Skip to main content
SourceMedium AI Analyst

The e-commerce analyst that shows its work.

Ask follow-up questions in Slack using the same governed definitions that power your dashboards. Analytical answers can expose their sources, assumptions, charts, and underlying SQL. Documentation and methodology answers cite their governing sources.

Book a platform walkthrough
AI Analyst

Ask questions about your data

Your AI analyst runs on your data in BigQuery.

This illustrative sample shows an analytical answer with sources, assumptions, a chart, and SQL. Ask follow-up questions in the same Slack thread or pull raw rows as a CSV.

# data-insights
MK
Marketing Lead 2:15 PM

What's our average CAC by channel in the last 30 days?

SourceMedium AI APP 2:15 PM

This 30-day result has two paid sales channels with measurable CAC: marketplace = $77.81 and online store = $399.99. Online store CAC is about 5.1x higher. If your target CAC is below $100, investigate that channel first.

Data: Average CAC by Channel (Last 30 Days)
ChannelAd SpendNew CustomersCAC
marketplace $393,270 5,054 $78
online_store $1,919,529 4,799 $400
Chart: CAC by Channel 
                                  -- Analysis: CAC by channel using executive summary table

-- Assumptions: timeframe=last_30_days | metric=cac=ad_spend/new_customer_count | grain=sm_channel | scope=channels_with_spend
SELECT
  COALESCE(NULLIF(LOWER(TRIM(sm_channel)), ''), '(unknown)') AS sm_channel,
  SUM(ABS(ad_spend)) AS ad_spend,
  SUM(new_customer_count) AS new_customers,
  SAFE_DIVIDE(SUM(ABS(ad_spend)), NULLIF(SUM(new_customer_count), 0)) AS cac
FROM `your_project.sm_transformed_v2.rpt_executive_summary_daily`
WHERE date >= DATE_SUB(CURRENT_DATE(), INTERVAL 30 DAY)
  AND ad_spend IS NOT NULL
  AND new_customer_count IS NOT NULL
GROUP BY 1
HAVING ad_spend > 0
ORDER BY cac ASC
LIMIT 10;

How it is grounded

One governed foundation. Evidence you can inspect.

The AI Analyst uses the same maintained definitions as dashboards and BigQuery, so the answer does not live in a separate black box.

01

Maintained data foundation

Connected commerce data is modeled, tested, and documented before it becomes analytical context.

02

Questions in Slack

Ask a question, review the response, and continue with follow-up questions in the same thread.

03

Evidence you can inspect

Analytical answers can expose the underlying SQL. Documentation and methodology answers cite the governing sources.

Three common ways teams use it

Diagnose a performance change

Move from what changed to the channels, cohorts, products, or customer segments that explain the result.

Review a definition or methodology

Ask how a metric is defined or how a model assigns credit, then inspect the governing documentation.

Take the analysis further

Review the query, export the relevant rows, or adapt the SQL for a deeper investigation in BigQuery.

Useful to the people who ask, verify, and act

Executives & Decision Makers

Decision-ready context without a dashboard scavenger hunt

  • 1
    Compare revenue, CAC, LTV, and margin using governed definitions.
  • 2
    See the sources and assumptions behind each answer.
  • 3
    Share a chart or the underlying SQL when the stakes are high.
Book a platform walkthrough

Ready to stop debating the numbers?

Book a platform walkthrough

In a 30-minute walkthrough, we'll review your current data stack, show how SourceMedium would replace or complement it, and demonstrate workflows relevant to your business.