Agency promptCopy prompt
Replace [variables] Facts ≠ hypotheses Action + follow-up metric
Trusted by marketing teams that talk to data every day
Three steps, from Catchr to your first answer. No code, and nothing to install on your machine.
Meta Ads
3 advertising accounts
Connected
Google Ads
2 advertising accounts
Connected
Google Analytics 4
1 web property
Connected
HubSpot
1 CRM portal
Connected
Open the Catchr MCP entry in the Claude connector directory, add it, then sign in to Catchr and review the access Claude asks for.
Replace account, period, KPI target, and threshold variables before you ask Claude.
Which one fits depends on what you need Claude to do, and on how many sources the question touches.
Compare your options
Sources
Sources
Sources
The data in your file
Connection
Connection
One Catchr workspace
Connection
Upload a CSV or Excel file
Next question
Next question
Query connected sources
Next question
Export again if the scope changes
Catchr gives Claude eight tools, and all eight of them read.
Claude can list the platforms and accounts connected to your workspace, browse the available fields, and run reporting queries against them. That covers spend, delivery, conversions, and creative performance, down to the level you ask for.
Claude cannot create, pause, or edit a campaign, move a budget, publish an ad, or change an audience. There is no write permission to grant, so there is none to supervise.
In Ads Manager. Decide in the conversation what to change and why, then act with the access and the review your team already uses.
Choose your role, open a workflow, then replace the bracketed variables before pasting the prompt into Claude. Every prompt asks for evidence, limits, and a concrete output.
“Why did CPA rise?” sounds simple. The answer is usually buried across exports, formulas, filters, and tabs before the analysis can even start.
Choose the right account, date range, breakdowns, and fields. Then wait for the file and do it again when the question changes.
Rename columns, fix formats, join tabs, and check formulas before you can trust the comparison.
Turn the numbers into a client-ready answer, then repeat the whole process for the inevitable follow-up.
Catchr MCP gives Claude access to the marketing accounts you select, so each follow-up starts with the same marketing context instead of a new manual export.
Clear answers for marketers comparing connectors, data sources, reporting workflows, and dashboard setup.
Claude can analyze Amplitude event records returned through Catchr, including event IDs, attribution identifiers, cities, IP addresses, client upload times, and server upload times. Nested event, user, group, and data properties are available as JSON strings rather than separate child tables.
Claude can organize returned Amplitude events by hour, day, day of the week, month, quarter, client upload time, or server upload time. Specify the period, desired time grain, and UTC handling in your prompt so the resulting analysis is clear.
Yes. When both timestamps are returned, Claude can compare the UTC client upload time with the server upload time to highlight differences in event delivery. This does not establish the cause of a delay, so ask Claude to separate measured timing differences from possible explanations.
Claude can flag patterns supported by the returned event records, such as changes by city, attribution identifier, hour, day, or upload timestamp. Ask it to show the relevant records and distinguish observations from hypotheses before recommending further investigation.
Useful prompts define the project, date range, time grain, filters, and required output. For example, ask Claude to summarize event records by city, compare client and server upload times, review attribution identifiers, or inspect selected JSON property fields for recurring patterns.
Amplitude group properties, user properties, groups, event properties, and data objects are returned as JSON strings. Claude can inspect their contents when present, but you should name the property or pattern you want reviewed instead of assuming each nested object has a separate schema.
Yes, provided each Amplitude project is connected and available through Catchr. Request separate project results first, then a combined comparison using the same period, time grain, timezone, and field definitions so differences between projects remain visible.
You need the Amplitude project API key and secret key, access to a Catchr workspace, and Catchr MCP enabled in Claude. Select the correct Standard or EU residency connection for the project, then Claude can analyze the Amplitude records available through that connection.
Claude analyzes the records returned by the Amplitude Export API and should not treat them as real time. Export windows use server upload time, exported data can lag ingestion by up to two hours, and available history depends on the requested period and the records successfully returned by the source.
No. Catchr provides Amplitude data to Claude for analysis; it does not authorize Claude to edit events, properties, projects, or account settings. Use Claude's findings to prepare an action plan, then make reviewed changes in the appropriate operational system.
Our team is always here to answer any questions you may have about our data connector.
Connect your marketing platforms to Claude with Catchr MCP and start investigating current campaign performance.