AI search news ·

Cold pitch emails by touch: 107 emails to 66 named people at AI-cited outlets, 23 openers, 3 replies, 0 tracked clicks, and every reply came from an opener within 2.1 hours

At 2026-09-06 14:11 UTC the Instantly API reads 107 pitch emails to 66 named people at outlets AI engines cite, sent over 55 days: 23 unique openers (35%), 3 replies (5%), 0 tracked link clicks. Touch 1 went to all 66; touch 2, one follow-up in the same thread, went to 41. Openers by touch: 19 of 60 measurable first emails (32%) and 13 of 41 follow-ups (32%). Replies: 1 after a first email, 2 after a follow-up; all 3 came from a person who had opened the email they answered, 0, 2, 2.1 hours after it. Method: Instantly API v2 campaign, analytics, sent-row, received-row and lead-row endpoints plus the 11 preserved touch-1 records, attributed one closed touch per preserved lead id (pitch-touch-cut-2026-09-06.json, 14 of 14 self-checks).

What was sent

One campaign, two sending inboxes on our own domain, one named recipient per outlet. The first email names the engine and the question on which the outlet's page was cited in our weekly panel, and offers the underlying rows; the follow-up is one reply in the same thread at least five days later, and nothing after it. Sends ran from 2026-07-13 to 2026-09-06 under a daily cap of 4. Link tracking and open tracking are on on this campaign, so an open is a pixel fetch and a click is a rewritten-link hit. The tool counts a follow-up as a new lead record (the record is deleted and re-added), which is why its analytics row reads 99 contacted and 37 unique openers where the address-level table below reads 66 and 23.

TouchEmailsRecords with a readable open countOpensUnique openersRepliesRepliers who opened that emailTracked clicks
1 (first email)66603119 (32%)110
2 (follow-up)41412313 (32%)220
Both1071015423 people (35% of 66)330

The follow-up opened at the same rate as the first email (32% against 32%) and produced 2 of the 3 replies from 41 sends, where the first email produced 1 from 66. Among the 36 people whose two emails both have a readable open count, 9 opened both, 6 opened only the first, 4 opened only the follow-up, 17 opened neither.

Opens per record, and where the replies sit

Opens on one email recordRecords
069
120
28
3 or more4
Readable records101

69 of 101 readable records (68%) were never opened. The 3 replies came 0 h, 2 h, 2.1 h after the email they answered, each from a record with at least one open, and none of the 66 people fetched a tracked link at any point: 0 clicks on 107 emails. Two of the three replies answered a follow-up, one a first email. The pattern this campaign has produced so far is that a reply is an immediate act by someone who opened the email and did not click through; the offered rows were asked for in the reply, not fetched from the link.

Sends by week

ISO week starting (UTC)First emailsFollow-ups
2026-07-1320
2026-07-20130
2026-07-2745
2026-08-03218
2026-08-1014
2026-08-17118
2026-08-2467
2026-08-3189

Pre-registered read, 2026-09-13

  1. 1Eight follow-ups went to the 2026-08-31 cohort on 2026-09-05 and 2026-09-06 (the last four at 14:05 UTC today). The follow-up reply rate to date is 2 of 33 before them. 1 or more replies among the 8 keeps the follow-up reading; 0 of 8 is reported as the follow-up rate falling to 2 of 41.
  2. 2Tracked clicks stay at 0 on every email sent, or the first click is reported with its touch and its hours-after-send.
  3. 3Any reply arrives within 24 hours of the email it answers, as all 3 have (maximum 2.1 h). A reply later than 24 hours falsifies the immediate-reply reading above.

Method

Five Instantly API v2 reads at 2026-09-06 14:11 UTC: the campaign object (status, tracking flags, daily limit), the analytics row, every sent row with its timestamp and subject, every received row, and every live lead row with its open, click and reply counts; 5 sent rows that are our own manual in-thread replies are excluded from the send count. A follow-up is done by preserving the lead record to a dated file, deleting it and re-adding it, so a first email's opens survive only in the preserved file; 11 preserved files were read, 0 skipped, and each preserved lead id that is no longer the live record is one closed touch, latest frozen count winning. Touch number is the order of sends to one address. A reply is attributed to the last email sent before it. 5 July leads received their follow-up as a sequence step on the same record, so their first email has no separate open count and is excluded from the touch-1 open denominator; 1 record was deleted after a bounce and has no open count at all. 14 of 14 self-checks pass, including sent rows equal to the analytics send count, replies equal to the analytics unique-reply count, clicks 0 on both reads, per-touch opens within 2 of the analytics total, every follow-up since 2026-08-11 sent as a reply thread, and no identifier in the artifact.

Limits

How many people opened, replied and clicked?

23 of 66 opened at least one email, 3 replied, 0 clicked a tracked link, across 107 emails over 55 days.

Did the follow-up help?

The follow-up produced 2 of the 3 replies from 41 sends and opened at 32%, the same rate as the first email; 4 people opened only the follow-up.

Why do replies come from openers who never click?

All 3 repliers had opened the email they answered and replied within 2.1 hours; 0 of 66 people fetched the tracked link. The offered data was requested by reply rather than fetched.

What is checked on 2026-09-13?

Replies among the 8 follow-ups sent on 2026-09-05 and 2026-09-06, whether tracked clicks are still 0, and whether any reply arrived more than 24 hours after its email.

See your number

See which businesses AI names when your client's buyers ask.

Running this for clients? The $249 agency 5-pack audits five businesses, white-labeled.

Check a client's AI visibility

Begin your check

Free · 60 sec

No account · No card · 3 buyer questions, 2 engines

By running a check you agree to our Terms and Privacy Policy.

Who runs this