A marketplace or gig platform should send SMS through one REST API, send transactional notifications to both sides of the market, get consent from each person when they sign up, and decide early whether it texts as itself or on behalf of the businesses on its platform. That last decision drives your registration, your sub-account layout and who answers when a recipient replies STOP.

Two-sided platforms are an awkward fit for generic texting advice. You have two kinds of users, often a third party (a merchant, a property, a venue), time-sensitive messages, and a strong temptation to let people text each other through your numbers. This post goes through the notification types, the sender models, the masked-number question and the consent rules that apply.

The messages a two-sided platform actually sends

Most marketplace texting is transactional: it's about something the person already asked for. Group them by who's receiving and how time-sensitive it is.

MessageToTime-sensitive?Notes
Booking or order confirmedCustomer, providerModerateOne text when the state changes
Provider on the way, arrivedCustomerHighShort, and only on real state changes
New job or request availableProvider, driverHighOften the highest-volume message you send
Schedule change or cancellationBothHighBoth sides need the same truth
Payout or payment noticeProviderLowKeep amounts out of long messages
Review request after completionCustomerLowMarketing-adjacent, so treat consent carefully
Verification codeEitherHighKeep it on a separate path from marketing

If one-time codes are part of your mix, our post on separating MFA from marketing numbers covers why those shouldn't share a lane with promotions.

A single send is one call: POST /v1/messages with an X-API-Key header, a to in E.164, a body, and consent_attested set to true. Add an Idempotency-Key header with one UUID per message and reuse it on every retry. That matters on a marketplace, where a retried "driver arrived" text is both annoying and confusing. For a same-text-to-many message, POST /v1/messages/bulk takes a list of recipients, up to 100 per call.

Decide who the sender is

This is the structural choice, and it comes before any code.

Model A: the platform is the sender. Texts come from your brand: "Your ride is arriving", "New job near you". You have one brand, one use case and one registration. A single account may be enough, and it's the simplest model. Most rider, driver and provider notifications fit here.

Model B: each business on the platform is the sender. If a merchant, venue or independent business texts its own customers through your product, each one is its own sender with its own brand. Ready gives each customer a sub-account with its own numbers and its own carrier registration (10DLC). You create one with POST /v1/sub-accounts and register it through the registration endpoints. See the sub-accounts docs, our multi-tenant SMS architecture post, and our 10DLC registration API guide.

Many platforms are both: Model A for the platform's own notices, Model B for businesses that text their customers. Keep them in separate accounts, so one tenant's problem doesn't touch the platform's own traffic.

Registration has a cost per sender: $25 a month per customer for the brand and its first campaign, plus $2 a month per number. Most approvals take 2 to 7 days. Ready registers businesses with an EIN. An individual gig worker without one can't register as their own sender, which is one more reason to prefer Model A for individuals.

The masked-number question

Many marketplaces want a rider and a driver, or a customer and a provider, to text each other without seeing each other's real numbers. That's a proxy or relay pattern, and it's a different thing from sending notifications.

Be careful here. This post doesn't describe a number-masking or proxy feature on Ready, and you shouldn't assume one exists. Confirm what a provider supports before you design around it. If it's not a native feature, the way to approach it is to relay in your own app:

  1. The sender writes a message inside your app or replies to a text from the platform's number.
  2. Your server receives it (the inbound message arrives by webhook; see receiving SMS with inbound webhooks).
  3. Your server forwards the body to the other party from the platform's number, with a label such as the sender's first name.

The cost and the risks are yours to weigh. You're now a relay: you see the content, you need a retention policy, and every relayed message is a billed segment. Replies are billed like a sent text, at the per-segment rate plus the carrier fee. Also be clear with both users that messages pass through your platform.

Sub-accounts per tenant, in practice

If you chose Model B, layout matters.

  • One sub-account per tenant keeps numbers, registration and billing separate and keeps a noisy tenant away from everyone else.
  • Store the sub-account ID against your tenant record, and never accept it from a request body. Derive it from the authenticated tenant.
  • Think about offboarding. When a tenant leaves, you need a plan for the numbers; phone number release and recycling for platforms covers it.
  • Know your tenant. A marketplace attracts all kinds of senders. Our KYC for SMS sub-accounts post lists what to collect before you give someone a sender identity.

Consent for both sides of the market

Consent is the sender's job, and on a marketplace there are three common traps.

  • Consent at sign-up must name texts. A rider or provider who ticks a general terms box hasn't clearly agreed to text messages. Add a specific line at the point you collect the phone number: what kinds of messages, roughly how often, that message and data rates may apply, and that they can reply STOP.
  • Transactional isn't a loophole. Even a booking notice goes to a person who gave you the number for that purpose. A promotion sent to the same number needs its own consent.
  • A provider's customer isn't your customer. If a tenant imports its own client list into your platform, the tenant has to hold consent for those people. Make them confirm that before a send.

Ready records your attestation when you set consent_attested, but it doesn't store a per-contact consent record: no recipient opt-in time, wording or page URL. You keep that evidence in your own database. See a consent record schema for developers.

STOP, quiet hours and timing

Opt-outs are automatic: an inbound STOP is honored, and clear phrases such as "please stop texting me" are caught too. A STOP always blocks that number in the sub-account that received it. Whether it also blocks the parent and other sub-accounts follows a Ready setting, so check GET /v1/opt-outs/:phone for a number before a send from another tool. Handle the case in your UI: if a driver replies STOP to a job alert, your app should show that SMS alerts are off and offer push or email.

Quiet hours matter for night-shift work. Ready enforces them by the recipient's area. For an API text, it's held only if the window opens within 6 hours. Otherwise the API answers 409 QUIET_HOURS_BLOCK and the text isn't sent, so your code sends it again once the window opens, or passes hold_until_allowed: true to hold it up to 24 hours. A text that replies to someone who texted the account in the last 24 hours goes out at any hour. A two-sided product with urgent jobs should read the error code list and decide, per message type, whether to wait or use another channel.

Frequently asked questions

Should every provider or driver on my platform get their own sender?

Usually not. Individuals can't register without an EIN on Ready, and most notifications come from the platform. Use sub-accounts for businesses that text their own customers.

Can riders and drivers text each other through my numbers?

You can relay messages through your server, but we don't describe a built-in masking feature, so confirm support before you plan on one. A relay also makes you responsible for the content passing through.

Do I need consent for transactional texts?

Yes. The person must have agreed to receive texts at that number. Put the agreement at sign-up, in words that name text messages.

Where to start

Pick the sender model first, write the consent line second, and wire up one notification type third. If you'd like to see the pricing for a planned volume, the pricing page has the figures, and the API quickstart gets a first send out.