What I Learned Building a Repeat-Customer System for a 42-Year-Old Fabric Shop in Lajpat Nagar

My family has sold fabric from a 240-square-foot shop in Central Market, Lajpat Nagar, Delhi, since 1984. I joined the business in 2017 after a decade in corporate retail. I am not a developer. I know enough SQL to query my own stock sheet and enough no-code to break and un-break things. But the lessons on repeat customers are the same whether you are selling embroidered organza or SaaS seats, so I thought I would share.
Here is what three years of trying to systemise a repeat-customer motion taught me.

1. The friction is never where you think it is I spent six months in 2022 convinced our problem was discovery — people did not know we existed online. We built a website, we loaded a catalog, we ran meta ads. Traffic came. Conversion was low. I sat on the shop floor for a week watching returning customers. The friction was not discovery. It was the fact that a woman visiting from Noida for a wedding would see three thaans she loved, buy one, and have no way to find the other two eight weeks later for a cousin's function. We did not record what she had touched. We did not WhatsApp her when a similar shade came in. The entire "I saw it last time" conversation was starting from zero every visit. The fix was not technical. It was a 3-column notebook on the counter. Date. Phone number. Three keywords for what she touched. Every evening my cousin transcribed it into a Google Sheet. Six weeks later I had a pattern. Three months later I had a WhatsApp broadcast list segmented by fabric type with a 34% click-to-visit rate. Technology was downstream of the behavior change on the counter.
2. The customer tells you the metric that matters A new customer asked me once: "Didi, kitne functions ke liye enough hoga ye thaan?" ("Didi, how many functions will this thaan be enough for?") That one sentence rewrote our product pages. Not metres. Not GSM. Functions covered. Three functions, five looks, one buy — that framing started outperforming any specification-first copy we wrote. Our bounce rate on bridal category pages fell from 71% to 48% in a quarter after we switched the H1 and the opening block to function-count. If you are building a product for a non-technical buyer and your conversion is low, go read your tickets. The language of the metric is in them.
3. Repeat is a cadence, not a campaign In fabric retail, there is a season. Oct-Jan is wedding. Mar-May is festive wear and summer wedding. Jun-Aug is slow but boutique owners restock. We used to blast every list in October. Open rates were fine. Reply rates were low. We switched to a cadence match. Each segment hears from us when their natural buying moment arrives, not when our calendar says send. A sister-of-bride hears from us 6-8 weeks before the function date she registered when she bought last. A boutique owner hears from us on the 2nd and 4th Tuesday of the month when their payroll cycle is finished. A wedding planner hears from us the Monday after a client signs — because the client WhatsApps the planner the moment they book a date. Replies went up 2.4x. We did not change our offer. We changed the clock we were running on.
4. The physical shop is a research lab I know every indie hackers piece says "go to where your users are." For a fabric shop, where they are is my counter. Three mornings a week I now block myself from WhatsApp till 1pm and I just sell. I watch what hands go to first. I note what word comes out of a confused buyer's mouth. I write it on a sticky note. Every landing page tweak I have made that worked, started on a sticky note.
5. Small inventory, small claims Last year I was asked to add 400 SKUs to the online catalog from a wholesaler I did not know well. The upside was obvious — more surface area, more search terms, more organic traffic. I said no. The reason: I could not verify fabric quality on 400 thaans sitting in another warehouse, and a bad return would cost me a repeat customer forever. We stayed at 180 SKUs we touch. Our Trustpilot and Google review score is 4.7. The warehouse model is building a shop that competes with every other shop. The 180-SKU model is building a shop that competes with no one. Narrow is a moat.

If you run a small physical business and are trying to systemise repeat behavior, these five are what I would start with. Boring beats clever. The counter notebook outperformed the CRM for a year. The function-count H1 outperformed six months of SEO consulting. Match the cadence to the buyer, not the calendar.
If you want to look at how we actually arrange a wedding-season buy, our online stock is at Paras Gallery Fabrics — the same thaans sit on our shop floor in Lajpat Nagar, Delhi.
Happy to chat in comments. Would love to hear what cadence shifts worked for your business.
Sneh, Paras Gallery Fabrics, Lajpat Nagar, Delhi.