A product subscription is when a customer chooses once what to receive and how often, and from then on the orders create and pay for themselves. It sounds simple, but between “we want subscriptions” and the first automatic charge there are seven decisions worth making deliberately. Here they are, in the order they get made.
Step 1. Check whether you need this at all
Subscriptions work on products with a consumption cycle: they run out, and the customer knows it. Coffee, tea, pet food, water, nappies, cosmetics, vitamins, contact lenses, filters, household supplies. They do not work on furniture, electronics, clothing or gifts — there is no “next time” to schedule.
The practical test is your own order history. Take the customers from the last year and look at which products are bought a second and third time by the same people. If there are no such products at all, subscriptions are not for you, and the time is better spent on other ways to grow sales. If there are at least five to ten, keep going.
Step 2. Pick the launch products
Do not switch subscriptions on across the whole catalogue. Choose 5–10 items on three criteria:
- Repeat rate. The ones from step 1: they are already bought repeatedly without any subscription.
- Reliable stock. A product that is regularly out of stock will miss a delivery — and a missed delivery on a subscription costs you the customer, not one order.
- Margin that survives a discount. Subscriptions almost always carry a 5–15% discount. On a 20% margin, a 15% discount makes no sense.
The typical mistake is starting with the most expensive product. Start with the one bought most often: it builds statistics fastest.
Step 3. Set the frequencies and the discount
Frequencies. Two or three, no more. For coffee that is “every 2, 3 or 4 weeks”; for pet food, “every 2 weeks or monthly”. More choice does not raise conversion, it creates doubt. Make the default the one that matches real consumption: a 250 g bag of coffee is two to three weeks for one person.
Discount. Its job is not to “lure” but to compensate the customer for giving up flexibility. 10% is the working standard; 5% goes unnoticed, 20% eats the economics. An alternative that often works better is free shipping on subscription: it costs you the same but is perceived as a bigger benefit.
Do the maths up front: an 80 average order and a 10% discount is 8 off each order. A subscriber places 8–11 orders on average instead of 2–3 one-off ones. The discount pays for itself from the second order. We worked through exactly how subscriptions change LTV in a separate article.
Step 4. Connect recurring payments
This is the technical core. Without automatic charging you do not have a subscription, you have a reminder newsletter.
The mechanics: at the first payment the customer consents to recurring charges, and the payment provider returns a token to the store — an anonymised card identifier. The store initiates the following payments by token, without the customer’s involvement. Card details are never stored on your side, which removes the hardest part of PCI DSS.
In Ukraine, recurring charges are supported by LiqPay (PrivatBank) and Plata by Mono. Both enable the feature within the existing acquiring agreement — no new legal entity needed. What exactly they require, how tokenisation works and what to do about failed charges is covered in the article on recurring payments in Ukraine; how Neocarts implements it is on the recurring payments page.
What to have in place before launch:
- Failed charges. The card expired, there is no money on it, the bank declined. You need a retry a day or two later and an email asking the customer to update the card — otherwise the subscription dies quietly.
- Pre-charge notification. Payment providers require warning the customer a few days ahead. It is more than a requirement: an email saying “in 3 days we will charge 72 for your coffee” removes the complaints about “unexpected charges”.
Step 5. Decide how it looks in the store
Three things the customer has to see, without which the subscription will not work.
A widget on the product page. Next to the “Buy” button, a toggle “One-off / Subscribe” with the frequency and the discounted price. The customer has to see the benefit before the product goes into the cart.
A customer account. Pause during a holiday, skip one delivery, change the frequency, swap the product, update the card, track the shipment. Every one of these actions that is missing from the account turns into a support email — and into a cancellation if support does not reply immediately.
One-click cancellation. It sounds counter-intuitive, but a hard cancellation does not retain customers — it retains unhappy customers who then write reviews. Easy cancellation plus a “pause instead” button next to it: half of those who came to cancel choose the pause.
Technically it is either an in-house build (account, widget, charge scheduler, failure handling — realistically two to three months of a developer’s time) or a ready-made service. Neocarts connects to a store through the API or a platform module: OpenCart, WooCommerce, Magento, Shopify, PrestaShop. The widget, the customer account, charging through LiqPay and Plata by Mono and Nova Poshta delivery are already inside; the fee is 5% on subscription transactions, with no monthly charge.
Step 6. Set up delivery and stock
Subscriptions give you something one-off sales do not: you know in advance how much of what ships next week. Use it.
- Reservation. Stock on active subscriptions must be reserved before one-off buyers take it. Otherwise it will not be there on delivery day.
- Shipping day. Group subscription orders into one or two days a week — picking and shipping is cheaper, and the customer gets used to “coffee arrives on Thursday”.
- Carrier integration. Automatic waybill creation from the details the customer entered once. If the address has to be confirmed every time, it is not a subscription.
Step 7. Launch and measure
The first launch is on those 5–10 products from step 2, with no announcements or banners. Just the widget on the product page. A month in, look at four numbers:
| Metric | What it shows | A normal level at launch |
|---|---|---|
| Share of subscriptions among orders on those products | Whether the widget is noticed and trusted | 10–20% |
| Successful charges from month two | Whether payments and reminders work | > 90% |
| Cancellations in the first month | Whether the frequency is right and the first delivery not too big | < 15% |
| Pauses | Whether people use the account instead of cancelling | anything > 0 |
If the subscription share is under 5%, the problem is the widget or a discount nobody can see. If cancellations are over 25%, the default frequency does not match real consumption and customers receive the product before the previous one ran out. How to measure churn and what to do about it is a separate article.
After the first month with numbers, expand to the rest of the consumable products, add an announcement to emails and the home page. Not before: an announcement without a working process brings customers you will then lose along with their reviews.
What usually goes wrong
Launching on the whole catalogue at once. Nobody needs a subscription to a gift set, and its presence turns the widget into noise.
A discount instead of service. 20% off and no customer account — customers come for the price and leave after the first problem with a pause.
No handling of failed charges. Three months in, a third of subscriptions are “active” in the database and not paying.
A first delivery that is too big. A customer who did not finish the first bag before the second arrived cancels and takes away the impression that “a subscription is when they push things on you”.
None of this is about technology; it is about the decisions in steps 2, 3 and 5. The technology only executes what you decided.