How to automate order processing for an online store
While you have five orders a day, a store lives happily on Excel and a phone. The trouble starts at order number thirty: the manager physically can’t call everyone back, waybills get issued until midnight, and the customer writes «where is my order?» because nobody updated the status. Meanwhile the owner is convinced they need to hire another person. In reality, in most stores 60-70% of the actions around an order are the mechanical transfer of the same data from one system into another. That is exactly what order processing automation takes over. Below: where the time is actually lost, what can be solved technically, and how many hours it really gives back.
Where order processing actually loses you time
Take a single order and honestly map its journey. The request arrives by email or in the site admin panel — the manager opens it and copies the name, phone number, product, and address into a working spreadsheet. That’s two or three minutes, and the first place where errors are born: a swapped digit in the phone number, a missing letter in the surname, the wrong product variant. Next comes the confirmation call. The customer doesn’t pick up, the manager calls again, the customer calls back two hours later — another five to seven minutes of pure time and a broken focus. Then the waybill is created by hand in the Nova Poshta account: the same full name, phone, branch, weight, and declared value again — three or four more minutes and one more chance to make a mistake. At the end the manager manually sets the status in the spreadsheet, or, more likely, forgets to.
Do the math: 12-15 minutes of routine per order. With thirty orders a day, that’s six or seven hours — a full working day for a person who sells nothing and merely moves data around. With fifty orders on peak days the store simply cannot keep up, and orders start to go stale: the customer didn’t wait for the call and bought from a competitor. Tellingly, not one of these operations requires a human decision. Copying data from a form, sending a message, creating a waybill from a template, updating a status — this is pure mechanics that an algorithm performs in a second and without mistakes. A human is needed where there is something non-standard: the customer wants to swap a product, asks for unusual delivery, or complains. Everything else is a job for the system.
Auto-confirmation: the first step of order processing automation
The confirmation call is the most expensive habit of Ukrainian stores. It appeared back when website forms were crude and customers made mistakes in their data. Today, in 80-90% of cases there is nothing to confirm: the person deliberately chose the product, entered a branch, and clicked «buy». Instead of a call, right after checkout the system sends a message on Viber, Telegram, or SMS: the contents of the order, the total, the delivery method, the branch, and two buttons — «confirm» and «I want to change something». The customer replies in ten seconds at a moment convenient to them, not while they are driving or in a meeting. Those who tap «I want to change something» land in the manager’s queue — and those are exactly the people worth spending time on.
The effect is visible in the first week. The manager makes three to five calls instead of thirty. Confirmations arrive at night and on weekends, when the store is closed, so on Monday morning part of the orders is already ready to ship. A typical mistake at this step is sending a dry «Your order has been received» with no details. Such a message doesn’t remove the customer’s anxiety, and they call anyway. List the items, the total, and the branch in the text: the person sees the system understood them correctly, and the question disappears on its own. A second nuance: if the customer stays silent for two hours, a reminder should fire, and if they stay silent for a day — then a manager’s call is warranted. Automation doesn’t abolish the phone, it simply reserves it for the cases where it is genuinely needed.
Nova Poshta integration: waybills and tracking numbers without manual work
Nova Poshta has an open API, and this is the quickest win in the whole automation story. As soon as the customer confirms the order, the system generates the express waybill itself: it fills in the recipient, phone, city, and branch from the order card, calculates weight and declared value from the items, and picks the payer and delivery type according to the store’s rules. The waybill is created in a second, the number is written straight into the order and flies to the customer in the same message together with a tracking link. After that the system polls the parcel status every few hours and notifies the customer of key events: «shipped», «arrived at the branch», «waiting 3 days», «returning to sender». The manager doesn’t open the Nova Poshta account at all — they simply print the labels in a batch at the end of the day.
Nova Poshta integration closes two more painful issues. First, the directory of branches and parcel lockers is always current, because it is pulled from the API rather than from a hand-typed list three years old. The customer picks a branch right in the form on the site or in the bot, and it is guaranteed to exist. Second, control over «stuck» parcels. The system sees packages that have been sitting at the branch for five days and sends the customer a reminder itself, and the manager an alert in Telegram. This directly reduces the number of returns, each of which costs the store double shipping and frozen inventory. One more detail that is often forgotten: the waybill should carry a correct description of the contents and declared value from the start — this saves you in disputes over damaged parcels.
Stock sync — the foundation of online store automation
The worst scenario for a store is selling something you don’t have. The customer has paid, the manager calls to apologize, and reputation and money go out the window together. This happens when stock levels live in three places at once — on the site, on the marketplace, and in a warehouse spreadsheet — and are reconciled by hand once a day. The solution is one shared warehouse as a single source of truth plus two-way synchronization. Sold a unit on a marketplace — the stock level instantly drops on the site. Received a delivery at the warehouse — the item appears across all channels. The reservation is placed at the moment the order is created, not at the moment it ships, so two customers can’t buy the last unit at the same time. Technically this is synchronization through each platform’s API every few minutes, not exporting files by hand.
Along with stock, status transitions get automated too. An order moves along the path «new — confirmed — being packed — shipped — received — paid», and each transition triggers its own action: a message flies to the customer, a picking task to the warehouse, a payment record to accounting. The key rule: a status must be changed by an event, not by a person. The waybill was created — the order became «shipped» by itself. Nova Poshta reported «received» — the order closed by itself. When statuses are set by hand they always lag behind reality, and the owner looks at numbers that cannot be trusted. When the system sets them, the order board shows the true picture at any second, and the customer stops writing in your DMs asking where their parcel is.
Abandoned carts, returns, and the morning sales digest
Once basic order processing runs itself, resources free up for the things you never got around to. Abandoned carts: a person added an item, reached checkout, and vanished. An hour later the system sends an unobtrusive reminder with a link back to the cart, and a day later a second one, sometimes with a small discount. On average this recovers 8-15% of such carts — pure revenue from traffic you have already paid for. Returns are worth putting through a scenario too: the customer fills in a short form or answers a bot, receives instructions and a return waybill, and the manager sees an already structured request instead of spending twenty minutes figuring out what is being returned and why. As a bonus, statistics on return reasons accumulate — a direct hint about which product or description needs fixing.
The final touch: every morning at nine the owner gets a short digest in Telegram — how many orders came in yesterday, for what amount, the average check, how many were shipped, how many are stuck without confirmation, which products are running out. Thirty seconds of reading instead of an hour digging through reports. Together, all the steps described strip those same 12-15 minutes of routine from every order and, at 30-50 orders a day, give the store back roughly six to eight hours every day — effectively a manager’s salary you don’t have to pay. These are exactly the systems we build at Devlly: linking the website, the warehouse, Nova Poshta, and Telegram around a store’s specific processes. Start with the most expensive pain — usually confirmations and waybills — rather than trying to automate everything at once.