Nobody Logs a Missed Text

Three people answered him, and none said don’t do it. They said it works, then each described the same kind of ending. What he decided is not recorded; his thread went up the same day this did.

One runs that exact setup for several customers. “There are companies out there that do allow texting on the main line,” he wrote. “It’s not that there’s anything inherently ‘no’ about it, more that their system isn’t really built for it.” His customers, he says, “tend to forget it’s setup this way, through internal employee churn or otherwise.”

A second warned that the cell “dies or reboots without you knowing.” The third worked through forwarding modes and caller ID pass-through, then finished on this: write the setup down and “retest every few months, because forwarding settings get lost when a phone is replaced or reset.”

None of them described a failure that announces itself.

One channel has instruments. The other has a person.

Once the calls move to a hosted system, they generate records as a byproduct of switching: which number rang, whether anyone picked up, what time, where it went after hours. Nobody asks for it; that is what the machine does while it works. The texts, under his plan, land in a messaging app on a handset, readable by one person on one device.

We are describing what each system records, not measuring how often either fails. The difference decides what he can find out later, and you cannot run a report on a gap. He can tell you his missed-call count for last month. He cannot tell you how many photos of broken equipment arrived while that phone was in a truck, on a charger, or in the pocket of somebody who had stepped out. Neither can we.

There is a fair reading in which this costs him little. A customer who texts a contractor and hears nothing often just calls, which puts the request back onto the channel that keeps records, and all he lost was the text. If that is his situation, he has a tidiness problem rather than a revenue problem, and we will not pretend to know which he has. Here is what we will put a stake in: he cannot tell either, and he is about to design his intake around the channel he cannot inspect.

Whether the main line can text is a purchasing question

Both answers he got were true at once. His experience that the main line can’t text is real for the products he has used, and the commenter telling him it can is right too, because this turns on the product and the paperwork rather than anything physical.

Two facts set the shape. A number can be text-enabled while its voice stays somewhere else: Bandwidth defines hosted SMS as “a service that allows Bandwidth customers to enable non-Bandwidth owned/operated phone numbers for messaging,” notes those numbers “will not route any voice traffic over the Bandwidth network,” and requires an approved letter of authorization. And once the texts come off a thumb and onto a platform, the sender has to be registered: Twilio’s compliance documentation says “anyone sending SMS/MMS messages over a 10DLC number from an application to the US must register for A2P 10DLC,” as a brand and a campaign, and that senders who skip it “will receive additional carrier fees” on top of heavier filtering. Those fees are published rather than mysterious: Twilio’s US pricing page says “US A2P 10DLC are subject to registration onboarding fees.”

So it is a filing, a provider who supports it, and a signature, and the whole thing collapses into one question for a salesperson: can my main number send and receive MMS, with more than one person signed in, and who holds the registration when I leave?

Moving the texts onto the phone system is not a free win

Read the same subreddit a few days earlier, where a reseller laid out six months of problems with the softphone app his customers use exclusively: calls that can’t be held, voicemail unreachable in the app, and the part that lands here: “SMS is very unreliable, sometimes texts don’t send or come through properly; MMS is even more unreliable and attachments fail to send or be received.” His tickets are open, and he is weighing moving his larger customers elsewhere.

Set that next to the first thread, where the photo of the broken thing is the work request. A text channel that quietly drops MMS has failed at the specific job the side of the truck advertised. Putting texting on the phone system does not end that exposure; it changes who you call about it. One reseller’s ticket queue cannot tell you how MMS behaves generally, and neither can we. It does tell you what to watch, which is why the drill below sends a photo rather than a word.

What ownership looks like at this size

Start with the record. Can a second named person, who is not the owner, read the last month of customer texts without borrowing somebody’s handset? If not, that channel belongs to a device and an employee, whatever name is on the number. Get the number into the company’s name with the authorization on file, and get the inbound thread somewhere that has logins rather than a SIM: the phone system’s own web or desktop client with several people signed in, or your help desk receiving the thread as a ticket.

Then put the drill on the calendar instead of in a note. Text the main number from a number nobody at the company recognizes, after hours, with a photo attached, and see how long it takes a named person who is not the owner to answer. Repeat it after any handset replacement, any carrier change, any departure. Tying it to those events is the version of that advice which survives the person who wrote the note.

If one person handles the texts, the volume is low, and that person is reliable, the cellphone forward is a defensible setup, and the honest spend is documenting it and running the drill. No project required, and no outside help either.

Ask the unglamorous question before the AI question

The AI being sold into this market answers the phone: a voice agent for inbound calls, after-hours triage, an assistant that books the appointment. Some of it genuinely works. Notice what it attaches to, though. An answering layer needs a queue, a log and an escalation path to plug into, which is exactly what the call side already has and the text side, in his plan, does not.

So ask the question that has nothing to do with AI first. Which channel does the work actually arrive on, and can you show that from a record, or only from memory? Automating the part of your intake that was already counted buys faster answers to the calls you were already answering.

Working out which of your channels carries the revenue, and which you can actually see, is where our coaching and project work starts. When the answer is that your setup is fine and somebody just needs to write it down, that is what we will tell you.