Monday 27 July to Sunday 2 August 2026. The biggest sending week of the engagement by a distance: three new segments went live on 27 July, 5,786 emails reached 3,097 new people, and 32 replies came back. One call was booked with Dylan Dickman, CEO of Known, and one CEO wrote back asking for a call outright.
Three new segments launched on Monday 27 July: classifier decision makers (the people who own classification inside an importer or broker), an internal champion angle, and a job signal list built from companies actively hiring for a customs desk. Between them they reached 3,097 new people in seven days, against 547 the week before. That is close to a six-fold step up, and the infrastructure carried it cleanly.
The result at the top of the funnel: 32 replies. The result at the bottom: one booked call with Dylan Dickman, CEO of Known, and one CEO asking for a call in writing without being pushed, Qasim Shahbaz at NTMK Logistics. Both are named with their full exchanges below.
Two things in this report are uncomfortable and are stated plainly rather than buried. First, the calendar invite for Dylan's call did not land in Adrian's diary, which Adrian flagged on 30 July, and there is a section on exactly what changes so it cannot happen again. Second, every one of the 32 replies came from one of the three new segments. The other two produced nothing, and the reason is in the copy rather than the list.
Open tracking stays switched off on all your campaigns, because open-tracking pixels hurt deliverability, so there is no open rate here and there will not be while we send this way. Replies below exclude automatic out-of-office bounces, of which there were 82 on the main segment alone.
| Metric | Result | Read |
|---|---|---|
| Emails sent | 5,786 | Across three new segments plus follow-ups on the July lists |
| New people reached | 3,097 | Up from 547 the week before, see the Week 4 report |
| Replies received | 32 | 1.03% of people reached, all from one segment |
| Calls booked | 1 | Dylan Dickman, CEO of Known, full card below |
| Direct meeting requests not yet booked | 1 | Qasim Shahbaz, CEO of NTMK Logistics |
| Bounces | 47 | 1.52% of people reached, comfortably inside the 2% action line |
This is the most useful table in the report, because it shows the whole week's result sitting inside one of the three lanes.
| Segment | People reached | Replies | Reply rate | Verdict |
|---|---|---|---|---|
| A. Classifier decision makers | 2,687 | 32 | 1.19% | The only lane producing conversations. Sequence now 94% complete, so this is close to its final number. |
| B. Internal champion | 381 | 0 | 0% | Silent. Too small a sample to condemn, but nothing to build on yet. |
| C. Job signal | 29 | 0 | 0% | Barely started. Needs volume before it can be judged at all. |
Cumulative since outbound began: 10,759 emails to 5,919 people, producing 45 replies. Just over two thirds of every reply this engagement has ever produced arrived in the last seven days.
LinkedIn has run out of road. Every campaign on both accounts has finished its sequence, which means no new connection requests went out at all this week. The channel is not broken, it is empty, and it needs a fresh shortlist loaded.
| Metric | This week | To date |
|---|---|---|
| Connection requests sent | 0 | 418 |
| Connections accepted | 2 | 107, a 26% acceptance rate |
| Messages sent | 12 | 152 |
| Replies | 3 | 25 |
A 26% acceptance rate is strong for cold LinkedIn, so the profiles and the targeting are working. The fix is supply, not approach, and reloading it is the second item in the plan below.
3,097 people reached produced 32 replies, which produced 5 live conversations, which produced 1 booked call and 1 direct request for one. Every stage of that funnel is now moving, including the last one, which is the stage that had never moved before this week.
The question that opens segment A, asking how a company gets HTS classification done and whether that is in house or paid per line to a broker, works on the people who buy classification. Sent to the people who sell classification, it reads as an insult, and three brokers said so in blunt terms this week. The list is right, the audience split inside it is not, and the fix is a rewrite of the broker lane rather than a change of targeting.
Five conversations came out of this week that are worth Adrian's time. They are ordered by how close each one is to a call. Every quote below is the prospect's own words, copied from the thread, and every card names the channel it arrived on, the inbox it landed in, the date and time in UK hours, and links straight to the person and their company. All five came in on cold email, in segment A, the classifier decision maker campaign.
Adrian raised this on 30 July, and he noted it had happened before during the podcast work. It is the single most damaging failure this engine can have, because a call that is booked but not in the diary is worse than no call at all.
Dylan was booked on Tuesday 28 July for Thursday 30 July. The booking went through the calendar link, which is the route that writes straight into Adrian's diary and sends a confirmation to both sides, and the notification fired correctly on our side. It still did not show up in Adrian's diary.
So the honest position is that the route we use is the right one and it should have worked, and we do not yet know why it did not. Guessing at a cause and calling it fixed would be worse than saying that plainly. The next step is to reproduce it in front of both of you rather than in isolation.
One segment out of three produced every single reply this week, and it is the segment aimed at whoever actually owns classification inside the business. That is 32 replies from 2,687 people, in a market that ignored the two adjacent angles entirely. The next expansion goes deeper into this lane rather than wider across new ones.
Three replies this week said the same thing in different tones. A compliance manager wrote "We are a Broker. Do your homework." Another wrote "We classify HTS for FREE for over 40 years now. Stop scamming people." To an importer, classification is a cost. To a broker, it is the product they sell. Pitching a broker on doing it cheaper reads as automating away their revenue. The broker lane is being rewritten around the opposite idea: they run it for their clients, their clients pay, and the brokerage keeps the billing.
This week's bounce rate was 1.52%, and the live segment ran at 1.45%. The June and early-July lists ran between 4.5% and 5.2%. Same inboxes, same sending pattern, different data source. The list building changed at the end of July and the delivery numbers moved with it, which protects the sending domains as volume climbs.
Zero connection requests went out across both accounts all week, because every campaign had finished its sequence and nothing new was loaded behind them. The channel itself is healthy, with a 26% acceptance rate to date, so this is purely a supply problem. It is fixed by loading a fresh shortlist, and it is on this week's list.
Five live conversations and one booked call are worth more than another five thousand sends, so the week is sequenced with the conversations first and the volume behind them.
The reply has already gone back from Adrian's inbox offering Tuesday or Wednesday with the booking link, so it sits with Qasim now. If nothing is on the calendar by Tuesday evening, Adrian sends one short line himself. A warm CEO who has already said yes should not be chased by a sequence.
Jordan books a slot on the same link while Adrian and Jonny are both on the call, and Adrian confirms on the spot whether it reaches his diary. Two minutes, with everyone watching, and it either clears the route or reproduces the fault in front of us. Detail is in the section above.
New copy for brokers and licensed customs brokers, built on the idea that they run the classifier for their own clients and bill for it as they do today, with the only free thing being a demo on a couple of lines of their own invoices. Wendy Weeks gets the first reply written that way, and she is the test of whether the frame holds.
A fresh shortlist drawn from the same classifier decision maker profile that produced every reply on email this week, filtered to people who have actually posted in the last thirty days so the connections are not spent on dormant accounts.
A rebuilt sequence is written and staged, anchored on running a sample of the prospect's own lines and showing them both numbers on a fifteen minute call, with Pentagon Freight as the proof. A further 1,274 lookalike companies are researched and ready behind it. Nothing is loaded until the broker rewrite above is finished, so the two do not go out in the wrong order.
Where this leaves us. Five weeks in, the engine is producing conversations with the right people at the right level: two CEOs and a licensed customs broker in one week. The constraint has moved from "will anyone reply" to "does the reply reach a calendar", which is a much better problem and a much faster one to fix. Week 6 is judged on booked calls held, and both Qasim and a re-run at Dylan are the named targets.