Starting September 21, 2026, Shopify improved how Analytics measures sessions. Sessions, conversion rate, and other visit-based metrics may look different. Shopify says this measurement update does not change orders, sales, or customer counts. A ratio moving across that date is therefore a prompt to investigate, not proof of a performance gain or loss. Read the Shopify changelog.
Mark the measurement break
The update aims to focus the default session experience on human visitors where the relevant filter is available, classify bot and system traffic more consistently, and preserve valid shopping journeys without a standard pageview. Each store has its own traffic mix, so the size and direction of any reporting difference will vary. Annotate September 21 on charts comparing periods across the change.
In a weekly report, put sessions, orders, sales, customers, and conversion rate in separate columns. Add significant promotions, stockouts, price changes, and site releases. This makes a measurement shift easier to distinguish from a real business change.
A simple example
Illustrative example: a store records 200 orders in each of two periods. The older report shows 10,000 sessions and the newer one shows 8,000. The calculated conversion rate moves from 2% to 2.5%, even though order count is unchanged. That arithmetic alone does not show that the product page or checkout improved.
The reverse can also happen: a higher measured session count can lower the apparent rate while sales rise. Keep the numerator and denominator visible whenever you discuss conversion. Avoid splicing the two session series into one continuous trend without a note.
Compare like with like
Export periods of equal length with comparable weekdays. Segment by channel and market when acquisition changed. Review orders and sales beside sessions in each segment. To assess a product-page test, add that page's views, cart additions, and purchases; a whole-store rate is too broad to assign an effect to one page edit.
Use Shopify's Analytics documentation for the definitions available in your account. Treat the post-September-21 session series as a new baseline, keep the earlier series dated, and state exactly which metric supports each conclusion.