Microsoft Clarity MCP Server: limits, use cases, and the Shopify alternative

What the Microsoft Clarity MCP server is

In mid-2025 Microsoft shipped an MCP server for Clarity. The release acknowledged what every analytics vendor was figuring out at the time: AI assistants are becoming the new query layer for data, and tools without an MCP path will sit outside that workflow. Clarity moved first among the major free web analytics tools.

The server bridges Microsoft Clarity's analytics with any MCP-compatible AI client. You install it via npx, paste a Clarity-issued API token, point Claude or ChatGPT at it, and ask questions in natural language. The technical execution is fine. The protocol implementation is correct.

Where it falls short is what gets exposed and at what rate.

What you can actually do with Clarity MCP

Within its limits, Clarity MCP answers basic web analytics questions in plain English.

The metrics exposed:

• Scroll Depth
• Engagement Time (total and active)
• Total Traffic
• Sessions and Bot Sessions
• Distinct Users
• Pages per session

The dimensions you can filter by:

• Browser
• Operating System
• Country / Region
• Device type

A representative query Clarity MCP can answer well: "What was our engagement time on mobile vs desktop over the past 2 days, broken down by country?" Three dimensions (device, country, time period), within the 3-day window, returns clean.

For a site doing demographic-level traffic analysis, this is workable. For a Shopify operator trying to figure out whypeople are or aren't buying, the data shape doesn't reach the question.

The four limits that matter

Microsoft documents the limits clearly. They are not hidden. But their cumulative effect on real workflows is rarely spelled out.

1. 10 API requests per day, per project

The most quoted limit, and the one that breaks workflows fastest. When you ask Claude one question via MCP, the AI typically makes 3-5 internal tool calls to answer it. It decomposes the question, runs sub-queries, validates. So 10 daily API requests usually means 2-3 real questions before the project is rate-limited for 24 hours.

Practical effects:

• A team of 2-3 people sharing one project: quota gone by lunchtime.
• Any agentic workflow that polls anomalies hourly: out in 30 minutes.
• Iterative drill-down ("now show by device, now compare to last week, now for returning users"): one drill-down session, then locked.
• Scheduled analysis cron jobs that depend on Clarity MCP: limited to 10 invocations per 24h.

2. 3-day maximum data lookback

Clarity MCP only exposes the most recent three days of data. Practical effects:

• Cannot do week-over-week comparisons.
• Cannot see seasonal patterns or recurring weekly cycles.
• Cannot answer "did anything change after the homepage redesign" if the redesign was four or more days ago.
• Anomaly detection that compares current to baseline becomes impossible because the baseline window does not exist.

3. 3 dimensions per request

Each request can apply up to three dimension filters. With only four total dimension types available (browser, OS, country, device), the combinations are quickly exhausted, and most real ecommerce questions need dimensions Clarity MCP does not have.

4. Web analytics, not ecommerce

This is the biggest gap for Shopify. Clarity MCP has no concept of:

• Products viewed or added to cart
• Cart value or composition
• Session outcomes (purchase, abandonment)
• Revenue attribution
• Funnel steps and drop-off positions
• Returning versus new customer behavior at the page level

For a non-ecommerce site these absences are fine. For a Shopify store, these are the only questions that matter. Clarity MCP cannot answer them because the data was never tied to Shopify events at ingest.

Where Clarity MCP makes sense

Being fair to it: Clarity MCP is genuinely useful for a specific audience.

• Personal blogs and content sites where 10 questions a day is more than enough.
• Casual exploration:"just curious about the mobile vs desktop split this week."
• Hobby projects on a $0 analytics budget that already use Clarity.
• Teams already invested in Clarity's dashboard who want a low-effort AI on-ramp without changing tools.

It is free. It works. For the right use case it is perfectly fine. The criticism here is not that the product is bad. It is that the product is web analytics, not ecommerce analytics, and running a Shopify store hits the ceiling fast.

ClickContext: how it differs for Shopify

We built ClickContext for the gap. It is the same MCP protocol, different priorities, designed from day one for Shopify and AI consumption rather than dashboards.

• One-click install from the Shopify App Store. No theme edits, no script tagging, no Google Tag Manager.
• E-commerce context attached to every behavioral event: product viewed, cart value, session outcome, revenue impact.
• 250,000 sessions per month on the Pro tier (free tier handles up to 5,000).
• 90-day data retention on Pro (14 days free). Enough to do week-over-week and month-over-month analysis.
• No daily API request cap. Agentic workflows, shared team access, iterative drill-down all work.
• Funnel tracking with revenue correlation. Each funnel step is tagged with the products and cart values at that step.
• Anomaly detection alerts on Pro.
• EU-hosted, merchant-owned data. Cleaner GDPR posture than a US-hosted free tool.

Honesty caveat: ClickContext costs money past the free tier. Clarity MCP is free. The trade-off is real and worth thinking about: at $99/mo, Pro is roughly the cost of a takeaway lunch a week. The question is whether removing the data quota and adding ecommerce context is worth that for your team.

Side-by-side comparison

Using them together

The two tools are not mutually exclusive. For stores that already run Clarity and want to add ecommerce-specific MCP, the right move is usually both.

The split:

• Clarity MCP handles broad demographic and traffic questions: who is visiting, on what device, from where, how engaged.
• ClickContext handles ecommerce-specific behavior: where are mobile shoppers abandoning carts, which products lose customers at the size chart, did the homepage redesign change behavior of returning customers.

Both can run inside the same Claude or ChatGPT window because MCP supports stacking multiple servers. The AI assistant picks which server to query based on the question. For a Shopify store that wants the broadest possible MCP-accessible analytics footprint, this is the answer.

Questions to ask when choosing

Quick decision-helper. Run through these honestly:

• Do you need data older than 3 days?
• Do you need to ask questions about products, carts, or revenue?
• Will your team or your AI agents make more than 10 queries per day?
• Are you running a Shopify store, or is your site non-ecommerce?
• Do you want to run scheduled or automated analyses?

If most of your answers are "yes," ClickContext fits better. If most are "no," Clarity MCP is probably fine.