Delivered
747
of 1,004 attempted
Replied
70
9.4% of delivered
Sent the offer
14
link + discount code
Opted out
31
4.1% of delivered
Delivery funnel
| Delivered to handset | 747 | 74.4% |
| Blocked — already on DND | 127 | 12.6% |
| Blocked — previously opted out | 111 | 11.1% |
| Undelivered — carrier reject | 14 | 1.4% |
| In flight | 5 | 0.5% |
238 were never contacted because they had already opted out before this campaign. The platform blocked every one at the gate.
What the 70 replies said
| Opted out (STOP) | 31 | 44% |
| Needed a human answer | 15 | 21% |
| Declined politely | 11 | 16% |
| Said yes | 8 | 11% |
| Asked for the info | 5 | 7% |
Every opt-out was the bare word “Stop” — no complaints, no negative commentary.
Audience segmentation — two message variants
| Past clients — base message | 797 | 61 replied · 7.7% |
| Realtors & agents — added “it’d do even more for an agent” | 207 | 9 replied · 4.3% |
Segmenting cost nothing — both variants fit in the same 2 SMS segments. Past clients out-engaged agents by 1.8×, the opposite of the expectation going in.
The reply automation
Inbound reply [tag = campaign] [channel = SMS]
│
├─ Not interested → remove tag, send nothing
│ no thanks · not interested · nope · not sure
│
├─ Interested → wait 2 min → MMS w/ link + code → remove tag
│ yes · yea · yep · sure · details · tell me · 👍
│
└─ Anything else → email the loan officer, contact gets nothing
The tag is self-maintaining. All three branches remove it, so it decays to mean exactly “texted, never replied.” Follow-up sends query the tag live — no list is ever hand-built, and nobody who answered can be re-contacted by mistake.
A 2-minute delay is deliberate. An instant reply reads as a bot. The wait is also gated to 8am–9pm so nothing fires during quiet hours.
Reliability finding
The CRM’s native “Customer Replied” trigger silently failed to enrol 6 contacts who replied with matching keywords. No error, no log entry — the reply simply never entered the workflow. Two were caught by the loan officer reading her own phone; four by tooling built the same afternoon.
Mitigation now running: an independent hourly sweep re-reads every recent conversation, finds anyone who asked and did not receive, sends it, and clears their tag. It will not fire inside the 2-minute window, and it stays out of any thread where a human has already replied.
Result: 14 of 14 people who asked for the offer received it. Six of those fourteen only got it because of the backstop.
The number that decides the follow-up
2.2 : 1
For every 1 person who wanted the offer, 2.2 permanently blocked the number.
14 served · 31 opted out.
A 4.1% opt-out rate is high — marketing SMS typically runs 1–3%. And STOP is not campaign-scoped: it is a permanent carrier-level block on the sending number. For a loan officer whose book depends on texting past clients, that is the real cost of the campaign, and it does not appear in any conversion metric.
Open decision: three further touches are scheduled. They target the 677 who did not respond — by definition the least receptive group. Recommendation on the table is to move the remaining touches to email, which carries no permanent channel loss.
All figures read directly from CRM conversation records — carrier-confirmed delivery status, not send-request acknowledgements.
Aggregate data only. No contact names, phone numbers or message content are published on this page.