[All articles](https://missblue.dev/blog) Miss Blue field notes # How to respond to a text message Most business replies fail for the same three reasons: they answer a question nobody asked, they arrive too late to matter, or they end without a next step. This is the structure that fixes all three, and the honest limits of letting software write it for you. Published August 27, 2026 15 minute read In this guide [Decide what the reply is for](#section-1) [Read the intent before you write](#section-2) [A reply structure that works](#section-3) [Reply patterns for common messages](#section-4) [How fast is fast enough](#section-5) [What an AI text message response generator is good at](#section-6) [What an automated responder should not do](#section-7) [Draft with software, publish as a person](#section-8) [Replying on a blue bubble](#section-9) [Responding to an unhappy message](#section-10) [Responding to a stop request](#section-11) [When not to reply at all](#section-12) [Measure reply quality, not reply volume](#section-13) Build with Miss Blue **Turn the next message into a real reply.** Get a blue line for your product, agent, or team. Use the Message Center today and connect the API anytime. [Create your account](https://missblue.dev/signup) [Explore the API](https://missblue.dev/imessage-api) Build with Miss Blue **Turn the next message into a real reply.** Get a blue line for your product, agent, or team. Use the Message Center today and connect the API anytime. [Create your account](https://missblue.dev/signup) [Explore the API](https://missblue.dev/imessage-api) 01 ## Decide what the reply is for Before you write a word, name the outcome. A reply usually does one of four things: it answers a question, it confirms something, it asks for the one piece of information you are missing, or it hands the conversation to a person who can finish it. A message that tries to do all four does none of them. The most common failure is a reply that restates the situation without moving it forward. The customer already knows what they asked. What they do not know is what happens next, who is doing it, and when. 02 ## Read the intent before you write Text messages are short, so intent is easy to misread. Sort the incoming message into a category first, and let the category pick the shape of your reply. Most inbound business texts fall into a small number of buckets. If a message could belong to two buckets, treat the higher-stakes one as correct. A message that might be a complaint is a complaint until proven otherwise. - A question with a factual answer - A scheduling or logistics request - A price or eligibility question - A complaint or a problem report - A yes or no to something you asked - An opt-out or a request to stop - Ambiguous — needs one clarifying question 03 ## A reply structure that works The reliable shape is: acknowledge, answer, next step. Acknowledgement is one clause, not a sentence of gratitude. The answer is the specific thing they asked for, stated plainly. The next step is a single, concrete action with an owner and a time. Keep it to the length of the screen it will be read on. Two to four short sentences is usually right. If the answer genuinely needs more room, say so and offer a call, a document, or a link rather than sending five paragraphs into a thread. - Acknowledge in a clause, not a paragraph - Answer the question that was actually asked - One next step, with a named owner and a time - Say who you are if the thread is new - Ask at most one question per message 04 ## Reply patterns for common messages Templates are useful as a starting shape, not as a finished message. Keep the skeleton and replace every generic noun with a specific one — the actual date, the actual price, the actual name of the person who will call. A template that ships unedited is how a business ends up sending “Hi {first\_name}” to a real customer. Treat the variable slots as a checklist, not as automation. - Question: “Yes — \[answer\]. Want me to \[next step\] today?” - Scheduling: “I have Thursday at 10 or Friday at 2. Which works?” - Price: “It is \[figure\] for \[scope\]. Happy to break that down if useful.” - Missing information: “To get you an exact answer I need \[one thing\].” - Delay: “Still working on this. I will have an answer by \[time\].” - Complaint: “That should not have happened. Here is what I am doing about it: \[action\], by \[time\].” - Stop request: suppress immediately, send nothing else. 05 ## How fast is fast enough Speed matters more in messaging than in email because the medium sets the expectation. A person who texts a business is usually mid-task and expects the same rhythm they get from a friend. A reply within a few minutes reads as attentive; a reply the next morning reads as an out-of-office notice. Rather than promising an unrealistic response time, publish one you can actually keep and then keep it. A reliable “we answer within two hours during business hours” beats an aspirational “instant” that fails a third of the time. Where you cannot staff the gap, an acknowledgement that names a real time is far better than silence. - Set a target you can staff, and measure against it - Acknowledge fast even when the answer is slow - Never promise a time you cannot hold - Route after-hours messages to a queue somebody actually opens 06 ## What an AI text message response generator is good at Automation earns its place on the parts of the reply that are mechanical: classifying intent, pulling the order status or appointment slot out of a system of record, drafting the first version, and catching the messages a human queue would have missed at 11pm. It is also good at consistency. A generated draft will reliably include the account reference, the correct hours, and the opt-out language that a tired person at the end of a shift forgets. - Triage and routing by intent - Looking up the fact the answer depends on - Drafting a first version for a person to approve - Answering a narrow, well-defined question set - Acknowledging out-of-hours messages with a real time 07 ## What an automated responder should not do The failure modes are predictable. An automated responder should not guess at facts it cannot look up, should not negotiate price or terms, should not handle a complaint that has escalated, and