Midpage MCP

@Midpage fetch the latest analytics summary for our landing page and show me the top 3 traffic sources from the past week.

Based on the Midpage data, your landing page received 2,847 visits in the past 7 days. The top traffic sources were: (1) Organic search — 1,203 visits (42.3%), primarily from Google with keywords related to 'project management tools'; (2) Direct traffic — 891 visits (31.3%), suggesting strong brand recall or bookmark usage; (3) Social referrals — 456 visits (16.0%), mostly from LinkedIn posts. Bounce rate averaged 38%, with mobile visitors spending 22% longer on the page than desktop users.

This example assumes read access to analytics data. The MCP requires OAuth2 authentication with appropriate scope to access page metrics. Response format and available metrics depend on what Midpage exposes through its API — actual data granularity may vary by account tier.

@Midpage create a new A/B test variant for the hero section with the headline 'Ship faster with your team' and allocate 30% of traffic to it.

A/B test created successfully. Variant B is now live with the headline 'Ship faster with your team' and will receive 30% of incoming traffic. The control (Variant A) retains your current headline and receives 70%. Test ID is `test_hero_2024_01_15`. Midpage will begin collecting conversion data immediately — you can view results in the dashboard or query them here after the test accumulates at least 100 conversions per variant.

This write operation modifies live page configuration and affects real visitor experience. Ensure you have edit permissions and understand the test's impact on conversion tracking. The MCP cannot roll back tests automatically — you must monitor results and manually conclude or adjust traffic allocation.

@Midpage compare conversion rates across all active page variants this month and recommend which version we should promote to 100% traffic.

Across 4 active variants this month, Variant C ('Collaborate in real-time') shows the strongest performance: 8.2% conversion rate over 1,456 visitors, compared to 6.1% for the control. Statistical significance reached (p=0.03). Variant C also reduced bounce rate by 11% and increased time-on-page by 34 seconds. Recommendation: promote Variant C to 100% traffic. Variants A and D underperformed (4.9% and 5.3% respectively) and can be archived. Note that Variant C's audience skewed 60% mobile — verify desktop performance before full rollout.

This synthesis example combines analytics retrieval with AI reasoning to surface actionable insights. The recommendation depends on sufficient sample size and statistical validity — premature conclusions from low-traffic tests can mislead. Always verify the AI's statistical claims against Midpage's native confidence intervals before making irreversible changes.

When OAuth2 matters more than tool count

A 3-person founding team runs weekly user-interview synthesis sessions in Switchy. They need to pull notes from multiple sources without storing API keys in a shared workspace. Midpage MCP uses OAuth2, so each team member authenticates once and the connection persists across sessions. This works if your team rotates who runs the sync and you want audit trails tied to real users. The trade-off: OAuth2 adds setup friction for solo founders who just want a quick API key. If you're the only person touching the integration and speed matters more than attribution, a simpler key-based MCP is faster. For teams of two or more who share workspace access, the OAuth2 flow pays off by the second week.

OAuth2 as the floor for regulated teams

A 6-person support team at a fintech startup uses Switchy to answer tier-2 tickets. They need to surface customer data from internal tools, but their compliance officer requires user-level auth for any integration that touches PII. Midpage MCP's OAuth2 flow meets that bar—each support agent authenticates individually, and access revokes when they leave. This scenario assumes the MCP will eventually expose tools that read sensitive records; if it stays read-only or public-data-only, the OAuth2 requirement is overkill. The boundary: if your compliance framework mandates user-attributed access logs, OAuth2 is non-negotiable. If you're pre-revenue or handling only public data, a simpler MCP saves setup time.

When OAuth2 scales across client boundaries

A 10-person agency uses Switchy to manage client projects. Each client has their own Midpage account, and the agency needs to pull data without asking clients to generate API keys. Midpage MCP's OAuth2 flow lets the agency authenticate once per client workspace, and clients can revoke access from their own admin panel. This works when you manage 5+ client accounts and turnover is frequent—OAuth2 centralizes credential management and makes offboarding clean. The limit: if the MCP's tool library stays empty or generic, the OAuth2 overhead isn't worth it compared to a direct API script. For agencies juggling multiple client tenants and needing durable, revocable access, the OAuth2 model is the right call by client three.