Quick answer
Omnivery classified 16.6% of Q2 2026 consumer email click events as automated. With newsletter archives included, the figure was 58%, a 41.4-point difference in cohort composition. A recovery email click alone is too weak to mark a shopper as returned or to suppress further help.
Key findings and sources
Adding newsletter archives raised the Q2 bot-click share by 41.4 points
You Left Something’s comparison of Omnivery’s Q2 2026 figures finds 16.6% classified bot clicks in consumer-only traffic and 58% after newsletter archives are included, a 41.4-point cohort-mix difference.
Omnivery, The State of Email Bots 2026
Omnivery’s first-party consumer email click stream and its separately identified newsletter archive traffic; figures are event shares. Subtract the reported shares for the two explicitly defined event cohorts.
The consumer-only bot-click share was about eight times the early figure
Omnivery reported about 2% bot clicks in its consumer email stream through 2023 and 16.6% in Q2 2026. These imply a 14.6-point increase and a ratio of about 8.3, based on an approximate earlier baseline.
Omnivery, The State of Email Bots 2026
Omnivery’s first-party consumer email click stream and its separately identified newsletter archive traffic; figures are event shares. Subtract and divide the reported early and Q2 consumer-only percentages.
The Gmail classification rate changed by roughly 16 points
Omnivery’s consumer Gmail series reported a move from roughly 6% classified bot clicks to about 22% in late February 2026, an increase of roughly 16 percentage points. The source does not establish the cause.
Omnivery, The State of Email Bots 2026
Omnivery’s first-party consumer email click stream and its separately identified newsletter archive traffic; figures are event shares. Subtract the source’s rounded pre-change rate from its rounded post-change rate.
Normalized unclassified-as-bot counts differed by 414 per 1,000 click events
Omnivery’s Q2 shares imply 834 consumer-only events per 1,000 that were not classified as bots, versus 420 per 1,000 events in the archive-inclusive stream. The two groups are normalized separately.
Omnivery, The State of Email Bots 2026
Omnivery’s first-party consumer email click stream and its separately identified newsletter archive traffic; figures are event shares. Take 100 minus each bot share and multiply by 10 to normalize per 1,000 events.
Compare the results
Adding newsletter archives changes the measured bot share
Automated clicks were present in both groups, so a recorded click alone is weak evidence that a shopper returned.
A recorded click does not show who returned
A customer gets a reminder with a link back to the cart. Your email report logs a click, but your store has no new checkout session, payment attempt or reply from that shopper. An automated service could have followed the link. A shopper could also have opened it on a device where the session did not load. Both possibilities matter if your recovery flow sends a different message after a click.
In Q2 2026 Omnivery reported a 16.6% automated share for its consumer click events. Including newsletter archives took the measured share to 58%. That is a 41.4-point change in which events are included. It does not imply that an individual store sees either rate. The figures come from a vendor’s own mixed consumer campaign stream, and checkout reminders are not identified as a separate group.
Read the difference in denominators before reading a trend
Think of two separately normalized groups of 1,000 events. The consumer-only share leaves 834 events that were not classified as automated, and the combined share leaves 420. Those numbers are useful for understanding the scale of the denominator change. They are not 834 and 420 people, or a count of events lost by filtering the same group. A recipient who clicks multiple times still contributes multiple events.
The report also put Gmail’s classified share near 6% before late February 2026 and about 22% after that point. A jump of roughly 16 points in that subset shows why a change in a click report can come from measurement or recipient mix. If your own email click rates rise but verified checkout resumes stay flat, check the relationship at the level of a confirmed session or order before crediting the message.
Keep recovery state tied to what the shopper actually did
Separate the email provider’s click event from a verified return to a store checkout and a paid order. A restored checkout with a matching ID and a subsequent customer action can support a return. A reply that asks for help can route to a conversation. A confirmed paid order should stop that checkout’s reminders, regardless of whether the email provider recorded a click.
Keep people eligible for the normal flow when the only evidence is an unverified click. Review a sample of high-click, low-return sends, compare the sequence across mailbox groups, and confirm that the latest cart, offer and purchase state are read at send time. Evaluate results per eligible shopper and retain the event-level diagnostic counts separately. That lets you fix broken return links without treating automated requests as proof of shopper interest.
Run a simple check before changing the flow: follow a test checkout link and close the page before the cart loads. Repeat with the cart restored, then with a paid test order. The first event should leave the shopper’s return status unresolved, the second should have a matching checkout event, and the third should stop the next reminder. This also helps distinguish an email tracking problem from a cart that opens but has lost the original item or offer.
Study data
| Measure | Value | Context |
|---|---|---|
| Consumer B2C click events classified as bots through 2023 | 2% | Approximate share before the reported rise. |
| Consumer B2C click events classified as bots in Q2 2026 | 16.6% | Consumer-only result, using the exact percentage stated in the report. |
| Click events classified as bots with archives included | 58% | Archive events are added to the consumer stream. |
| Consumer Gmail click events classified as bots before the step change | 6% | Approximate reported level before late February. |
| Consumer Gmail click events classified as bots after the step change | 22% | Approximate sustained level after the step change. |
| Consumer Outlook and Hotmail click events classified as bots | 30% | Approximate provider level. |
| Consumer iCloud click events classified as bots | 12% | Approximate provider level. |
| Newsletter archive click events classified as bots | 90% | Approximate share for archive tools alone. |
| Archive click events attributed to Meta’s crawler | 84% | Approximate share that the vendor attributes to one crawler. |
| Reported archive click volume increase | 19 times | Approximate change in archived-newsletter click volume; exact base volume is not reported. |
| Consumer mailbox provider families covered | 126 providers | Vendor-reported coverage count, not number of senders. |
| Consumer geographic regions covered | 6 regions | Vendor-reported regional coverage, not a representative population sample. |
What changes when you compare the figures
| Comparison | Result | Calculation |
|---|---|---|
| Effect of adding archives to the measured Q2 share | 41.4 percentage points | = 58 - 16.6 |
| Consumer click bot-share change from 2023 to Q2 2026 | 14.6 percentage points | = 16.6 - 2 |
| Ratio of consumer Q2 to earlier bot share | 8.3 times | = 16.6 / 2 |
| Approximate Gmail step change | 16 percentage points | = 22 - 6 |
| Consumer-only events not classified as bots per 1,000 | 834.0 events per 1,000 | = (100 - 16.6) * 10 |
| Combined events not classified as bots per 1,000 | 420 events per 1,000 | = (100 - 58) * 10 |
| Normalized difference in events not classified as bots | 414.0 events per 1,000 | = (58 - 16.6) * 10 |
Methodology and sources
We read the first-party B2C and newsletter-archive findings. The denominator is click events in the stated group, not delivered emails, people or orders. We subtract the two published Q2 shares to get the cohort-mix gap and calculate complements per 1,000 events independently within each group. Gmail’s before/after figures are approximate and do not identify why the change occurred.
Download the source data (CSV)
Download the calculations (CSV)
Related reading
Abandoned cart email statistics and how to read revenue per recipient
Cart recovery SMS benchmarks for conversion and unsubscribes
Email open rate statistics and what Mail Privacy Protection changes
Abandoned cart email inbox placement statistics by mailbox provider
Ecommerce push notification statistics for checkout recovery
Send a Shopify checkout recovery link that opens the right cart
Sources and definitions
Omnivery, The State of Email Bots 2026
The source data includes the population, period, denominator and limitations for every figure.