Prepare your next client review with answers from connected Amplitude data. Catchr MCP lets freelancers and agencies ask ChatGPT or Claude about exported event volumes, geographic breakdowns and upload timing.
Connect your client's Amplitude project, choose your AI destination and start with a reporting question. Count events, compare countries or investigate when events reached Amplitude.
“Compare paid media performance across my client accounts and flag the biggest changes.”
Amplitude
Google Ads
GA4
Used by the world's leading companies
Build your analysis around Number of Events and the available event dimensions. Amplitude export date ranges follow Server Upload Time, so keep that basis explicit when comparing periods or explaining results to clients.
Sum Number of Events to measure exported event volume. Each exported event contributes one, giving you a clear starting point for comparing reporting periods.
Event Id identifies an event within the export payload. It is a dimension, not a metric to sum, and event volume should not be presented as a count of people.
Use Country, Region and City alongside Number of Events to see how exported activity is distributed geographically. Compare locations between reporting periods to prepare focused questions for a client call.
Keep missing locations visible in the results. These fields describe the geography recorded on events; City is derived from the event IP address.
Compare Client Upload Time with Server Upload Time to examine the gap between client upload and server upload. Export windows use Server Upload Time, rather than the original event timestamp, and exported data can lag ingestion by up to two hours.
User Properties and Data are available as JSON text. Discover the available columns and inspect the returned values before selecting properties to analyze; nested keys should not be assumed to exist as separate columns.
A focused event investigation, a recurring client report and a question in ChatGPT or Claude serve different needs. Choose the approach that fits the Amplitude question and the output your client needs.
Use a focused breakdown to investigate event activity, a report to present recurring results, and Catchr MCP to ask follow-up questions. Across all three approaches, count exported events with Number of Events and keep upload-based date ranges clear.
Connect your client's Amplitude project to Catchr using its project API key and secret key. Use the Standard or EU connection endpoint that matches the project's data residency.
Then connect ChatGPT or Claude through Catchr MCP. Start by discovering the available Amplitude fields, select the columns you need and specify your reporting dates. Keep credentials out of your analysis prompts.
Replace the bracketed values with your client's project, reporting dates and thresholds. Each prompt defines a specific analysis and keeps the meaning of exported event data explicit.
Using Catchr, compare total Number of Events for Amplitude project [project] between [first date range] and [second date range]. Sum event_count for each period and show the absolute and percentage change. If the first period has zero events, mark the percentage change as undefined. Label both periods as Server Upload Time windows and flag any period that could be affected by the export lag of up to two hours.
Copy promptUsing Catchr, break down Number of Events by Country for Amplitude project [project] across [first date range] and [second date range], using Server Upload Time for both windows. Show each country's totals and absolute change, keep missing country values in a separate row, and highlight changes larger than [event threshold]. Rank countries by the size of their absolute change and suggest questions to discuss with the client without assuming a cause.
Copy promptUsing Catchr, retrieve Event Id, Client Upload Time and Server Upload Time for Amplitude project [project] within the server upload window [start date] to [end date]. Calculate the gap between the two upload timestamps where both are present. Show events with gaps above [minutes threshold], identify missing timestamps or negative gaps separately, and explain that this measures upload timing rather than time since the original event occurred.
Copy promptUsing Catchr, discover the available Amplitude columns for project [project] and retrieve my selected columns [selected columns], including User Properties and Data, for the server upload window [start date] to [end date]. Inspect the JSON text and summarize which property keys appear in the returned records, with missing or unparseable values identified separately. Limit the summary to the retrieved records and ask me which observed keys to analyze next.
Copy promptNumber of Events counts exported events: each event contributes one, and summing the metric gives the total. Event Id is an identifier, not a count. Event totals do not represent unique people.
You need the Amplitude project's API key and secret key, plus the correct Standard or EU endpoint for its data residency. Connect the project in Catchr, then use ChatGPT or Claude through Catchr MCP to ask questions about the available fields.
Export date ranges use Server Upload Time, not the original event timestamp. An event can therefore fall into a different reporting window. Exported data can also lag ingestion by up to two hours, which matters when checking recent periods.
User Properties and Data are exposed as JSON text rather than separate columns for each nested key. Start by discovering available fields and inspecting returned values, then choose the observed properties you want to analyze. Do not assume every event contains the same keys.
Connect your sources to Catchr, add the MCP server to your AI assistant, and ask about the work that matters now.
