Monday 24 August to Sunday 30 August 2026. The first full week with everything sending. Nine campaigns went out, 8,696 emails landed, and the result is unusually clear: one offer produced almost every reply and both opportunities, and the rest produced close to nothing. This report shows what worked, what did not, what we got wrong, and what changed on Monday.
That is a reply rate of 0.16 percent across the whole account. For cold email aimed at this market we would expect somewhere between 1 and 3 percent, so this is well below where it needs to be. The useful part is that the failure is not spread evenly. It sits almost entirely in the offers, not in the sending.
Delivery itself was healthy. 79 emails bounced out of 8,696, which is 0.91 percent, comfortably inside the safe range and well under the level where we would pause anything for deliverability.
| Campaign | Offer | Emails Sent | Replies | Opportunities |
|---|---|---|---|---|
| H | Software replacement | 1,281 | 11 | 2 |
| E | Trade classification | 2,226 | 1 | 0 |
| C | Trade classification | 2,038 | 0 | 0 |
| J | Customs Health Review | 1,135 | 0 | 0 |
| I | Document processing | 1,071 | 0 | 0 |
| A | Customs brokers | 285 | 2 | 0 |
| D | Oil and gas | 276 | 0 | 0 |
| G | Export control test | 201 | 0 | 0 |
| B | Forwarders | 183 | 0 | 0 |
Two results deserve to be called out plainly rather than buried in the table.
Campaign J sent 1,135 emails and produced zero replies. Not a low number of replies, zero. Campaign A, which carried the same idea to licensed brokers, sent 285 and produced 2. We have now tested this offer properly and it does not earn a reply from a cold audience.
This is not a comment on the product. It is a comment on the ask. The message tells someone we will look over work they have already done, which reads as a favour they have to make time for rather than something they want. There is also no number in it, so nothing in the message tells them what they stand to gain.
Campaign H asks what a company pays each month for the systems its team runs on, then offers to build a replacement at no charge and run it for about half. 1,281 emails, 11 replies, 2 opportunities. On its own that is a 0.86 percent reply rate, roughly five times the account average, from the smallest budget of attention.
The sample is still small and we are not treating it as proven. It is, however, the only offer that has produced anything, and it is the one Adrian raised on the call as the strongest thing Ripple sells.
Across the four LinkedIn campaigns running on Adrian's and Jonny's accounts, acceptance ran at 28.6 percent, which is a healthy number and in line with what we saw in earlier weeks. Of the 40 conversations that opened, 6 people replied.
Set against email that is a striking contrast. LinkedIn produced 6 replies from 40 conversations. Email produced 14 from 8,696 sends. The audiences overlap heavily, so the difference is the channel and the fact that a LinkedIn message arrives from a person rather than an inbox.
Adrian flagged a booking on Monday from a company that had nothing to do with customs. He was right, and it was our error, not a stray inbound. The person came from our own LinkedIn campaign running on Jonny's account and replied on 24 August.
Two separate checks failed on the same record. Our data had her at a large freight forwarder, which is a real target for Ripple, while her own profile says she now runs a technology company she co-founded. She had moved on and our source had not caught up. Her location is also listed as Hong Kong, against a buyer definition that is United States first.
What changes. Employment is now checked against the person's current profile rather than the record we were given, and any lead outside the agreed countries is dropped before it can be contacted. Both checks run before anyone is loaded into a campaign, not after.
All sending is paused, at your request, and nothing has gone out since. The reason was sound. The replacement emails named systems that are used in the United Kingdom and a large majority of the people receiving them are in the United States, so the question landed as nonsense for most of the list and Adrian was right to stop it.
One offer and one audience instead of nine campaigns pulling in different directions. The software replacement offer, aimed at freight forwarders, which are the only audience that has ever produced a booked call on this engagement. Everything else parked until that has had a fair test with real volume behind it.