It removes sessions Shopify has identified as automated, and only those. Since the September 2026 update, the filter is on by default in session reports.
Your orders don't change: the filter only affects metrics calculated on sessions. A bot has never bought anything, but it has dragged down a lot of rates.
The benchmark
Shopify classifies each session as "human" or "bot" based on everything that happens in it, and the filter has existed since October 27, 2025.
In reports that offer the "Human or bot session" filter, you can show bots again. Search and product recommendation reports remove them automatically and don't give you the choice.
To find your share of automated traffic, take a period after the update and compare the same report with and without the filter.
What the overall number doesn't tell you is where bots land. At a fashion brand audited this spring, desktop conversion had fallen from 1.95% to 0.33% in a few months, while mobile barely moved by a tenth of a point.
Orders, meanwhile, were stable. All the excess traffic came in on a single device and through a single channel, direct.
The first hypothesis was an influx of unfiltered bots, which run mostly on desktop. Bot traffic isn't spread evenly: it concentrates in one segment, and that's the segment to read.
The rule
Always present your numbers in human sessions. Keep the view with bots for one use only: diagnosis, when a traffic spike makes no sense.
The steps
- Bot share. Measure it over four recent weeks.
- Breakdown. Look at it by device, source and country: bots aren't spread at random.
- The logbook. Write this number down in it. It will explain future gaps.
- Other tools. Check that GA4 and ad platforms aren't telling a different story.
- The line for leadership:
"Our total traffic is down, our human sessions are stable, and the filter is what changed."
The catch
The key word is "identified." A bot that Shopify doesn't recognize stays in your human sessions, and Shopify admits its classification is cautious: it would rather miss a bot than label a real person as one.
A traffic spike with no add-to-carts is always worth a look. And the filter changes nothing on your ad bill: a bot's click on an ad is still a paid click.
