Your Shopify Conversion Rate Changed on September 21. Your Store Didn't

Somewhere between September 21 and 23, your Shopify sessions number moved. So did your conversion rate, add-to-cart rate, bounce rate and a handful of others. If you happened to look that week, you might have thought a campaign took off or a theme change broke something.

Neither happened. Shopify changed how it counts. Its help center is blunt about it: "A change in these metrics doesn't necessarily mean that your traffic, orders, or customer behavior changed."

This would be a minor footnote in March. Seven weeks before Black Friday it matters a lot, because BFCM planning runs on comparisons, and one side of that comparison just got redefined.

What Shopify changed

The rollout ran September 21 to 23 and made three changes to how sessions are counted:

• Sessions now end after 30 minutes of inactivity. Before, they ended at midnight UTC.
• Some sessions without a pageview now count, such as a shopper going straight to checkout from a cart link.
• Identified bot sessions are filtered out of session reports by default.

Nine metrics can look different as a result: sessions, conversion rate, add-to-cart rate, reached-checkout rate, checkout conversion rate, bounce rate, pageviews per session, online store visitors and searches.

Orders, sales and customer counts didn't change. That's the most useful sentence in the whole update.

Which way did your numbers move?

It depends on which change hit your store hardest.

If you had a lot of bot traffic, removing it shrank your session count and pushed your conversion rate up. If you get a lot of traffic from cart links (SMS campaigns, email "complete your order" buttons, influencer links that drop straight into checkout), those sessions now count, which pushes conversion rate down. The midnight-UTC change mostly affects stores whose shoppers browse late at night in time zones far from UTC. A session that used to split in two at midnight now counts as one.

Shopify says session counts "may increase or decrease depending on which factor dominates." So you can't apply a standard correction. You have to look at your own store.

The year-over-year trap

Here's where BFCM planning gets risky.

Shopify's note says bot filtering applies to sessions from October 7, 2025 onward, so last year's Black Friday traffic has bots filtered too. That helps. But the update doesn't clearly say whether older sessions were recalculated under the new 30-minute rule or the new pageview rule. Shopify tells merchants to "review reports using consistent filter settings" when comparing periods, and to treat post-update data as a new baseline.

My reading: be careful with any comparison of session-based metrics that crosses September 21. If your report says conversion rate is up 0.4 points versus last November, part of that might be the counting change. Orders and revenue are safe to compare. Conversion rate across the change needs an asterisk.

One more open question. AI agents can now browse and check out on Shopify stores through WebMCP tools inside a shopper's browser. Shopify hasn't said whether those sessions count as human, bot or something else. If agent shopping grows this season, it could move your session numbers in ways nobody has documented yet.

October 1: script tags stopped accepting changes

The second change also hits measurement. Since October 1, apps can no longer create or update script tags on your storefront, per Shopify's developer changelog. Existing script tags keep running until March 1, 2027, and then they stop entirely.

Shopify's reason: script tags "run JavaScript directly on a storefront without appearing in the theme editor, which makes them difficult for merchants to find and control." Apps are supposed to move to app embed blocks, or to web pixels for tracking.

What this means for you right now: if an older app on your store tracks conversions or adds widgets through script tags, it can't change that code anymore. If the app's developer needs to update its tracking for BFCM (a new pixel event, a fixed bug), the fix has to come through the new method. An app that hasn't migrated is frozen in place.

You don't need to read code to check. Open Online Store, Themes, Customize, and look under App embeds. Apps that have migrated usually appear there. For any tracking or marketing app that doesn't, email the developer and ask directly: "Do you still use script tags, and when are you moving off them?"

Why clean numbers matter more this year

Adobe's 2026 holiday forecast, published September 28, expects US online holiday sales of $275.1 billion, up 6.7%. It also expects traffic from AI platforms to US retail sites to rise 130% year over year, with the biggest jump on Thanksgiving (up 159%), then Black Friday (up 95%) and Cyber Monday (up 88%).

Shopify is seeing the same trend in its own data. On its Q2 call it said AI-driven traffic and orders to Shopify stores each tripled year over year, per TechCrunch.

So this season you'll be trying to answer new questions, such as how much of the lift came from AI referrals, using session data that was just redefined. If you set a clean baseline now, you can answer them in December. If you don't, you'll be guessing.

What to do before November

1. Take an October snapshot. Write down sessions, conversion rate and add-to-cart rate for October 1 to 31, under the new counting. That's your real pre-BFCM baseline, not last October.
2. Compare BFCM with last year on orders, revenue and average order value. Use conversion rate only against post-September data.
3. Pick one report filter setting and stick with it. If a report offers the "Human or bot session" filter, decide whether you include bots and use the same setting every time.
4. Check your app embeds. List every tracking, review and marketing app, and confirm each one has moved off script tags or has a date for it.
5. Tell your team and your agency. Anyone building a BFCM recap deck needs to know conversion rate changed definition in September, or they'll report a win or a loss that didn't happen.
6. Add an AI referral view. In your reports, segment sessions and orders by referrer so you can see ChatGPT, Perplexity, Copilot and Gemini traffic separately when it peaks on Thanksgiving.

If a report in late November surprises you, check orders and revenue before you believe it. Those didn't change definition.