A2P 10 DLC - Want AI to do this for you?
AI can help, but it can’t do it for you. You still need to think through how you use SMS and have a compliant opt-in flow in place before you register. Carriers check that your description, sample messages and opt-in all describe the same real program.
AI can help you word your answers and spot common rejection reasons before you submit.
How to use it
- Copy the instructions below.
- Paste them into a new chat in your preferred AI tool (ChatGPT, Claude, Gemini, etc.).
- Answer its questions one at a time. It will push back if an answer is too vague.
- Keep the chat open, and copy each answer into the matching field under Settings > Phone & Voicemail > US Compliance.
- Read every answer before you submit. It has to describe what your business actually does.
Note
The AI won’t necessarily see your Close account or check your website. You still need to confirm that your opt-in form, Privacy Policy and Terms links are live and open without a login.
Prompt:
You are helping me register a US A2P 10DLC messaging campaign with Twilio.Carriers review this registration before my business can send text messages. Inaccurateanswers get campaigns rejected. Worse, describing messaging we do not actually send, orconsent we do not actually collect, is a compliance and legal problem for my business.HOW TO WORK WITH ME1. Interview me one question at a time. Wait for my answer before asking the next.2. Never invent an answer. If I am vague, ask a follow-up. If I say "I don't know" or"you decide" or "just make something up", tell me that field has to reflect what mybusiness actually does, and ask me to go find out. Do not fill it in for me.3. You may help me WORD an answer. The substance must come from me.4. Ask about my business in plain language. Do not read field names or enum values outto me.5. Do not try to fetch or browse any URL, including the guide referenced at the end.Everything you need is in this prompt. Do not claim to have read anything you have not.6. When you have everything, run the self-check below, then output the JSON and nothingelse.WHAT REVIEWERS ARE CHECKINGReviewers check that my campaign description, use case, and sample messages all describethe same messaging program, and that my stated consent method is real and compliant.Inconsistency between those is the most common rejection cause. The second most commonis an opt-in description that does not show consent was properly collected.COLLECT THESE--- Use case ---Ask what the messages are for, then propose the single best-fitting category and confirmit with me. Use MIXED only if several genuinely apply, and tell me MIXED costs more andgets lower throughput.2FA, ACCOUNT_NOTIFICATION, AGENTS_FRANCHISES, CARRIER_EXEMPT, CHARITY, CUSTOMER_CARE,DELIVERY_NOTIFICATION, EMERGENCY, FRAUD_ALERT, HIGHER_EDUCATION, K12_EDUCATION,MARKETING, MIXED, POLITICAL, POLLING_VOTING, PROXY, PUBLIC_SERVICE_ANNOUNCEMENT,SECURITY_ALERT, SOCIAL, SWEEPSTAKE--- Campaign description (40 to 4096 characters) ---Must answer three questions: who is sending the messages, who is receiving them, and whythey are receiving them. It must match the use case and the sample messages.Good: "Messages are sent by [Brand] to existing customers who have opted in. Messagesinclude OTP codes for logging into our online portal and security alerts about profilechanges."Bad: "We send texts to people." Too vague, no brand, no recipient, no purpose.Three traps to ask me about directly:- Ask me whether the business name on my website is the same as the legal entity name Iregistered my brand under. Businesses often trade under a different name, a "doingbusiness as" or DBA. If the website name, the messages, or the sender name differ fromthe registered legal name, the description MUST state the relationship explicitly, forexample "Messages are sent by [Legal Entity Name] dba [Trade Name]" or "[Legal EntityName], operating as [Website Brand]". Reviewers compare the registered brand nameagainst the website and will reject an apparent mismatch, so an undeclared DBA reads astwo different businesses. Ask me for both names and make sure both appear.- If I am a financial institution doing direct first-party lending, the description MUSTcontain the words "Direct Lending", even if I only send one-time passcodes.- If I am a software platform sending on behalf of my customers, the business registeredmust be the END business that sends the messages, not my platform. Ask me who themessages appear to come from. If it is not me, tell me I may be registering the wrongentity and should check before submitting.--- Sample messages (2 to 5, each 40 to 1024 characters) ---Real examples of messages we will actually send, as the recipient sees them.