Texting ski resort guests: the rules to settle before the season

In the US, a text message counts as a call under the Telephone Consumer Protection Act (TCPA): the FCC said so again in 2015 (FCC 15-72, paragraph 107). On top of the law, the wireless industry’s own rules, CTIA’s Messaging Principles and Best Practices (May 2023), sort business texts into three kinds by how much permission each needs (CTIA, Exhibit II). They are industry guidance, not law, but they set what the messaging ecosystem expects of senders.

• A reply to a guest who texted first. CTIA: if the guest starts the exchange and you only respond with relevant information, “no verbal or written permission is expected”. The FCC has said much the same for a one-time text sent in response to a consumer’s request, as long as it is requested, sent straight away and contains only the information asked for, “with no other marketing or advertising information” (FCC 15-72, paragraph 106). That ruling is about one answer; a longer back-and-forth is what CTIA calls conversational messaging, with the same implied consent.
• Reminders and alerts. CTIA calls them informational: the guest gives you their number and asks to be contacted, and agrees to receive texts “for a specific informational purpose”. A guest who opted in to lesson reminders has not opted in to snow alerts.
• Promotions. CTIA asks for express written consent before any promotional text, and warns that adding a call to action such as a coupon code to an informational text may make it promotional. Federal law goes the same way for marketing texts sent with an autodialer: they need the guest’s prior express written consent (47 CFR 64.1200(a)(2)), a written agreement, which can be signed electronically, that names the number and tells the guest that signing is not a condition of buying anything (64.1200(f)(9)).

Whether a given texting platform counts as an “autodialer” under the TCPA is a legal question we do not answer here. Do not build your plan on the assumption that it does not.

• The words that end consent. Under the FCC’s rule, a reply of “stop”, “quit”, “end”, “revoke”, “opt out”, “cancel” or “unsubscribe” revokes consent, and so do other words a reasonable person would read as a request to stop (47 CFR 64.1200(a)(10), 2025 edition). CTIA asks senders to act on plain-language requests too, regardless of capitals or punctuation.
• One confirmation, then silence. CTIA: send one final message confirming the opt-out, and nothing after it (section 5.1.3).
• HELP. For recurring messages such as reminders, your opt-in confirmation should say how to reach customer care, for example a phone number or “HELP command instructions”, how to opt out, how often you will text and any fees or charges (CTIA, section 5.1.2.1). Your 10DLC campaign registration also asks for your HELP keyword and the reply guests get when they send it, as it does for STOP (TCR CSP user guide). The HELP reply tells guests how to reach you; it does not end anything.
• The rules are changing. On 30 September 2026 the FCC adopted changes to how opt-outs work (FCC 26-67). A sender may treat an opt-out from an informational text (one with no advertising) as covering only that kind of informational message, and may make one or more of the opt-out methods the rule lists, such as replying STOP, the only ones it accepts, as long as each text says so. An opt-out from a marketing text still ends all marketing texts from that sender. The changes take effect 30 days after publication in the Federal Register, which had not happened by 15 October 2026, and the FCC is consulting on further changes. Ask your messaging provider which version applies when you launch.

The simplest policy survives every version: when a guest says stop, stop at once, for everything, and confirm it once.

To text guests from an ordinary 10-digit number, the resort registers for 10DLC, the carriers’ system for business texting. The Campaign Registry (TCR) lists AT&T, T-Mobile, Verizon, US Cellular and Interop as launched and requiring it (TCR, August 2025). You do not register with TCR yourself: “direct registration with TCR is not available for Brands”, so it goes through your messaging provider (TCR). There are two steps (TCR CSP user guide, April 2026):

• The brand: your company. Legal company name, country, EIN or tax ID, address, website and contact details. A brand that cannot be verified cannot register campaigns, and TCR names the most common reason: the legal company name and the EIN do not match what is filed with the federal government. Use the exact name on your tax records, not the resort’s trading name.
• The campaign: what you will send. One per use case, with sample messages. Answering guests fits “Customer Care”, defined as “all customer care interaction”, including account management and customer support. Promotions go under “Marketing”. “Mixed” covers 2 to 5 uses on one campaign.

Leave time for review. TCR says campaign review can take 1 to 4 weeks, and that TCR itself “does not review, approve, or reject campaigns”: the providers and partners in the chain between you and the carriers do. So no one can promise you a go-live date, and a rejection means another round. Start in the summer or early autumn, not the week before opening day.

The alternatives. Short codes (5 or 6 digits) are typically used for high-volume messages such as airline flight delays and bank account alerts, and run through the Common Short Code Administration (CTIA, section 5.5.4); for answering guests they are more than you need. Toll-free numbers can also be enabled for texting by the business that holds them (CTIA, section 5.5.6); ask your provider about its verification process and timing before you choose.

Apple Messages for Business puts your resort in the Messages app on iPhone and iPad. It works differently from texting (Apple FAQ, version 3.1, 21 July 2026):

• The guest starts the conversation, by tapping a Messages button on your website, app, emails, a QR code or your Apple Maps listing.
• You need an Apple-approved Messaging Service Provider. Apple provides the channel itself free of charge; the provider may charge for its service.
• A person must be available. Apple requires a blend of automated and human agents, and “deployments without live agent support will not be approved”.
• Follow-ups only to guests who wrote to you, and only in their interest: Apple says proactive messages should not be used for promotional or marketing content unless the guest explicitly asked for it, and asks you to support the STOP and UNSUBSCRIBE keywords.
• Call it “Apple Messages” in what guests see, not “iMessage”.
• Allow a few weeks. Apple says its approval can take a few days, and the experience review, in which Apple signs off on automated agent interaction, 1 to 2 weeks (Apple, end-to-end guide).

