I own a small bakery in Minneapolis. We do morning pastries, sourdough loaves on Tuesdays and Saturdays, custom cakes by order, and we run a small wholesale account with two local cafes. Annual revenue is somewhere in the mid-six-figures. For the first five years I was obsessed with new customers. I now think that was almost entirely the wrong thing to be obsessed with.
Eighteen months ago I built a loyalty system from scratch. Not a punch card. A real system, with tiers and a community and earned perks. As of last month, customers in the program represent 41% of revenue while making up about 22% of our total customer count. The average loyalty member spends 3.4 times what a non-member spends per year. Here is exactly how the system works and what I learned building it.
The wrong loyalty system I tried first
Before this one I had a 'buy 10 get 1 free' punch card. Customers loved getting them, lost them within two weeks, and approximately nobody ever redeemed one. We probably gave out 4,000 cards over two years and gave away maybe 200 free pastries. The punch card was theater, not loyalty.
It also was not building anything. A redeemed punch card was a one-time discount. It did not tell me who my best customers were. It did not give me a way to talk to them. It did not unlock anything special as a customer kept coming back. It was just a 10% discount applied randomly and inefficiently.
What I built instead
The new system has three tiers and is tracked through our point-of-sale by phone number. Customers join by giving their phone number at checkout. There is no app to download. There is no card to lose. They get a text the first time saying welcome and explaining the tiers.
Tier one: Regulars
Anyone who has spent $50 cumulative. About 60% of joiners hit this within a month. Regulars get one free morning pastry on their birthday and a $5 credit they can apply any visit. The text on their birthday goes out automatically and is the single highest-converting message we send. Birthday visits are usually our highest-ticket visits of the year for that customer because people bring friends.
Tier two: Locals
Anyone who has spent $300 cumulative. About 25% of Regulars reach Locals within a year. Locals get all of the above plus a 10% standing discount on Saturday sourdough (which moves them onto a slower-day visit so we even out our weekly demand) and early access to seasonal pre-orders (Thanksgiving pies, holiday cookie boxes, etc.). The early access is the killer perk. Last Thanksgiving we sold out our pie pre-orders in 36 hours, almost entirely to Locals.
Tier three: Family
Anyone who has spent $1,000 cumulative. About 30 customers right now. Family gets a private text thread with me (I batch responses every couple of days, not in real time, but they know I see them). They get to suggest a flavor for each month's special croissant. They get a free custom cake on a milestone of their choice once per year. They get invited to two members-only events per year, usually a Thursday evening tasting with the staff. They never wait in line.
The Family tier is the only marketing channel I have ever had that improved my mood instead of draining it. These are my favorite customers. The perks are mostly things I would have done for them anyway. Now they are formalized.
Why the tiers matter
A flat loyalty program rewards everyone the same and so motivates no one specifically. The tiers create a ladder. Customers can see where they are and what comes next. Regulars look at the Saturday sourdough discount and want it. Locals look at the private text thread and want in. The ladder is the engine.
I did not invent this. Hotels and airlines have known it forever. What I did was apply it to a 28-employee bakery in Minneapolis, which was not how this was supposed to work, and it has worked better than any other thing I have ever done in the business.
The numbers, after 18 months
- Loyalty members: 612 active phone numbers.
- Regulars: 458. Locals: 124. Family: 30.
- Loyalty members as a percentage of total customer count: ~22%.
- Loyalty members as a percentage of revenue: ~41%.
- Average non-member annual spend: $84.
- Average loyalty member annual spend: $287.
- Birthday text redemption rate: 71%.
- Pre-order sellouts to Locals: 4 out of last 4 seasonal pre-orders.
- Cost of the program (free pastries, $5 credits, etc.) as a percentage of revenue: ~3.2%.
What I did not expect
The Family members became my best market research. When I am thinking about a new product, I run it past them. They are honest and specific in a way that focus groups never are. They tell me things like 'I would buy that as a 4-pack but not as a 6-pack' or 'that should be $7, not $9, because the croissant next to it is $6'. That information is worth thousands of dollars in saved guesses.
The Family members also became referrers without me asking. About a third of my custom cake orders now come from Family members recommending us to friends having a birthday. The referrals are higher value than any other channel I have because Family members only recommend us to people they know will actually order, not generic acquaintances.
What I would do differently if I were starting today
- Start the program day one of opening, not year five. The customers I lost over years one through four because there was no path to deeper loyalty — I cannot get those back.
- Make tiers public from the start so customers know they are climbing a ladder.
- Build the birthday automation first. It is the single highest-ROI piece and works on its own from week one.
- Be careful about how often Family-tier events happen. Two a year is right. Four a year felt like work and stressed me out.
- Track everything by phone number, not email. Email opt-ins were 15%, phone opt-ins are over 80%.
A note about software
I duct-taped the first version of this together with our point-of-sale, a Google Sheet, and a Twilio account for the texts. It was clunky for about a year. I now run the whole thing through Social Perks, which lets me set the tiers, the perks, the automated messages, and the referral mechanics in one place. The 14-day free trial was enough time to know it was the system I had been duct-taping together for years. The work is the same. The work is just less work now.
If you are reading this and you have been telling yourself you need to spend on ads to grow, ask yourself first whether your existing customers have a reason to come back this month. If they do not, do that first. New customers cost five times what existing ones do. The math has been the same forever. Most of us, including me for years, ignore it.