How Do You Open an Online Store? From Zero to First Sales
How do you open an online store? Platform choice, payment collection, delivery and the first sale — a complete from-scratch roadmap for Azerbaijan.

To open an online store, first verify the product's unit economics, then set up business and tax registration, the sales terms, the platform, payment, delivery and measurement in one order flow.
A store is not just a domain and a pretty design. It is an operating system that works from the moment a customer finds the product to the last exception where money is refunded. If the storefront is ready but the stock is wrong and the returns process has no owner, launching just moves the problem onto the internet.
This guide puts the decisions in the right order. The goal is not listing hundreds of products on day one, but building profitable, legally compliant, repeatable orders with a small number of products.
A short decision framework
- Before the platform, confirm to whom, which product and at what margin you will sell.
- Include in the price not only the product but packaging, payment, delivery, returns and customer acquisition costs.
- Show the seller information, TIN, price, tax, delivery and withdrawal terms clearly before the order.
- Test not only the ordinary order but also failed payments, stock mismatches, cancellations and refunds.
- Make the launch decision by the completed order's contribution margin and repeat purchases, not by traffic.
How do you choose a business model for an online store?
The business model determines who owns the product, who carries the stock risk and who does the work after the order. Your own stock, pre-orders, direct shipping from a supplier and marketplaces create different cash flows under the same name "online selling."
| Model | Advantage | Main risk | Starting test |
|---|---|---|---|
| Your own stock | Control over speed and quality | Capital tied up in unsold goods | One category, small stock |
| Pre-orders | Seeing demand before stocking | Delays and lost trust | A clear delivery date |
| Direct shipping from a supplier | Low initial stock burden | Stock, quality and returns control | A trial order sent to yourself |
| Marketplace | A ready audience and payment flow | Commission, rules and platform dependence | Margin and data export |
I explain the differences between e-commerce models separately. Here a simple rule is enough for the choice: do not turn an obligation you cannot supervise daily into a sales promise. A supplier's "we usually deliver in two days" is not a date to give a customer.
How should the product be tested?
The first product must pass three tests: a real need, a healthy margin and manageable operations. Search volume can signal demand, but it is not proof of purchase. Five competitors selling the same product does not validate the market either; sometimes everyone repeats the same weak assumption.
Plan a limited pilot of 10–20 real orders. That number is not a market norm; it is a working threshold to keep stock and ad risk limited. In the pilot, write down which questions customers ask, at which price they hesitate, how they react to delivery and why they return items.
If you choose the dropshipping model, do not settle for product photos and the supplier's catalogue. Order the product to your own address, open the packaging, see the actual delivery time and walk the defective item's return path yourself. This test is cheaper than advertising.
How are the unit economics calculated?
Unit economics shows how much money one completed order leaves in the business. Subtracting only the product's purchase price from the sale price is not the margin; packaging, payment fees, delivery subsidies, a returns reserve and customer acquisition cost also belong to that order.
Contribution margin = sale price − cost of goods − all variable costs per order.
| Illustrative order | Amount | Role in the calculation |
|---|---|---|
| Sale price | 55.00 AZN | Actual revenue per tax and campaign rules |
| Cost of goods | 22.00 AZN | Purchase and the import/supply load per item |
| Packaging | 2.00 AZN | Box, bag, label and protective material |
| Payment cost | 1.10 AZN | A 2 percent assumption for the example |
| Delivery subsidy | 4.00 AZN | The part the customer does not pay |
| Cancellation and returns reserve | 3.00 AZN | An average share updated as facts accrue |
| Customer acquisition cost | 9.00 AZN | Ad spend / first orders |
| Contribution margin | 13.90 AZN | 55 − 22 − 2 − 1.10 − 4 − 3 − 9 |
These figures are not pricing advice. The important part of the calculation is this: 55 AZN of turnover does not appear as profit. From the 13.90 AZN, the monthly platform, labour, warehouse, photography, accounting and other fixed costs still have to be paid.
How is the starting budget written?
Build the budget with three envelopes, not one number. The first envelope is for setup: the domain, platform, integrations, legal texts, product photos and initial data. The second is for 30 days of operations: stock, packaging, delivery, support and advertising. The third is for mistakes: returns, failed payments, losses and rework.
