Repeat purchase is the business: retention economics for D2C food brands

A ghee brand judging every campaign on the margin of a single jar will always conclude that acquisition is too expensive, because on one order it usually is. That single misreading is behind more stalled food brands than any bidding mistake.

Food is a consumable with a natural reorder rhythm. That changes what a first order is worth, what you can afford to pay for it, and which number you should be managing. It also creates a specific trap that looks like health right up until growth stops.

The first order is not supposed to pay for itself

The same spend reads completely differently once the second and third orders are counted, and those are largely won before the first jar even arrives: by the product doing what the ad promised, by a reason to remember the brand between purchases, and by a reorder prompt that lands near the point the jar runs out.

On Pure Whites, which sells farm-fresh ghee, honey, oils and mithai, a third of revenue now comes from repeat orders, alongside 550-plus new customers a month. That repeat share is what makes the rest durable, because revenue arriving without fresh acquisition cost pulls down the blended cost of every sale and raises the CAC the business can afford to pay. When 25% or more of revenue comes from repeat buyers at near-zero marginal cost, you can outbid competitors for new customers without being more efficient than them.

The trap: retention strength hiding an acquisition problem

My Pahadi Dukaan sells a Himalayan pantry range and had a loyal base that kept reordering unasked. That sounds like a healthy business until you notice it was the whole business. New customers barely trickled in, ROAS had flatlined, and the same tired creative was going to the same tired audience. A genuinely special brand was coasting at ₹12 lakhs a month.

Repeat buyers carrying a brand is a compliment and a trap. The compliment is obvious: the product is good enough that people come back. The trap is arithmetic. A revenue line made mostly of repeat orders looks stable and reports a flattering blended return, which hides that the top of the funnel has stopped working. Meanwhile prospecting is buying almost nothing, and the creative serving it has usually been running long enough that the intended audience has learned to scroll past it.

Three corrections did the work. Fresh creative formats broke the monotony, and traction returned almost immediately, which is diagnostic in itself: when new creative works quickly on a flat account, the problem was never the audience or the product, only the asset. Cohort-based segmentation meant first-time scrollers, repeat buyers and fence-sitters each got messaging built for their specific hesitation. And spend was pointed at hero products that actually close rather than spread across a catalogue.

Eight months later the brand was doing ₹1.2 crores a month, a 10× increase, with a steady month-on-month climb rather than a spike, and ROAS held as spend scaled. Retention flows ran underneath the whole time, so every new customer stacked on top of existing revenue instead of replacing someone who had quietly churned. That is why the line climbed steadily rather than sawing up and down.

Separate the numbers before you separate the budget

The fix starts in reporting, not in the ad account. Report new-customer acquisition on its own, with its own cost and its own return, and the true state of the funnel becomes visible at once.

• Track cost per new customer, not blended CAC. Blended is the number that lies.
• Split repeat revenue out of campaign reporting entirely, so warm conversions stop flattering cold prospecting.
• Report repeat rate by cohort and by month of first purchase, not as one lifetime average, because an average hides a base that is quietly ageing.
• Watch time to second order. It tells you when your reorder prompt should fire and whether it is firing too late.
• Watch how much of this month's revenue came from customers acquired more than six months ago. If that share keeps rising, you are harvesting.

A brand that cannot say what a first order costs does not know whether it is growing or harvesting.

Engineer the second order before the first one ships

Retention in food is mostly timing and memory. The buyer is not deciding whether your ghee is good, they decided that in the kitchen. They are deciding whether to reorder from you or pick up something in the weekly grocery run.

• Time the prompt to consumption, not to the calendar. A 500 ml jar in a family of four runs out on a schedule you can estimate from SKU size and average household use. Fire the reminder just before, not a month after.
• Make the reorder one tap. A saved cart, a repeat-order link in WhatsApp or a one-click reorder in the account area removes the friction that sends a buyer to a marketplace instead.
• Give a reason to be remembered between purchases. Pantry buying cycles are long, so the brand needs to be recalled rather than retargeted at forever. On Barosi, content and social gave the brand a voice worth following between orders, which is exactly what that gap is for.
• Do not pay to re-acquire people you already own. Segment existing buyers out of prospecting audiences and speak to them with their own message and offer.

Order value is the other half of retention economics

Repeat rate and basket size do the same job from different ends: both raise what a customer is worth so that acquisition can cost more without hurting.

Barosi's monthly revenue went from ₹25,000 to ₹21 lakhs, and bundles and smarter pricing were a deliberate part of that: they raised the value of every order so growth did not depend on discounting the brand into a commodity. Conversion on site went from 0.8% to 6.8% over the same period and CAC halved, which together mean the same media budget bought far more orders.

Pushti Organics shows the sequencing. The brief carried a hard floor of minimum 3.5× ROAS, so the funnel was rebuilt from the bottom up: remarketing first, because the cheapest conversions in any account belong to people who already visited and did not buy, then reactivating old buyers, then growing order value with bundles, which lifts revenue per order without another rupee of acquisition spend. Only then was prospecting scaled. The result was 570% growth in four months with ROAS held at 3.5× to 4×, and founder-led video cut CAC by 28% while lifting repeat purchases at the same time.

Starting at the bottom of the funnel is counter-intuitive and correct. Prospecting is the most expensive traffic in the account and the slowest to teach the platform anything. Bottom-of-funnel work produces conversions at the lowest cost available and gives the algorithm real purchase signal to learn from, which is what prospecting is then judged against.

Subscriptions: worth trying, but know the plumbing

Subscriptions look like the obvious answer for a consumable, and for some brands they are. Understand the payment rails before you build the offer around them.

The RBI's Digital Payments E-mandate Framework, 2026, issued on 21 April 2026, consolidates the earlier circulars on recurring transactions across cards, UPI and prepaid instruments. The parts that shape your customer experience:

• Registration of the mandate requires additional factor authentication, and so does the first transaction.
• The issuer must send a pre-transaction notification at least 24 hours before every debit, carrying the merchant name, amount, date and time, the e-mandate reference and the reason for the debit, along with a facility to opt out of that transaction or the mandate.
• A post-transaction notification follows every successful debit.
• Recurring transactions may be authorised without additional factor authentication up to ₹15,000 per transaction. Above that, authentication applies each time.
• No charges may be levied on the customer for using the e-mandate facility.

The practical consequence is that your churn moment is that 24-hour notification. Every cycle, the customer is reminded, given the amount and offered an exit. A subscription that is genuinely convenient survives that. One that exists to trap people does not, and you will find out monthly. Keep the cycle value comfortably inside the ₹15,000 threshold so buyers are not asked to authenticate every time, and time the shipment so the product arrives before the previous one runs out.

What to fix, in order

• Split new-customer cost and return out of blended reporting. Everything else depends on seeing this honestly.
• If prospecting has stalled, change the creative before you change the audience. Fatigue is the most common cause and the fastest to test.
• Build cohorts. First-time visitors, one-time buyers, repeat buyers and lapsed buyers all need different sentences.
• Set the reorder prompt against consumption timing, and make reordering a single tap.
• Raise basket value with bundles rather than lowering price with discounts.
• Only then scale prospecting, judged against a baseline the bottom of the funnel has already set.

A loyal base is a foundation, not a business. When repeat orders carry the revenue, they also hide the fact that nobody new is arriving.

Where this becomes work rather than a diagnosis, it is performance marketing and strategy together, with CRO and web keeping the reorder path short. More of this category sits under food and beverage and nutrition.