Most phone loyalty enrollment fails for a reason that has nothing to do with the offer. It fails because it happens at the wrong second of the call, and it asks for more than a person can give over a phone line.
The version that works is short enough to be forgettable. A recognized number gets credited silently. An unrecognized number gets one sentence, after the order is done, with a reason attached. That is the whole design.
The order comes first, always
A caller dialed you to get food. Anything that delays that is friction they did not sign up for, and the cost shows up as abandoned calls rather than as declined enrollments, which makes it easy to miss.
Put the ask after the order is complete. The caller has already gotten what they wanted, the transaction is settled in their head, and a fifteen-second addendum reads as a small favor rather than an obstacle. Put it before the order and it reads as a toll booth.
There is one exception worth making. If a caller volunteers a phone number early, because they are a regular and expect to be recognized, take it then and skip the ask entirely later.
Fifteen seconds means one field
The realistic ceiling on a phone call is a phone number. Possibly a first name.
Email addresses are the trap. Spelled aloud, a moderately unusual address takes twenty seconds and gets transcribed wrong often enough that the resulting list is dirty. Names with unusual spellings have the same problem. Birthdays, which loyalty programs love because they drive a reliable annual offer, feel intrusive when a stranger asks for one mid-order and the refusal rate reflects that.
So collect the number, and let everything else arrive by text. A phone number is the one field a caller can give quickly, accurately, and without feeling examined, and it doubles as the identifier your POS actually uses.
Why the number is the useful field anyway
The phone number is what ties a caller to their history, which is what makes per-customer rules enforceable. It is the same identifier that lets you cap a one-per-customer offer, which is the practical problem in honoring promo codes on phone orders. Without it, phone orders are anonymous and every limit is theoretical.
Say what they get, not what it is called
"Would you like to join our rewards program?" is a question about paperwork. It gets declined because it describes an obligation rather than a benefit.
"I can put this order toward your rewards, you'd be about a third of the way to a free entree" is a question about food. It describes something the caller already earned by placing the order they just placed.
The specific mechanic matters less than the concreteness. Points-per-dollar language is abstract; "one more order and your next appetizer is on us" is not. If your program's math does not translate into a sentence like that, the program is the problem, not the script.
Recognition is the part that pays off
The enrollment is the setup. The payoff is what happens on call two.
When a returning number is recognized, the agent can greet by name, offer the last order as a starting point, and skip the address confirmation for delivery. That saves the caller thirty seconds and saves you a repeated data entry. It also makes the system feel like it belongs to your restaurant rather than to a vendor, which is the difference most operators are actually worried about when they consider automation on the phone. Voice AI and loyalty programs covers how that recognition layer connects to the POS side.
The corollary is that asking an enrolled member to enroll is worse than not asking at all. It is a small failure that tells the caller nobody is home. Check membership before the prompt fires, and if the lookup is unavailable, skip the prompt rather than guessing.
Where phone enrollment is a bad idea
Say it plainly: if your loyalty program requires an app install before anything accrues, phone enrollment mostly produces accounts that never activate. You will book an enrollment number that looks fine and a redemption number that does not move.
The fix is not a better script. It is making the phone number sufficient on its own, so the visit is credited immediately and the app becomes a way to see a balance that already exists. If your provider cannot do that, the honest answer is to capture numbers for marketing and stop calling it loyalty enrollment.
Similarly, if you run a punch-card program that lives on paper at the counter, there is no phone version. Do not build one out of a spreadsheet somebody has to reconcile.
Handling the caller who says no
The decline is more common than the acceptance, and how it is handled determines whether the next ask lands. One "no thanks" should end the topic for that call and, ideally, for the next several.
A system that asks the same caller on every order is training them to talk over the prompt, and once a caller has learned to ignore one part of the call they tend to stop listening to the rest of it, including the readback. That is a real accuracy cost paid for a marginal enrollment gain.
Set a suppression window. If a number declines, do not ask that number again for a month. If it declines twice, stop asking it entirely. The people who want your program will join early; the ones who do not are not persuaded by repetition, they are just annoyed by it.
The same restraint applies to callers who give a number but refuse the rest. Take what they gave, credit the order, and do not push for the email. A partial record that gets used is more valuable than a complete one you never earned.
What to send afterward
One text, immediately, containing the confirmation and the enrollment in the same message. The order confirmation is the part the caller wants, and the loyalty link rides along with it, which is why text confirmations on voice orders are worth having in place before you add enrollment on top.
Do not send a separate marketing message that day. The enrollment was a small yes, and immediately spending it on a promotional blast is how a program gets muted in the first week.
Measure two numbers, not one
Enrollment rate is the obvious metric and the less useful one. Track it, but track second-order rate next to it: of the numbers enrolled by phone this month, how many placed another order within sixty days.
A program that enrolls forty percent of callers and never sees them again is collecting phone numbers, not building repeat business. A program that enrolls fifteen percent and brings a third of them back is working. If the offer is what changes behavior, the second number moves when you improve the offer, and it barely moves when you improve the script.
Run it for one month before changing anything. Then change one variable, keep the placement in the call fixed, and compare against your own prior month rather than against a figure from someone else's restaurant.