AI search news · ·
0 of 30 cold emails to addresses verified before the send have bounced at 3 to 4 days, against 5 of 62 sent before the gate; 4 of those 5 bounced within an hour of the send
At the 2026-09-13 read, 0 of the 30 cold first emails sent to addresses that passed a mailbox-verification read before the send had bounced, at ages of 71 to 96 hours. The 62 first emails sent before the gate went live bounced 5 times, 8.1%. 4 of those 5 bounces were recorded within an hour of the send and the fifth at 24 hours, so every gated row is past the window in which every earlier bounce arrived. At the pre-gate rate the 30 rows would carry 2.4 bounces. The gate cost 10 plan-included credits across 40 reads and removed 5 addresses that read invalid before any email went out. The read that scores the gate is dated 2026-09-15 and its rule was written on 2026-09-08; this is the interim count.
Method, stated before the numbers
Two populations, one campaign. The first email of a cold sequence went to 92 named people at US and UK web agencies between 2026-09-02 and 2026-09-10 (132 sends in all, 40 of them follow-ups). An address is in the gated population when its load record carries a mailbox-verification reading taken before the load: from 2026-09-08 each address was submitted to Instantly's email-verification endpoint, an invalid result was dropped by name, a verified result was loaded, and a verified result on a catch-all domain was loaded with the MX read green and the status written down. 30 gated addresses had a first email by the read (5 more are loaded and unsent because the campaign paused on 2026-09-11). The ungated population is every address whose first email predates the gate; those 62 were verified on 2026-09-08, after their sends, in the post that set the gate.
A bounce is a lead the sending platform moved to its bounced status, read live or from the record we keep when a bounced lead is deleted; the campaign's bulk counter reports the same 5. Bounce timing is the bounced-status timestamp against the first-email timestamp from the sent list. Age at read is the first-email timestamp against the read time, 2026-09-13 13:56 UTC. Verification classes: verified and not catch-all, verified on a catch-all domain, invalid. Gate cost is one read per address the gate touched, loaded or dropped, at 0.25 credits each from the plan's included allowance. The computation is the committed script `content/news/gate_four_day_cut.mjs`; it writes `content/news/gate-four-day-cut-2026-09-13.json` with counts, days and hours and no addresses, and 15 of 15 self-checks pass, among them that the two populations sum to the campaign's first-email addresses, that the bounce count equals the bulk counter, and that no gated row reads invalid.
The two populations at the read
| Population | First emails | Verified, not catch-all | Verified, catch-all domain | Invalid | Bounced | Bounce rate | Age at read |
|---|---|---|---|---|---|---|---|
| Verified before the send (gated) | 30 | 18 (60.0%) | 12 (40.0%) | 0 (dropped before load) | 0 | 0.0% | 71 to 96 h |
| Sent before the gate (ungated) | 62 | 40 (64.5%) | 13 (21.0%) | 9 (14.5%) | 5 | 8.1% | 6 to 11 d |
The gated population carries a larger catch-all share, 40.0% against 21.0%, because the gate loads a catch-all reading on the MX read alone and the sourcing runs of 2026-09-08 to 2026-09-10 drew more Google Workspace domains. In the ungated set the 5 bounces fell 2 on verified-not-catch-all, 2 on catch-all and 1 on invalid; the gated set has 0 in every class.
When the pre-gate bounces arrived
| Bounce | Hours after the first email |
|---|---|
| 1 | 0 min |
| 2 | 0 min |
| 3 | 0 min |
| 4 | 10 min |
| 5 | 24.3 h |
4 of 5 were recorded within an hour of the send and the last at 24.3 hours. Every gated row was at least 71 hours old at the read, 3 times the longest bounce delay the campaign has produced. That is the reason the interim count is worth stating at 30 rows: a bounce that was going to arrive on the pattern this campaign has shown would have arrived. It is not a claim about bounces that take days, which this campaign has not produced and cannot rule out.
