Results do not match expectations
The numbers seem wrong
Run these checks before treating the result as a pipeline issue.1
Check the date range in the SQL
“Last week” could mean the previous 7 days or the previous calendar week. The SQL shows which one ran.
2
Verify the metric
“Revenue” can be gross, net, or total. Check what the SQL summed against Metric definitions and Revenue fields.
3
Check data health
Ask “How is my data health?” If a table you depend on is stale, recent days are missing from the result.
4
Check filters
Confirm the query used the same filters as the report you are comparing against: channel, store, order validity, customer type, subscription status, product, and date basis.
5
Compare complete days only
Leave today out. Use a finished date range that should be fresh according to Data freshness.
How to investigate a discrepancy
When SourceMedium does not match Shopify, an ad platform, Amazon, or Recharge, start with Why would external reports not match the SourceMedium dashboard?. Then pin down the exact comparison.Documentation explains definitions and expected freshness. If a discrepancy survives the checks above, it usually needs account-specific investigation by SourceMedium support.
High (direct) / (none) in results
This is a tracking issue, not a query issue. The analyst is reporting what is in your data. Ask “How is my attribution health?” and see Data and attribution health.The analyst misunderstood the question
It answered a different question
Rephrase with more detail.It used the wrong table
Name the domain.It ran an analysis when I wanted rows
Use pull or export language and list the fields: “Pull order ID, order date, UTM source, and revenue for last week.” Naming dimensions alone is not enough. See Asking questions.No results returned
”No data found” or empty results
- Check the date range. The period may have no activity.
- Check filters. A filter may have excluded every row.
- Check data health. The table may be empty for your account.
The question timed out
A single question never runs longer than 7 minutes. Large date ranges and row-level detail take longest. Try:- A narrower date range
- An aggregate instead of row-level detail
- Splitting the question into smaller parts
Chart not generated
No chart appears when:- The result is a single number
- You asked for a raw data pull or export
- The data does not suit a chart
- Chart generation failed
Access or installation denied
”This workspace is not authorized”
Your Slack workspace or your email does not match the workspace configured for your organization. Contact your SourceMedium team.”Installation denied” or “access denied”
Only approved workspaces and users can use the analyst. Ask your SourceMedium admin to add your domain or exact email.Refused in a shared channel or group DM
The conversation probably does not share a store, or someone still needs the app installed. See Shared Slack and Slack Connect. For a connected assistant that cannot reach SourceMedium, see AI assistant troubleshooting.Slow or queued responses
Response time depends on query complexity, table size, and current load. Deep Analysis runs two analyses, so it takes longer than a single question. A single question never runs longer than 7 minutes.My question says it is queued
A thread runs one analysis at a time so follow-up context stays consistent, and holds one more question in a queue. If you ask while both slots are taken, your message is not rejected. The analyst tells you it added your follow-up to the next response and answers it in order.I hit the concurrent analysis limit
Each user can have 2 analyses running at the same time, across all threads. The message lists the threads that are still running. Wait for one to finish before starting another.The analyst said it cannot help
The analyst declines requests that:- Ask about data outside your SourceMedium warehouse
- Ask it to take an action, such as changing data or sending an email
- Are unrelated to analytics
Common questions
Can I ask follow-up questions?
Can I ask follow-up questions?
Yes, in the same Slack thread. In a DM, type normally. In a channel, mention
@SourceMedium in the reply. See Threads and follow-ups.Why can I not switch stores in the same thread?
Why can I not switch stores in the same thread?
Once a store is confirmed, the thread stays on it. That prevents mixed-store analysis. Start a new thread to switch. Your available stores are listed in the Data Availability Report. See Store scope.
Why did my typed reply confirm the store?
Why did my typed reply confirm the store?
When a thread is waiting for store confirmation, any reply counts as confirmation. The question you already asked then runs on its own.
Why did the bot ignore my channel reply?
Why did the bot ignore my channel reply?
In channels, a reply without
@SourceMedium does not run. Reply again with the mention.Why did it ask me to run an analysis first?
Why did it ask me to run an analysis first?
Summaries and recommendations use the thread’s findings. If the thread has no analysis yet, run one, then ask again.
Why did it miss an older detail from this thread?
Why did it miss an older detail from this thread?
Thread context is condensed as the thread grows. Restate the detail in your next prompt.
How do I reset context?
How do I reset context?
Start a new thread.
How often does data refresh?
How often does data refresh?
Most tables refresh daily. Data is complete through yesterday in your reporting timezone. Today’s data is not guaranteed. See Data and attribution health.
Can I export the results?
Can I export the results?
Yes. Data answers attach
query.sql, results.csv, and the chart image if one was made. Data pulls attach results.csv automatically.Can several people use it at once?
Can several people use it at once?
Yes. Each user is capped at 2 running analyses. Each thread runs one analysis at a time and queues the next.
Why do I see different results than my teammate?
Why do I see different results than my teammate?
Everyone queries the same data. Check whether you used the same time range and the same metric, and whether data refreshed between the two questions.
Getting more help
- Read the SQL. It shows how the analyst interpreted your question.
- Ask a simpler question. Start with one metric and one range, then add detail.
- Ask about schema. “Which table has order data?” explains the data structure.
- Contact support. Open a support request for persistent issues.