Guests need an iPhone, iPad or iPod touch to use it (Apple Support, 5 June 2026), so Apple Messages sits alongside texting, never in place of it.

If you host guests from abroad, some will want to reach you on WhatsApp rather than text a US number. Through the WhatsApp Business Platform (Meta, pricing page updated 30 September 2026):

• Opt-in first. You can only message people who have opted in to hear from you.
• The 24-hour window. Each message from the guest opens or resets a 24-hour customer service window. Inside it you can reply freely; outside it, only with a template Meta has approved (marketing, utility or authentication).
• No promotions to US numbers. “WhatsApp does not currently deliver marketing template messages to WhatsApp users with United States phone numbers” (Meta, per-user marketing limits). Guests on foreign numbers can receive them if they opted in.
• Replies are charged again. Meta charges per delivered message, by category and by the country code of the guest’s number. Since 1 October 2026 that includes service replies, whether a person or an AI writes them, after 1,000 free ones per business phone number per month. Messages from guests to you are free. Our guide to WhatsApp service message pricing has the details.
• A way to a person. WhatsApp’s business policy allows automation inside the 24-hour window, but you “must also have available prompt, clear, and direct escalation paths” (WhatsApp Business Messaging Policy, 23 September 2026). Whether an AI agent may answer your guests there at all: yes, and we explain why in Is AI allowed on WhatsApp Business?

Our recommendation is simple: the agent says it is an AI in its first message, and says so again whenever a guest asks. It costs nothing and meets the California and Utah rules below. If Colorado’s HB26-1263 applied to your agent, it would ask for more: the disclosure at the start of each day’s first interaction, and again at least every three hours in a continuous conversation unless a notice stays visible. What the state laws actually say:

• California. Business and Professions Code section 17941 makes it unlawful to use a bot to communicate with a person in California online “with the intent to mislead the other person about its artificial identity”, in order to encourage a purchase or influence a vote. There is no liability if you disclose that it is a bot, clearly and conspicuously. “Online” means a public-facing website, web application or digital application (section 17940). In force since 1 July 2019 (California Legislative Information).
• Utah. A business (a “supplier”) that uses generative AI to interact with someone in a consumer transaction must disclose that they are talking to AI, not a person, if they clearly and unambiguously ask (Utah Code 13-77-103, effective 7 May 2025).
• Colorado. Two 2026 laws set duties from 1 January 2027. SB26-189 covers automated decision-making technology used to materially influence “consequential decisions” in areas such as employment, housing, lending, insurance and health care (SB26-189). Its definition of a consequential decision leaves out low-stakes or routine processes that do not materially influence one, naming routine scheduling, customer service triage and communication of decisions among them, and also leaves out advertising and marketing. Its definition of the technology leaves out tools that answer consumers’ questions in plain language, provided they are not intended for consequential decisions and an acceptable use policy forbids that use. HB26-1263 requires operators of a “conversational artificial intelligence service”, one open to the public that primarily simulates human conversation, to disclose that it is AI; its definition excludes software primarily designed for commerce-related or transactional help, “including” customer support and customer service, and features of a theme park or location-based entertainment that keep to that venue’s topics (HB26-1263). Employment is one of SB26-189’s covered areas, worth knowing if an agent ever screens seasonal job applicants. Whether your agent falls inside either law, or inside one of their exclusions, is a question for your counsel.

Apple and WhatsApp add their own expectation that a guest can reach a person (sections 04 and 05). An agent that says what it is and hands over cleanly is the common ground of all of them.

Marketing email follows the CAN-SPAM Act. The FTC’s guide for business lists what each commercial email needs (FTC):

• Honest headers and subject lines, and a clear statement that the message is an advertisement.
• Your valid physical postal address.
• A clear way to opt out, honoured within 10 business days.
• Responsibility stays with you even if an agency or a platform sends the email for you.

Transactional or relationship messages, such as a booking confirmation, are exempt from most of these provisions, though their routing information must still be accurate. A plain answer to a guest’s question, with nothing promoted in it, is usually neither commercial nor transactional: the FTC calls it other content.

1. Pick the number and the provider for texting, and start brand and campaign registration with your exact legal name and EIN.
2. Write the sample messages for your campaign: replies to guests, plus any reminders you plan to send, and your opt-in, STOP and HELP replies.
3. Fix your opt-in wording for reminders (one purpose each) and, separately, for promotions (written consent, never a condition of purchase). Keep a record of who agreed to what.
4. Test STOP and HELP on every number: STOP gets one confirmation and then nothing; HELP gets your contact details.
5. Keep offers out of replies. Brief your team, and your agent if you use one.
6. Decide on Apple Messages and WhatsApp, with a provider for each, and leave a few weeks for Apple’s review.
7. Name who answers when the agent hands over, and during which hours.
8. Write the AI disclosure into the first message.
9. Check your marketing emails for a postal address and a working unsubscribe link.
10. Have counsel review the plan once, before opening day.

Answering guests who write first, the first kind of message, is where an AI agent helps most, including at 7 a.m. on a storm day. It answers by text and WhatsApp (and Apple Messages, through an Apple-approved provider) in the same words, says it is an AI, keeps offers out of replies, stops at STOP, and hands anything that needs a person (a refund request, a complaint, an injured guest) to your team with the conversation attached. It quotes lift status and snow only from your own system, with the time it was last updated, and never promises an opening date or a refund. Reminders go only to guests who asked for them.

That is how we build our AI agent for ski resorts, the same AI agent we build for customer service, set up for a resort; the AI agent for river outfitters shows the approach for another weather-dependent season. You can try it with your own guests’ questions in the demo, set up for a ski resort.