Skip to main content
Most data in your dashboard and warehouse is complete through the previous day. Yesterday and earlier are the days you can rely on. The current day may already have some rows, but those rows are rarely complete. Do not use the current day for a decision or a reconciliation. Changes made to your Configuration Sheet are picked up during regular refreshes. The visible effect in a dashboard or warehouse table depends on the next refresh for the report or table using that configuration. If a configuration change is still not reflected after the expected refresh window, contact your SourceMedium team with the sheet tab, rule, date range, and dashboard or table where you expected to see the change.
  • Amazon Seller Central. Recent days settle over time. Orders count in dashboards once they settle. Until then, the warehouse order tables hold the last 3 days of orders from Amazon’s live feed as pending rows (is_pending), with no revenue. Gross revenue is generally stable enough for an operating read after about 4 to 7 days. Net revenue, refunds, taxes, discounts, and customer-level metrics settle later. See Amazon Seller Central: Data nuances and limitations.
  • Amazon Vendor Central. That Seller Central order curve does not describe Vendor Central. Connection steps are on the Amazon Vendor Central integration.
  • Amazon Ads. Spend stabilizes in about 4 days. Attributed conversions and ROAS need about 15 to 22 days. See Amazon Ads: Data nuances and limitations.
  • Amazon DSP. The Seller Central and Amazon Ads windows above do not describe DSP. Connection steps are on the Amazon DSP integration.

​
What this means for reconciliation

Freshness is only one reason reports differ. For the full cross-tool checklist, see Why would external reports not match the SourceMedium dashboard?. On the freshness side specifically, when comparing SourceMedium to Shopify, an ad platform, Amazon, Recharge, GA4, or another external report:
  • Use completed dates only. Do not use the current day as the basis for a discrepancy.
  • Compare the same date basis. Order date, refund date, click date, conversion date, and platform reporting date can all produce different totals.
  • Check whether the source has a longer expected lag before escalating.
  • Re-run the comparison after the next expected refresh if the missing data is recent.
  • If the source data is fresh but the result still differs, check the metric definition and filters rather than treating it as a freshness issue.

​
Scheduling downstream jobs

SourceMedium publishes a freshness SLA, not a fixed customer-wide wall-clock refresh time for every table. If you run scheduled queries or downstream exports from your warehouse:
  • Schedule them after the normal overnight refresh window for your reporting timezone.
  • Build them to tolerate a missed or delayed upstream refresh.
  • Check Data Health before using the output for a time-sensitive report.
  • Avoid scheduling against current-day data when the job depends on complete results.
  • Contact SourceMedium if you need a table-specific refresh expectation for a critical workflow.
For table-specific freshness questions, use Data Health to confirm the latest available data, then align your downstream schedule to complete data rather than a guessed exact time.