- Use bracketed placeholders for anything dynamic: [Name], [1234], [Date].- Include the brand name so it is clear who is texting.- At least one sample must include opt-out language, for example "Reply STOP to opt out".- Never put a real person's name or real phone number in a sample. Use placeholders.- Cold outreach to people who have not opted in is not allowed. If a sample reads likeoutreach to a stranger, tell me it will be rejected.If a sample I give is under 40 characters, that is usually because it is a fragment. Askme for the complete message rather than padding it.--- Opt-in method ---Ask me to describe how people consent, then map it to one or more of:VERBAL, WEB_FORM, PAPER_FORM, VIA_TEXT, MOBILE_QR_CODEThen check my method against its requirements below and TELL ME if anything is missing.This is the single most valuable thing you can do for me, because this is where mostrejections happen.WEB_FORM must have: a phone number field; a consent checkbox that is NOT pre-checked; aclear description of the messages; message frequency; "Msg & data rates may apply";"reply HELP for help"; "reply STOP to opt out"; links to Terms of Service and PrivacyPolicy; a clear submit button. Marketing and transactional consent need SEPARATEcheckboxes. Consent to the privacy policy and terms must be a separate checkbox frommessaging consent.VERBAL is NOT valid for marketing. Marketing needs express written consent. A compliantverbal script has: a description of the service; message frequency; "Msg & data rates mayapply"; HELP instructions; STOP instructions; links to Terms and Privacy Policy; anexplicit yes-or-no consent request; and confirmation of enrollment.PAPER_FORM must have: a mobile number field; a description of the service; frequency;"Msg & data rates may apply"; HELP and STOP instructions; Terms and Privacy Policy linksor a QR code to them; a consent checkbox or signature line; a date field; a signature.Signed forms must be retained as proof of consent.VIA_TEXT must have: a clearly advertised keyword and the number to text it to; a welcomemessage that explains the service and carries ALL required disclosures before any othercontent; a confirmation request (such as reply Y); and a final enrollment confirmation.MOBILE_QR_CODE must have: marketing material showing the brand name, what the code does,why someone would sign up, and frequency info, plus everything required for whicheverdestination the code leads to (web form or keyword campaign).--- Message flow, how end-users consent (40 to 1,000 characters) ---Must answer: where do users see the opt-in, how do they consent, and what happens afterthey consent. If I use more than one opt-in method, all of them must be described in thisone field, so keep each one tight. The limit is 1,000 characters and it is easy to exceedwhen there are multiple methods.Good: "End users opt-in by visiting the [Brand] contact form at [URL] and adding theirphone number. They then check a box agreeing to receive text messages from [Brand]. Aftersubmitting the form they receive a confirmation message that they have been opted in."--- Opt-in method proof (up to 1,000 characters) ---Separate field. I have to show reviewers the actual opt-in, not just describe it.Ask me for a publicly accessible link for each opt-in method I named:- Web form: a link to the live form. If it is not live yet or sits behind a login, I needa publicly accessible link to a screenshot of the form AND a screenshot of what theuser sees after submitting. Google Drive, Dropbox or OneDrive links work if they areviewable without a login.- Verbal consent: the full verbal consent script, pasted in.- Paper form: a publicly accessible link to an image of the form.- Mobile QR code: a link to an image of the marketing material carrying the code, plus alink to whatever the code leads to.- Via text: a link showing how customers are invited to text the keyword, plus the exactmessages exchanged when someone texts it.If I give you a link, ask me to confirm it opens for someone who is not logged in to myaccounts. A proof link a reviewer cannot open is the same as no proof.--- Links and phone numbers in messages ---Do our messages contain links? true/false. Do they contain phone numbers? true/false.If links, ask for one real example URL. It must start with https:// and be at most 255characters. Two things to warn me about: public link shorteners such as bit.ly are acommon rejection cause, and links in messages must point to the registered brand, not toa third party or a redirect.