SMS open rates sit around 98%, and most texts get read within minutes. Marketers have measured those numbers for years, and email has never come close. For a restaurant, a text sent at 17:15 can drop your kitchen into the dinner decision of hundreds of past customers at once, for a cent or two apiece. The flip side is that a lazy blast lands on 98% of your list too. Every text you send is either welcome or resented. Nothing just sits there ignored.
The channel comparison, honestly
Email is cheap and can run as long as you like, but a typical restaurant email gets opened by about one in five people, hours or days after it lands. That's fine for a newsletter and useless for "order dinner now." Push notifications cost nothing and arrive instantly, but they only reach customers who installed your app and left notifications on. SMS reaches anyone who ever gave you a phone number, app or not, smartphone or not. The sensible setup uses all three: push first because it's free, SMS for the people push can't reach, email for anything long.
The cost math: one order pays for the month
Gateway list prices start at $0.0083 per message (Twilio's US rate as of mid-2026), and carrier surcharges of $0.0025–0.0065 bring the realistic all-in cost to about 1.1–1.5 cents. Text 1,000 customers and you've spent roughly $15. If two of them place a $30 order, the campaign already paid for itself, and a decent offer to a warm list converts a lot better than two in a thousand. Set that against winning the same order on a marketplace at a 25–30% commission: $7.50–9.00 on a single $30 ticket. The $15 that covers two marketplace orders' worth of commission covers a message to your whole customer base.
Five texts that earn their two cents
Start with a fence around your slowest day: "Tuesday only: free delivery on orders $25+, order at [link]. Reply STOP to opt out." Send it Tuesday at 16:30, to everyone. Because it's pinned to your weakest day, it adds orders instead of shaving margin off the Friday you'd have sold anyway.
The lapsed regular is the one to guard: "Hey [name], it's been a while, 15% off your next order this week: [link]." Send it once, 45–60 days after someone's last order. This is the most valuable text on the list, a customer you were about to lose brought back for pennies.
Expiring credit almost writes itself: "Your $5.80 cashback expires Sunday. Spend it: [link]." Send it to wallet holders who never installed the app, since the offer is their own money.
A launch note works when you keep it to fans: "You've ordered our ramen 4 times. The new spicy tonkotsu drops Thursday, want first crack? [link]." Only the segment with matching order history should get it.
Then the transactional workhorse: order confirmed, courier on the way, order delivered. None of it is marketing, but these texts teach customers that your number is worth reading, and that habit is what keeps them from blocking the promotional ones.
Dots segments customers by order history automatically — the right text, the right people, the right Tuesday.
Frequency: the rule that protects the asset
Cap it at two promotional texts per customer per month. That's the ceiling, not the target. It sounds painfully conservative until you price the alternative. A customer who opts out is unreachable for good, and US carriers increasingly filter senders with high complaint rates, so one spammy month can drag down deliverability for the entire list. Low frequency is only workable if you segment. Group customers by recency and order history (the RFM method takes about 20 minutes), and each segment gets the single text that applies to it instead of everyone getting everything.
Consent is not optional paperwork
US law (TCPA) requires express written consent before you send marketing texts, an unambiguous opt-out in every message, and immediate honoring of STOP replies. In practice that means collecting the phone number at checkout on your ordering website with a clear marketing checkbox, keeping transactional and promotional streams separate, and never buying lists. A delivery restaurant has it easy here, because every order already comes with a phone number and a natural reason to ask. A platform-level setup handles STOP processing and quiet hours for you through the same marketing tools that run your push and cashback campaigns, and revenue per campaign shows up in reports.
Where SMS fits in the bigger machine
Across the 3M+ orders we've processed, the pattern holds: SMS is the bridge channel. It reaches the customers who ordered by phone or website and never installed the app, and its best long-term job is to move them into the app, where messaging is free and conversion runs as high as 35%. A text that says "install the app, get $5 credit" pays for itself on the first push notification you no longer have to buy.
Start here. Export the customers with no order in 45+ days, send them the lapsed-regular text above, and count what comes back this week. That one segment usually funds the whole SMS program.
See what a coordinated retention stack does to your repeat rate. Live in about two weeks.