Troubleshooting
Solutions to common issues including connection problems, missing data, authentication errors, and performance questions.
Verifying Your Connection
Before troubleshooting a specific issue, confirm that Presso is properly connected and your data sources are accessible.
Quick Connection Test
Ask a simple question that requires a data source response:
List my Shopify products (limit to 3)
If you get product data back, the Shopify connection is working. Repeat with a simple query for each data source you use:
Show me my GA4 sessions for yesterday
List my Google Ads campaigns (limit to 3)
Show me my Meta Ads account summary for yesterday
Check MCP Server Status
Presso runs as an MCP server within Claude Code. If nothing is responding, verify that the MCP server is configured and running:
- Open your Claude Code MCP configuration file (typically
~/.claude/or your project's.mcp.json) - Confirm that the Presso MCP server entry is present and correctly configured
- Restart Claude Code to reinitialize the MCP connection
Presso Is Not Responding
Symptom: You ask a question and Claude Code does not use any Presso tools, or it responds with general knowledge instead of your actual data.
Possible Causes and Solutions
The MCP server is not running. Claude Code needs to have the Presso MCP server configured and active. Check your MCP configuration and restart Claude Code if needed.
The question is too general. If your question does not clearly involve e-commerce data, Claude Code may answer from general knowledge instead of querying Presso. Be explicit:
Using my Shopify data, show me yesterday's revenue
Network connectivity issues. Presso needs internet access to reach your data sources. Check your network connection and try again.
The MCP connection timed out. If the connection was idle for a long time, it may need to be reestablished. Restart Claude Code or start a new session.
No Data Returned
Symptom: Presso connects and attempts to query, but returns empty results or says no data is available.
Possible Causes and Solutions
The date range has no data. If you ask for data from a period before your data source was connected or before your store existed, there will be no data. Try a recent date range:
Show me orders from the last 7 days
The data source is not connected. You may be querying a data source that has not been connected yet. Verify which integrations are active in your Presso account at presso.now.
Permissions are insufficient. The OAuth token may not have the required scopes to access the data you are requesting. Reconnect the data source to ensure all necessary permissions are granted.
The query filter is too narrow. If you are filtering by a specific product, campaign, or customer that does not exist, no results will be returned. Broaden your query:
Instead of: "Show me orders for product SKU-XYZ-999"
Try: "Show me all orders from the last 7 days"
The data source has no data for that metric. Not every data source tracks every metric. For example, Search Console will not have revenue data, and Shopify will not have impression data. Make sure you are asking the right source for the right data.
Authentication Errors
Symptom: You see an error related to authentication, invalid tokens, expired credentials, or authorization failures.
Possible Causes and Solutions
Your OAuth token has expired. OAuth tokens have a limited lifetime. When they expire, Presso can no longer access your data. Reconnect the affected data source:
- Go to your Presso account settings at presso.now
- Find the data source showing authentication errors
- Disconnect it
- Reconnect it by going through the OAuth flow again
Permissions were revoked on the source platform. If someone changed permissions in your Shopify admin, Google account, or Meta Business Manager, the existing Presso token may no longer be valid. Reconnect the data source to generate a new token with current permissions.
Multiple accounts or wrong account connected. If you have multiple Shopify stores, Google accounts, or Meta Business accounts, you may have connected the wrong one. Disconnect and reconnect, making sure to select the correct account during the OAuth flow.
The data source platform is experiencing an outage. Rarely, the source platform itself (Shopify, Google, Meta) may have an API outage. Check the platform's status page:
Slow Responses
Symptom: Queries take a long time to return results, or they time out.
Possible Causes and Solutions
You are requesting too much data at once. Asking for daily data over a 12-month period or listing all products in a large catalog takes longer. Narrow your request:
Instead of: "Show me daily revenue for the past year"
Try: "Show me weekly revenue for the past 3 months"
Instead of: "List all my products with inventory"
Try: "Show me my top 20 products by revenue with their inventory levels"
You are querying multiple platforms simultaneously. Cross-platform queries require multiple API calls. If speed is a priority, query one platform at a time:
First: "Show me Google Ads ROAS for last month"
Then: "Now show me Meta Ads ROAS for the same period"
The source platform API is slow. Occasionally, the underlying platform API may be slow due to high load. This is outside of Presso's control. Try again after a few minutes.
