Yes, modestly. AI helps a small café turn occasional visitors into regulars by making follow-up cheap: spotting who has stopped coming, writing offers that feel personal, keeping your Google profile and social posts active, and replying to every review. It can't make the coffee better or greet people by name. The gains come from steady contact with past visitors.
That focus on existing customers is simple arithmetic. A regular who buys a $4.50 coffee three times a week is worth about $700 a year. Moving twenty occasional customers up to that habit is worth more than a busy month of one-off visitors. But AI can only follow up with people you can reach, so the first job isn't AI at all. It's collecting contact details, with permission, from the customers you already serve.
The regulars sum for your own café
Before any tool, work out where your customers sit today. If your till has a loyalty feature or a digital stamp card, export the last three months of member visits. Sort members into four groups and put a value on each. An illustrative 30-seat café with 400 loyalty members:
| Group | Definition | Members | Average spend per visit | Value per year |
|---|---|---|---|---|
| Regulars | 2+ visits a week (about 3 on average) | 60 | $6.20 | About $970 each |
| Occasionals | 2-7 visits a month (about 4 on average) | 140 | $7.80 | About $370 each |
| Drifting | Visited before, not in 3-6 weeks | 90 | $7.50 | Falling |
| Lapsed | No visit for 6+ weeks | 110 | n/a | Nearly zero |
The planning sum: if 10% of occasionals become regulars (14 people, each worth about $600 more a year), that's about $8,400. If 20% of the drifting group (18 people) settle back into an occasional habit worth about $370 a year, that's another $6,600 or so. Together, roughly $15,000 a year in extra takings for this café, before counting anyone new. You'll get different numbers, but the shape is usually the same: the drifting group is the cheapest to win back, and the occasionals are the richest seam.
A chat assistant can do this sorting for you from an export (with names and emails removed, or replaced with member numbers). Ask it for the counts in each group and check one group's count yourself with a filter in the spreadsheet.
Capture the customers you already serve
Most small cafés know their regulars by face and their occasionals not at all. Ways to change that, from easiest to most effective:
- A digital stamp card or loyalty feature on your till. The best option if your till supports it, because it links visits to a person without extra effort at the counter.
- A table or counter card with a QR code offering something small (a free pastry on the fifth visit, early notice of new bakes) for joining the list.
- Wi-Fi sign-up with a separate, unticked box for marketing messages.
- Email receipts, if your card machine offers them, with an opt-in at the same time.
Whatever the route, the permission wording matters, because messaging people who didn't agree to it is a legal problem in most places and a reputational one everywhere. A plain version:
Join our regulars' list: a short email most weeks with new bakes,
quiet-hour offers and events. Your fifth coffee is on us.
[ ] Yes, email me. You can unsubscribe any time.Only customers who ticked the box go into any of the AI jobs below. Keep the ones who didn't in your loyalty scheme for rewards, and leave them out of marketing.
Four AI jobs that raise visit frequency
1. Spotting who is drifting away
Every two weeks, export member visits and ask the AI to list members whose gap since their last visit is more than double their usual gap. That's a better signal than "hasn't been in a month", because a weekly regular who vanishes for three weeks has changed habit, while a monthly visitor hasn't. A prompt:
Attached: loyalty visits for the last 90 days (member number, visit
date, amount). For each member with 4+ visits, work out their usual
gap between visits and the days since their last visit. List members
where days since last visit is more than twice their usual gap.
Show member number, usual gap, days since last visit, and their
most common purchase if it's in the data.An illustrative extract of what comes back: "Member 1043: usual gap 3 days, last visit 11 days ago, usually buys oat flat white and a cheese scone." That's specific enough for a message that doesn't feel like a mass mailing. Before sending anything, check two or three of the flagged members against your till records; if a member number was mistyped at the counter, the AI will report a "lapse" that never happened.
2. Offers timed to your quiet hours
Regulars are made in the hours they can actually come. If your 2pm to 4pm slot is empty, an offer aimed at people who have visited at that time before is far more likely to create a habit than a general discount. The method, including how to read quiet periods from till data, is in how cafés use AI to fill quiet hours with targeted offers.
3. Reviews and your Google profile
People deciding between two cafés on the same street look at the Google profile, and regulars notice when you reply to their reviews. AI makes replying to every review a five-minute weekly job: paste in the week's reviews, ask for short replies in your voice that mention something specific from each review, and edit before posting. A weekly profile post (this week's bake, an event, new opening hours) keeps the listing current; writing Google Business Profile posts with AI every week has a routine for it.
