Bulk SMS for Insurance Agents: A Six-Step Deliverability Diagnostic

The short answer

If your bulk texts are not landing, the cause is almost never the send button. It is one of six things: unregistered or misregistered 10DLC, message content that trips carrier filtering, sending volume concentrated on one number, identical message bodies at scale, a list full of dead or wrong-type numbers, or sends outside permitted hours that your platform allowed through.

This is the diagnostic order to work through, cheapest fix first.

Start here: what "delivered" actually means

Before diagnosing, get the vocabulary straight, because platforms report these differently and agents conflate them.

If your platform only shows you "sent," you cannot diagnose anything. Delivery tracking per message is table stakes.

The six-step diagnostic

Step 1 — Confirm your 10DLC registration matches reality

Application-to-person messaging on US long codes runs through the 10DLC framework. Your brand and each campaign are registered, and carriers assign throughput and trust based on that registration.

The common failure is not being unregistered — it is being registered under a use case that does not match what you are actually sending. A campaign registered as "customer care" that carries pure marketing content will get filtered, and the filtering will not be labelled helpfully.

Check: does your registered use case describe your real messages? Does your sample message in the registration resemble what you actually send? Is your opt-in description accurate?

Step 2 — Read your own message like a carrier filter would

Filtering is heuristic and unpublished, but the patterns are consistent. Content that raises risk:

The fix is usually smaller than agents expect: identify yourself in the first message, say why you are contacting them, keep it under 160 characters where you can, and include an opt-out instruction.

Step 3 — Look at how volume is distributed

One number carrying an entire day's send is the single most common self-inflicted deliverability wound. Carriers evaluate per-number behaviour, and a number that goes from zero to two thousand messages in an afternoon looks like exactly what it looks like.

Use a number pool with load balancing so volume distributes, and select caller ID by area code or state so the number sending to a Florida lead looks local to Florida. APEX supports both, along with consent-based number sharing across a team hierarchy.

Then add throttling. Sending 2,000 messages over four hours behaves very differently from sending them in four minutes, and the slower send is not meaningfully slower in business terms.

Step 4 — Vary the message body

A thousand byte-identical messages leaving one platform in one window is a strong fingerprint. Spintax — writing a message with alternative phrasings so each recipient gets a variant — breaks that pattern without changing your meaning.

A simple example: {Hi|Hello|Hey} {{first_name}}, this is Maria with {{agency}}. produces three openings from one template. Combine two or three spintax points and you have enough variation to matter.

APEX generates spintax variations as part of its AI copilot, and applies them at send. Write the message; let the variation happen automatically.

Step 5 — Audit the list, not the platform

Sending to a list with 18% landlines and 9% disconnected numbers produces failure rates that look like a deliverability problem and are actually a data problem. Failures also feed back into your number's reputation, so a bad list compounds.

Do this before your next campaign:

Step 6 — Check the clock and the calendar

Texts are subject to the same 8:00 a.m. – 9:00 p.m. window as calls, in the recipient's local time, and several states are stricter — Florida and Oklahoma stop at 8:00 p.m., and Texas does not permit solicitation before noon on Sunday. Texas SB 140, effective 1 September 2025, allows treble damages and mandatory attorney's fees for texting violations.

A platform that enforces this by blocking rather than warning removes the whole category of problem. Every suppressed send should appear in an audit log with a reason.

Building a bulk send that behaves

Once the diagnostics are clean, here is the shape of a send that tends to work.

Segment first. One message to one cohort. A blast to your whole database is a blast that is generic to everyone in it.

Lead with identification. Name, agency, and reason for contact in the first line. This is both a compliance-friendly and a deliverability-friendly habit.

One idea, one ask. A question that can be answered with a word gets more replies than an invitation to a lengthy process.

Throttle it. Spread the send. There is no business reason to compress it.

Give it somewhere to land. Replies come back to a two-way inbox where you can actually answer them from a template, not to a number nobody watches.

Stop on reply. If the send is the first touch of a sequence, the rest of the sequence must halt the moment someone responds.

What to measure

Track four numbers per campaign, and review them weekly:

Reply rate is the one to optimise. Delivery rate is a hygiene metric — you want it high, but a 99% delivery rate on a message nobody answers is a well-delivered failure.

Related reading

FAQ

Why are my texts sending but not being received?

Almost always carrier filtering. The message was accepted and then dropped, which is why your platform shows success. Work through 10DLC registration accuracy first, then message content, then volume concentration on a single number. Filtering is rarely announced, so the diagnosis is by elimination.

How many texts can I send per day?

Throughput depends on your 10DLC registration, your brand trust score, and your number pool size — there is no universal number. The more useful framing is that you should send at a rate your number reputation can sustain, spread across multiple numbers, rather than at the maximum your platform technically permits.

Do I need consent to text an insurance lead?

Yes. Prior express written consent is required for marketing texts to wireless numbers under 47 CFR 64.1200(f)(9), and the number must not be on the Do Not Call registry at the time you send. The FCC's one-to-one consent rule was vacated in January 2025, but that removed a specific constraint on how consent is captured, not the consent requirement itself.

Does spintax actually help deliverability?

It addresses one specific signal — large volumes of byte-identical message bodies — and it does not fix anything else. If your registration is wrong or your list is dirty, spintax will not save the campaign. Treat it as one layer, applied after the fundamentals are right.

Should I use a link in the first message?

Prefer not to. First messages containing a shortened link from a shared domain are among the most commonly filtered. If you need a link, use a branded domain, and consider earning the reply first and sending the link in message two.

What is a good reply rate for insurance SMS?

It varies enough by list source, vertical, and message that any published benchmark is likely to mislead you. Measure your own baseline across a few campaigns and optimise against that. The comparison worth making is your campaign against your campaign, not your campaign against a vendor's marketing number.