Next to every amount write "one-off," "per order" or "monthly." Do not calculate ROI without comparing the same period. Mixing a five-thousand-manat first month with a two-hundred-manat steady month misjudges both the store and the expectations.
What are the TIN, tax and sales rules in Azerbaijan?
Online selling does not fall outside business and consumer obligations just because there is no physical storefront. The form of activity, the tax regime, record-keeping and receipt rules must be confirmed with an accountant or tax adviser to match the product and turnover.
According to the State Tax Service's tax guide, an individual's tax registration can be completed electronically; the application also indicates the taxation system to be applied. "It's an Instagram page, not a company" does not automatically turn continuous commercial activity into personal selling.
Articles 5 and 8 of the Law "On Electronic Commerce" give a practical checklist for the store. Before ordering, the buyer must clearly see:
- the seller's name, address, contact details and TIN;
- the product's main characteristics and its price including taxes;
- delivery costs, payment and performance terms;
- the right of withdrawal, the price's validity period and the contract stages;
- the ability to see and correct errors in the order;
- immediate order confirmation, a complaints address, warranty and after-sales service.
Article 9.1 of the law gives the buyer the right, subject to certain exceptions, to withdraw from performance of the contract within 7 business days without giving a reason and without penalty. The separate consumer rights law provides a right to exchange non-food goods of proper quality within 14 days under certain conditions. Do not write these as one rule; verify the product exceptions and application conditions with a lawyer.
Which legal pages must the store have?
The minimum package includes seller information, terms of use/sale, payment and delivery rules, a cancellation-and-returns policy, a privacy notice, warranty terms and a contact/complaints channel. An "I agree to the terms" checkbox does not create consent to unread text; the rules must be in plain language, saveable and accessible before the order.
If names, phones, addresses, order history and behavioural data are collected, write the purpose and the legal basis or consent, minimum volume, access, retention, transfer and deletion rules. The Law "On Personal Data" makes the choice of foreign platforms and analytics tools part of the data map too.
How do you choose an online store platform?
Platform choice is not a choice of code and design templates; it is a decision about how the team will manage orders. Test the same five scenarios in two candidate systems: an ordinary order, an out-of-stock product, a failed payment, a partial refund and a data export.
| Path | Fitting case | Hidden burden | Exit test |
|---|---|---|---|
| Managed SaaS | A small team, a fast start | Monthly plans, extra apps and rule dependence | Product, order and customer export |
| An open-source CMS plugin | Technical control and flexibility | Hosting, updates, security and compatibility | A full database and media backup |
| A custom-built system | Processes that fit no standard platform | Upfront cost, a permanent team and technical debt | Documentation, code, data and handover |
| Marketplaces and social channels | A fast demand test | Commission, audience and account risk | The boundary of customer and order history |
Do not compare a Shopify setup only by design and the monthly price; currency, local payments, delivery, tax, language, export and app costs must be considered together. Selling via Instagram can test demand, but tying search, orders, payment and customer data all to a single message thread gets expensive as you grow.
The same question applies to selling on Trendyol and other marketplaces: what stays in the business after the ready traffic? Before the contract, see the margin after commission and advertising, the returns obligations, access to order history if the account is suspended, and the path to your own store.
What is the technical minimum?
Your own domain, HTTPS, mobile speed, backups, multi-factor login, role-based admin permissions, an order log and a test environment are the minimum. A design template update can break the order flow; "the site loads" is not a technical acceptance test.
The platform must unify all order data — product variant, payment state, delivery code, discount, tax information and returns — under one order number. Every discrepancy fixed in a separate spreadsheet creates the risk of later sending the customer a different status.
How are the catalogue and product pages prepared?
The catalogue must quickly answer the buyer's "is this right for me?" Category names should follow the customer's selection logic, not your warehouse language; filters should show real differences like colour, size, material, compatibility and use case.
Prepare a minimum data model for each product:
- a unique product and variant code (SKU);
- a clear name, real photos and size/scale;
- material, composition, compatibility and usage limits;
- the actual price, stock and delivery time;
- box contents, warranty, care and return terms;
- differences between variants and answers that aid the choice.
A product description that sells is not a pile of adjectives. It shows who the product suits, what job it does, when it is not the right fit and which fact backs each claim. AI can provide a draft, but it may not invent the material, dimensions, warranty or safety facts.