4. Remembering regulars' orders
Nothing makes someone a regular faster than "the usual?" Staff know this for the top twenty customers and forget it for the next fifty. A simple regulars' sheet helps: first name or member number, usual order, and one friendly detail they've volunteered. AI can turn staff's scribbled notes into a tidy sheet, and a weekly export can suggest each member's usual order from their purchase history. Keep it light and limited to what customers would expect you to remember, like a drink order or a dog's name. Health details, family situations and anything overheard don't belong on it.
Find out what your regulars already love
Regulars rarely come back for the coffee alone. They come for a seat, a person, a routine. Your reviews say which, and AI is good at pulling the pattern out of a year of them. Paste in your Google reviews (they're public, but leave out reviewers' names anyway) and ask:
These are our café's reviews from the last 12 months. List the
specific things people praise more than once (not "great coffee" but
what exactly), the things that come up in complaints more than once,
and any mention of a routine (a regular time, day, seat or order).
Quote the review text for each point. Don't include anything you
can't quote.An illustrative result for the café in this tutorial: eleven reviews mention the window bench, seven mention the owner remembering orders, five mention it being "quiet enough to work" on weekday mornings, and four complain about slow service on Saturdays. That's a marketing brief in four lines. The weekly email started featuring the weekday-morning calm ("laptop-friendly until noon, Monday to Friday"), and the Saturday complaints went to the rota rather than the marketing plan.
Events that turn visits into a habit
A fixed weekly event is the most reliable habit-builder a small café has, because it gives people a reason to come on a particular day, every week. Typical examples: a parent-and-baby morning, a Thursday book swap, a run club that finishes at your door, a quiet hour for remote workers, or a monthly local makers' table. AI's job is the admin around them: the poster, the reminder message, the social post and the sign-up list.
An illustrative case: the café above started a Wednesday 10am parent-and-baby morning with a small discount on flat whites. AI drafted the poster and a two-line reminder email sent on Tuesday evenings to the 35 people who'd asked to hear about it. Eight weeks in, 12 to 18 parents came each week, most staying over an hour, and several became weekday regulars outside the event. The cost was the discount and some floor space for prams. The success came from the fixed slot; AI just made it easy to keep the reminders going every single week, which is where most café events quietly die.
An eight-week plan for a 30-seat café
An illustration with round numbers, for the café in the table above. It takes about 520 transactions a week, and loyalty members account for about 38% of them. The owner has two hours a week for marketing, mostly on Monday afternoons.
- Week 1 (3 hours): switch on the till's digital stamp card, print the counter card, and brief staff to mention the card at every order for a fortnight.
- Weeks 2-3 (1 hour a week): sign-ups grow by about 250. The owner writes a short welcome email with AI's help and sets it to send automatically.
- Week 4 (2 hours): first drifting check finds 60 members. Each gets a short personal-feeling email: "Haven't seen you for a bit. Your usual oat flat white is on us this week." Fourteen come back within seven days.
- Weeks 5-8 (1.5 hours a week): one weekly email (new bake, a quiet-hour offer on Tuesday and Wednesday afternoons, a book-swap evening), review replies every Monday, and a second drifting check in week 6.
Illustrative outcome at week 8: the share of transactions from loyalty members rose from 38% to 45%, active members' average visits went from 1.6 to 1.9 a week, and the Tuesday and Wednesday 2pm to 4pm slots each gained about 15 transactions a week. The fourteen free coffees in week 4 cost about $15 in stock (about $63 at menu prices) and paid for themselves in the first week. Time spent: about 14 hours across eight weeks, most of it in the first three.
What didn't work: a Friday "happy hour" discount pulled in people who would have come anyway, and the owner dropped it after two weeks. Offers work best where there's spare capacity; on a busy Friday they just cut the margin.
Prompts a café owner can copy
The drifting-customer message.
Write a 2-sentence email to a café customer who used to visit every
few days and hasn't been in for about 2 weeks. Their usual order is
[order]. Offer that drink free this week. Warm but not gushing, no
exclamation marks, no "we miss you", no mention of data or tracking.
Sign off as [owner's first name] from [café].An illustrative first result: "Hi [first name], it's been a little while since we've seen you at [café], and we hope all is well! As a thank-you for being such a loyal customer, your next oat flat white is on us." Two things to fix: the exclamation mark the prompt banned, and "loyal customer", which sounds like a template. The edited version: "Hi [first name], it's been a couple of weeks. Your oat flat white is on us any day this week, just show this email at the counter. [your name], [café]." Shorter, and it tells the customer exactly how to claim it.