Large Shopify stores. Stores with tens of thousands of products or orders may experience slower queries for listing operations. Use filters and limits to reduce the data set:
Show me orders from the last 3 days (not "show me all orders")
Incorrect or Unexpected Data
Symptom: The numbers Presso returns do not match what you see in your platform dashboards.
Possible Causes and Solutions
Timezone differences. Presso queries may use a different timezone than your platform dashboard. If yesterday's revenue does not match, it may be because "yesterday" is defined differently. Be explicit about timezones when precision matters, or use specific date ranges:
Show me revenue for February 15, 2025 (UTC)
Data freshness and caching. Some platforms have a delay before data is fully processed:
| Platform | Typical Data Delay |
|---|---|
| Shopify | Real-time for orders, near real-time for inventory |
| GA4 | 24-48 hours for some reports, 4-8 hours for real-time |
| Google Ads | Up to 3 hours for conversion data |
| Meta Ads | Up to 3 hours for standard metrics |
| Search Console | 2-3 days for search performance data |
| Amazon Ads | Up to 12 hours for campaign data |
| Klaviyo | Near real-time for profiles and events; campaign reports may have up to 1 hour delay |
| ShipStation | Real-time. Some endpoints (carriers, rates) may have slower response times |
| Judge.me | Real-time for reviews and ratings |
If you are comparing Presso data to your dashboard and the numbers differ slightly, check whether the dashboard has been updated more recently.
Attribution model differences. Google Ads, Meta Ads, and GA4 each use different attribution models, which means they may report different conversion numbers for the same period. This is expected behavior. When comparing:
Show me conversions as reported by Google Ads, Meta Ads, and GA4
for last month, and note which attribution model each uses
Metric definitions vary by platform. "Revenue" in Shopify (total order value) may differ from "revenue" in GA4 (which may exclude taxes or shipping depending on your configuration). When numbers do not match exactly, ask Presso to clarify what each metric includes:
What does the revenue figure include for Shopify versus GA4?
Refunds and cancellations. Shopify revenue may include orders that were later refunded. Depending on when you query, the numbers may not account for recent refunds:
Show me net revenue (after refunds) for last month from Shopify
How to Reconnect a Data Source
If a data source is not working, follow these steps to reconnect:
- Go to presso.now and sign in
- Navigate to your account settings or integrations page
- Find the data source you want to reconnect
- Click "Disconnect" to remove the current connection
- Click "Connect" to start a fresh OAuth flow
- Select the correct account and grant the required permissions
- Return to Claude Code and test the connection with a simple query
After reconnecting, your previous conversation context will not know about the new connection. Start a new Claude Code session for best results.
Known Limitations
Understanding what Presso can and cannot do will help you set the right expectations.
Data Access
- Read-only access. Presso can query and analyze your data but cannot modify it. You cannot create orders, update inventory, change ad bids, or edit products through Presso.
- No real-time streaming. Presso fetches data on demand when you ask a question. It does not provide live dashboards or push notifications.
- API rate limits apply. Each data source imposes API rate limits. If you make a very large number of queries in a short period, some may be throttled.
Query Scope
- Historical data depends on the platform. Shopify retains order history for the lifetime of the store. GA4 data depends on your property's retention settings (default is 14 months). Ad platforms vary.
- No raw database access. Presso queries data through each platform's official API. Some metrics or dimensions available in platform-specific reporting may not be accessible through the API.
- Complex custom calculations. While Presso can perform comparisons and basic calculations, highly complex statistical analysis or custom attribution modeling may require dedicated analytics tools.
Cross-Platform
- Data does not automatically reconcile. Revenue in Shopify and revenue in GA4 may differ due to attribution, timing, and definitions. Presso shows you what each platform reports.
- No unified customer identity. A customer in Shopify and a user in GA4 are not automatically linked. Cross-platform customer analysis has inherent limitations.
Getting Help
If you have tried the troubleshooting steps above and are still experiencing issues:
- Check the status page at presso.now for any known service issues
- Start a new Claude Code session to clear any stale MCP connections
- Reconnect the affected data source following the steps above
- Contact support from the contact form. Include:
- A description of the issue
- The question you asked
- Any error messages you received
- Which data source is affected
- When the issue started