By send day
| First-email day | Population | First emails | Bounced | Bounce rate |
|---|---|---|---|---|
| 2026-09-02 | ungated | 10 | 0 | 0.0% |
| 2026-09-03 | ungated | 20 | 0 | 0.0% |
| 2026-09-04 | ungated | 13 | 3 | 23.1% |
| 2026-09-07 | ungated | 19 | 2 | 10.5% |
| 2026-09-09 | gated | 10 | 0 | 0.0% |
| 2026-09-10 | gated | 20 | 0 | 0.0% |
What the gate cost
| Item | Count |
|---|---|
| Addresses read at the gate (loaded or dropped) | 40 |
| Credits, plan-included, 0.25 per read | 10 |
| Dropped as invalid before any email | 5 (3 on 2026-09-09, 2 on 2026-09-10) |
| Loaded | 35 |
| Loaded and sent by the read | 30 |
| Loaded and unsent (campaign paused 2026-09-11) | 5 |
| The 2026-09-08 read on the ungated addresses, credits | 15.5 |
5 of 40 addresses read at the gate were dropped, 12.5%, against 9 of 62 (14.5%) reading invalid on the ungated set after the fact. Each dropped address had already passed the named-person, source-page and MX gates; the verification read was the only gate that removed it. Nothing was bought; the credits are the plan's included allowance.
What this read does not show
30 rows is a small cell. At the pre-gate bounce rate the cell would carry 2.4 bounces, so 0 is below expectation but not far below it, and a single bounce on the next gated cohort moves the rate to a number this post cannot predict. The two populations were sourced on different days from different towns and the gated one has a larger Google Workspace share, so the comparison is between cohorts, not a controlled split. The campaign is paused with 52 leads pending and 40 complete; no reply and no report load has been recorded on either population, which is the campaign's own record and not this post's subject.
The pre-registered read, 2026-09-15, restated
- 1(A) Cold rows loaded from 2026-09-08 under the gate bounce at or below 8.1%, the rate of the ungated population, on the 2026-09-15 read, with at least 20 such rows having a first email. 30 rows qualify today; the interim count is 0.
- 2(B) The 9 addresses that read invalid on 2026-09-08, re-submitted on 2026-09-15, read invalid again in at least 7 of 9.
- 3(C) The 4 bounced addresses that read verified on 2026-09-08, re-submitted on 2026-09-15, read verified again in at least 3 of 4.
The rule was written on 2026-09-08 and is not amended by today's count. Any of the three outside its statement on 2026-09-15 is published as a falsification with the same script and a dated artifact.
Does verifying an email address before sending stop bounces?
On this campaign, 0 of 30 first emails to addresses verified before the send had bounced at 71 to 96 hours, against 5 of 62 sent without the gate. The scoring read is 2026-09-15; the cell is small and the two groups were sourced on different days.
How quickly does a cold email bounce?
On this campaign 4 of 5 bounces were recorded within an hour of the send and the last at 24 hours. No bounce arrived later than that on 92 first emails.
What does a mailbox-verification gate cost?
40 reads at 0.25 plan-included credits each, 10 credits, to load 35 addresses and drop 5. The reads take about 75 seconds per batch before the first email is written.
Artifact
`content/news/gate-four-day-cut-2026-09-13.json` holds both populations' counts, classes, bounce timing in hours, age at read, the send-day tables, the gate cost lines and one row per first email with its population, send day, class, bounce flag and hours to bounce (no addresses), with 15 self-checks (0 failed). It is produced by `content/news/gate_four_day_cut.mjs` from the sending platform's sent list, lead list and bulk counters, the preserved bounced-lead records, the 2026-09-08 verification ledger and the per-load verification readings; the per-address ledgers stay in the repository's private prospects directory.
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.
Who runs this
- Built and operated by Sensara LLC, Atlanta, Georgia — about us and how the audit works.
- See what the report looks like before you run anything — score per engine, the competitors AI names instead of you, and a fix plan.
- We run the same audit on ourselves every week and publish the result: in the latest run AI named AskedAbout in 1 of 144 answers. We report our own numbers the way we report yours.