should not keep replying after a person has asked for a human. It also should not pretend to be a person. Manufactured familiarity — a fake first name, an invented anecdote, a “sorry, just saw this!” from a system that never sleeps — is the fastest way to lose a customer who later works out what happened. - No invented facts, prices, or commitments - No continuing after a stop or a handoff request - No pretending automation is a named human - No more than a bounded number of turns before escalation - No sending at all when confidence is low — queue it instead 08 ## Draft with software, publish as a person The workable arrangement in most businesses is draft-and-approve. The system proposes a reply with the relevant record attached; a person reads it, edits the one clause that is wrong, and sends. Approval takes seconds and removes almost all of the risk. Move to unattended sending only for the intents you have measured: a specific question, a specific answer shape, a specific accuracy rate over a real sample. Expand the unattended set one intent at a time, and keep the ability to switch it back off in a single click. 09 ## Replying on a blue bubble On iMessage the reply lands in the same thread as messages from the customer’s family. That familiarity raises the standard rather than lowering it. Typing indicators and read receipts also make your timing visible in a way email never does. Practically: keep messages short, avoid multi-message bursts, do not send a follow-up before they have had a chance to read the first one, and make sure the name and identity on the line match the business the customer thinks they are talking to. 10 ## Responding to an unhappy message Answer the substance before the emotion. A short acknowledgement that the outcome was wrong, one concrete action, and a time is more calming than an elaborate apology with no remedy attached. Do not explain internal process; the customer is not asking how your systems work. Move it off text if the thread is getting long or the facts are contested. Two people typing at each other in short bursts is a bad medium for a disagreement. Offer a call, name the person who will make it, and follow up in the thread afterwards so there is a record. 11 ## Responding to a stop request Treat any clear request to stop as immediate and permanent. Suppress the contact across every campaign and workflow, not only the one that sent the last message, and do not send a marketing “sorry to see you go” afterwards. CTIA’s Messaging Principles and Best Practices set the industry expectation that senders honor opt-outs, and applicable law may impose stricter requirements. The request will not always use the word STOP. “Please don’t text me”, “wrong number”, and “take me off this list” are all opt-outs. Any automation you run needs to recognize the intent, not only the keyword. 12 ## When not to reply at all Silence is occasionally the right answer. Do not reply to reopen a conversation the customer has closed, do not reply to a stop request with anything other than suppression, and do not reply purely to bump the thread when you have nothing new to say. A “just checking in” with no new information is a message the recipient has to spend attention on and gets nothing for. If you need to follow up, bring a reason: a new slot, a new price, a resolved issue, an answer you owed them. 13 ## Measure reply quality, not reply volume Messages sent is not a result. Track the numbers that describe whether the conversation worked: how long the customer waited, how many exchanges it took, whether the question was actually resolved, and how many people asked to stop hearing from you. If you introduce automated drafting, measure the same set before and after on comparable conversations. A responder that halves your response time while doubling your opt-out rate has made the business worse, and only the paired metrics will show it. - Median and 90th-percentile time to first reply - Turns to resolution - Resolved-on-first-reply rate - Escalation rate to a human - Opt-out rate - Correction rate on generated drafts Frequently asked questions ## Quick answers What is the best way to respond to a customer text message?+ Acknowledge in a clause, answer the specific question that was asked, and close with one concrete next step that names an owner and a time. Keep it to two to four short sentences and ask at most one question. Should I use an AI responder for text messages?+ Use it for triage, record lookup, and drafting, with a person approving the send. Expand to unattended replies only for narrow intents you have measured, and always keep a fast path to a human. How quickly should a business reply to a text?+ Fast enough to match the expectation the medium creates. Publish a response target you can actually staff, acknowledge quickly even when the full answer is slow, and measure your 90th-percentile time rather than your average. What counts as an opt-out?+ Any clear request to stop, not only the word STOP. Suppress the contact across every workflow immediately and permanently, and do not send a farewell marketing message afterwards. Primary sources ## Read the documentation. - [CTIA: Messaging Principles and Best Practices (May 2023)](https://api.ctia.org/wp-content/uploads/2023/05/230523-CTIA-Messaging-Principles-and-Best-Practices-FINAL.pdf) - [Apple Support: About iMessage](https://support.apple.com/guide/iphone/about-imessage-iph4e9799206/ios) Explore the platform ## Turn the research into a working conversation. [For developers ### iMessage API Build two-way blue-bubble conversations into your product or agent. Explore](https://missblue.dev/imessage-api) [For teams ### Message Center Use a complete shared conversation platform without writing code. Explore](https://missblue.dev/features/message-center) [For industry teams ### iMessage use cases See practical playbooks for real estate, plumbing, HVAC, roofing, and home services. Explore](https://missblue.dev/use-cases) [Miss Blue Research ### iMessage vs SMS study Read the beta observations, reported benchmarks, methodology, caveats, and pilot framework. Explore](https://missblue.dev/imessage-response-rate-study) Ready to build? ## Send your first blue bubble with Miss Blue. [Explore the iMessage API](https://missblue.dev/imessage-api)