The weekly email.
Here are this week's notes: [new bake, event, offer, anything else].
Turn them into a short email for our regulars' list: a subject line
under 45 characters, three short items, one line each, and a sign-off
from [name]. Don't add anything that isn't in my notes.The line "Don't add anything that isn't in my notes" earns its place. Without it, an illustrative draft described the café's scones as "award-winning" and its coffee as "ethically sourced from a single estate". Neither was true. Customers who notice a claim like that stop trusting the rest of the email.
Mistakes that push regulars away
- Messaging too often. A café that moved from weekly to three emails a week saw unsubscribes jump. Regulars already see you in person; the email is a nudge, not a feed.
- AI photos of food you don't sell. A generated cinnamon bun that looks nothing like yours sets up a small disappointment at the counter. Use phone photos of the real thing.
- Every message a discount. Customers learn to wait for the offer. Mix in new bakes, events and news, and keep discounts for quiet hours and win-backs.
- Staff who don't know about the offer. A customer shows the email and the barista has never heard of it. Put every live offer on a card by the till.
- Fake or AI-written reviews. Tempting when you're new, and a fast route to a warning on your Google profile. Ask real customers instead.
- Generic greetings. "Dear valued customer" tells people they're on a list. First names and their usual order tell them they're known.
Where AI isn't worth the effort for a café
Some AI marketing that suits other businesses rarely pays for a small café:
- A website chatbot. Café visitors want the menu, the hours and the address. A clear page and an up-to-date Google profile answer those faster than a chat window.
- Heavily personalised offers for a tiny list. With under a hundred members, you and your staff already know who they are. A notebook by the till does the job.
- Generated videos of drinks and food. Phone footage of your real barista pouring your real latte beats a generated clip, and costs nothing.
- Daily social posting. Two or three good posts a week, with real photos, do more than seven AI-written ones.
The AI work that pays is the unglamorous part: finding who has drifted, writing the nudge, replying to reviews and keeping the weekly email going.
Checking whether regulars actually grew
Pick three numbers and check them monthly against the same month before you started:
- Regulars count: members with two or more visits a week. This is the number the whole plan is meant to move.
- Win-back rate: of drifting members messaged, the share who visited within a week. Somewhere around one in five is a reasonable illustration; far lower means the message or the offer needs rethinking.
- Unsubscribe rate: if more than one or two people in a hundred leave the list after a typical email, you're sending too often or the content isn't useful.
Ask your staff too. They usually notice "the new regulars" before the spreadsheet does, and they'll tell you which offers customers actually mention at the counter. For a longer-term version of this plan, with rewards tiers and personalised offers, see building a customer loyalty programme with AI personalisation, and for bringing back the lapsed group specifically, winning back lapsed customers with AI-personalised emails goes into message sequences and timing.
Café owners also ask
Is it worth paying for a loyalty app for a small café?
Usually yes, if it connects to your till and gives you customer contact details with marketing permission, because that list is what every AI follow-up depends on. A paper stamp card rewards regulars but tells you nothing about who has stopped coming. Check the monthly fee against a simple target, such as the extra visits from ten customers a week.
How often should a café message its customer list?
For most small cafés, one short email or message a week at most, plus the occasional targeted nudge to lapsed customers. More than that and unsubscribes rise quickly. Make each message worth opening: a new bake, a quiet-hour offer, an event. If you have nothing new that week, skip it rather than sending filler.
Can AI write our specials board and menu descriptions?
Yes, and it's one of the quicker wins. Give it the actual ingredients and your tone, ask for short descriptions, and check every allergen and ingredient claim yourself before it goes on the board. AI will happily add details such as 'locally sourced' or 'homemade' that may not be true, so strike anything you can't stand behind.
Further reads
- Best AI Tools for Cafés and Coffee Shops, Grouped by Task — The tools behind each job in this plan, grouped by task.
- What AI Can and Cannot Do for an Independent Café — A wider view of where AI helps a café, and where it doesn't.
- Can a Café Use AI to Take Bookings and Answer Messages? — Handle table bookings and customer messages with AI.
- How to Get More Google Reviews With Automated Review Requests — Ask regulars for reviews automatically, within the rules.
- What Can AI Do for Your Email Marketing? 10 Practical Uses — More ways to use AI in the weekly email to regulars.
- How to Segment Your Email List With AI, Even a Small One — Split even a small list into regulars, occasionals and lapsed.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: general guidance from till and loyalty system help pages on customer exports; Google Business Profile help on posts and review replies.