agency customer Layer, client reports that took 10 hours of manual spreadsheet work now take 20 minutes, and 17 of those minutes go into dropping the finished data into the client's Google Sheet, as reported by Anthropic.
  • Layer recorded no hallucination or incorrect data point in its multi-week validation of live marketing data, as reported by Anthropic.
  • Annual media plans, once too time-consuming to build thoroughly, are now standard practice at Layer, and teams that optimized weekly now optimize daily, as reported by Anthropic.

These figures are as reported by Anthropic in Anthropic customer story ↗. Anthropic sells the product used here. AgentGid did not measure or independently verify them.

Takeaway. Marketers trust an AI assistant with client numbers when every figure carries its source and timestamp and every campaign change waits for a human to approve it.
AgentGid's take

This suits agencies and in-house marketers who already live in several ad platforms and spend hours stitching exports into reports; it suits less a team whose single dashboard already answers its questions. You pay twice: Claude, and Supermetrics' AI chats subscription (Supermetrics' docs say a trial starts at first login). Per AgentGid's Claude Cowork page (pricing checked Nov 4, 2026), Claude has a free tier, Pro is $17 a month on annual billing ($20 monthly), Max $100, Ultra $200, Team standard seats $20 per seat a month annual ($25 monthly) and premium $100 ($125), Enterprise $20 per seat a month annual with a 20-seat minimum plus usage. Caveats: every figure comes from Anthropic's own customer story, with no publication date; the report time is one agency's experience, and "no hallucinations" is that agency's own check, not an audit. Keep campaigns paused by default and approve each change. If you want an open-source route, n8n can schedule the same data pulls into Slack, per our catalog.

The agent used here

Similar use cases

Guides