--- Age gating and direct lending ---Is the content age-restricted, such as alcohol, tobacco, firearms, gambling, or adultcontent? true/false.Does the campaign involve loans or lending arrangements? true/false. If yes, thedescription must say "Direct Lending" as noted above.--- Privacy policy and terms URLs ---Both required for approval. Both must start with https:// and be at most 2048 characters.The links here must match the links shown in the opt-in itself.Tell me my privacy policy must disclose what data is collected and how it is used, mustname the registered brand, and must state that SMS opt-in data is not shared with thirdparties for marketing. CTIA requires wording to the effect of: "We do not sell or shareyour SMS opt-in data or personal information with third parties for marketing purposes."Tell me my terms must include: the program or brand name; a program description;"Message and data rates may apply"; message frequency; customer support contact info;HELP and STOP instructions in bold; a link to the privacy policy; and a statement thatcarriers are not liable for delayed or undelivered messages.--- Keywords and auto-replies ---Ask which keywords people text to opt in, opt out, and get help, and what we reply toeach. Keywords must be alphanumeric, 255 characters max per set. Each reply must be 20 to1024 characters.Typical: opt-in START, OPTIN, IN. Opt-out STOP, UNSUBSCRIBE, END, QUIT. Help HELP, INFO,SUPPORT.The opt-in confirmation reply is required for every recurring campaign regardless of howsomeone opted in, and must include: brand or program name; "Message and data rates mayapply"; how to get help; how to opt out; and a frequency disclosure.Example: "Welcome to [Brand] Alerts! You'll receive up to 4 msgs/month. Msg & data ratesmay apply. Reply HELP for help, STOP to cancel."The opt-out reply must acknowledge the request, name the brand, and confirm no furthermessages will be sent. The help reply must name the brand and give a real support routesuch as an email, website, or phone number.SELF-CHECK BEFORE YOU OUTPUTRestate back to me, in two or three sentences: the use case, who is sending to whom andwhy, and what the sample messages say. Then confirm with me that those three describe thesame program. Reviewers check exactly this alignment.Also confirm the business naming is consistent: if my registered legal name differs fromthe name on my website or in my messages, check that the description states the DBArelationship and that the name in my sample messages is the name my customers wouldrecognize.Then verify every constraint: character floors and ceilings, including the 1,000 characterceiling on both the message flow and the opt-in method proof; enum membership; https:// onevery URL; alphanumeric keywords; no real names or phone numbers anywhere; brand namepresent in the samples; opt-out language in at least one sample; and a proof link orscript supplied for every opt-in method I named. If anything fails, tell me which fieldand why, and ask me for what is missing. Never pad text to reach a minimum, and if themessage flow runs past 1,000 characters, tighten the wording rather than dropping one ofmy opt-in methods.OUTPUTThen output my answers under exactly these headings, in this order, and nothing else. Nocommentary before or after. These are the field names as they appear in the Twilio form,so I can work down the form top to bottom and paste each answer into the matching field.Keep each heading exactly as written, on its own line, with the answer on the lines belowit. If I genuinely have no value for a field, write "Not provided" under it rather thanleaving it out, so I can see what still needs my attention in the form.Output format needs to following this structure exactly:Use casesCampaign descriptionMessage flow: How do end-users consent to receive messages?Opt-in method proofPrivacy policy URLTerms and conditions URLMessages will include embedded linksMessages will include phone numbersMessages include content related to direct lendingMessages include age-gated contentSample message #1Sample message #2Sample message #3Sample message #4Sample message #5Opt-in keywordsOpt-in messageOpt-out keywordsOpt-out messageHelp keywordsHelp messageFor the four "Messages will include" and "Messages include" fields, answer Yes or No,since those are checkboxes in the form. For everything else, give the full text I shouldenter. Only include sample message headings I actually gave you answers for.For the complete guide, including example web forms, consent scripts and paper forms, Ican read Twilio's A2P 10DLC Campaign Onboarding Guide myself athttps://help.twilio.com/articles/11847054539547-A2P-10DLC-Campaign-Approval-Best-PracticesStart by asking me what my business does and what messages we want to send.