How does Google find the product?
Google's e-commerce site structure guidance recommends reaching categories, subcategories and products through ordinary links from the menu. Because Googlebot does not type queries into your site's search box, products visible only through filters and search can stay out of the index.
On purchasable product pages, the Product and Offer structured data must state the same facts as the product name, image, price, currency and offer. Google's merchant listing document explains how price, stock, delivery and returns data qualify for rich results; markup is not used to "fix" information that is absent from or stale on the page.
How is online payment set up?
Online payment does not end with "money was charged to the card." The order amount, the payment attempt, the bank's response, the success notification, cancellation, refund and accounting reconciliation must be tracked under the same transaction identity.
The Central Bank of Azerbaijan's register of payment institutions shows the licensed entities and their permitted activities. A provider's logo and familiar name are not enough; check the legal entity you contract with, its licensed activity and its current register status.
Compare the online payment options with these scenarios:
- After a successful payment, the order is created only once.
- Even if the customer closes the page, the server notification recognises the payment correctly.
- If the same notification arrives twice, the order and payment are not written twice.
- Failed and pending payments are correctly separated from stock.
- Full and partial refunds appear identically in the order, payment and reports.
- At day's end, the provider statement reconciles with the store and accounting records.
Instead of storing card data on your own server, use a provider-managed or tokenised flow. The PCI Security Standards Council's merchant resources explain how the level of contact with card data changes the security obligations. Confirm the technical boundary in writing with the provider and a security specialist.
How are delivery, returns and support set up?
Delivery is not a courier choice; it is tracking the order from the store to the customer's acceptance, and onward to a possible return. City, district, weight, dimensions, temperature, breakage risk and payment method may not all work at one tariff.
Beyond the price table, test the courier and logistics choice with these events: the address is incomplete, the buyer does not answer, the parcel is damaged, delivery is late, cash-on-delivery is refused and the product comes back. When the courier writes "not delivered," the owner of the next step must be known inside the store.
| Status | What the customer sees | Internal owner | Guardrail metric |
|---|---|---|---|
| Order received | Product, amount, address and number | The store system | Duplicate orders |
| Being prepared | The expected handover time | The warehouse | Stock mismatches |
| Handed to the courier | Tracking and contact | Logistics | Late handovers |
| Delivered | Acceptance confirmation | The courier and support | Wrong items |
| Return opened | Conditions, address and timeframe | Support | Resolution time |
| Money refunded | The amount and expected bank time | Finance | Reconciliation with the order |
The support channel must be written down with working hours, response times and an escalation owner. WhatsApp, phone, email and Instagram must not create separate truths for the same order. Wherever the customer writes from, one history must open under the order number.
How do you bring customers to an online store?
A customer acquisition plan is not a channel list; it is a set of trials where each channel has an intent, a cost and a metric. Search captures existing demand, social content creates interest, ads give a fast test, and email and messaging continue the relationship with existing customers.
In e-commerce marketing, do not open every channel at once in the first month. Choose one high-intent channel, one re-engagement channel and clear metrics. If the product page is weak, more traffic just creates more expensive hesitation.
The minimum structure for SEO
- a unique SEO title, meta description and main heading;
- ordinary server-rendered links leading to the product;
- indexable category and product URLs with correct canonicals;
- an XML sitemap and Search Console verification;
- sized files, alt text and stable dimensions for product images;
Product/Offerdata matching stock, price and variants;- indexing and crawl rules for filter URLs;
- visible pages for returns, delivery and seller information.
Multiplying a hundred similar product descriptions with AI can create an index count, but no selection value. In Google's current approach favouring people-first content, the key question is not which tool wrote the text but whether it gives the user original, reliable value. Long copy without dimensions, materials, real photos, compatibility and usage limits is weaker than a short, precise card.
What should an online store measure?
Measurement should start with order count and end with order quality. If turnover grows while contribution margin falls and cancellations and returns rise, the store does more work but does not become a better business.
| Metric | Formula | What it changes |
|---|---|---|
| Product view to cart | Add-to-carts / product views | The offer and product page |
| Checkout entry | Checkouts started / carts | Price, delivery and trust |
| Checkout completion | Paid orders / checkouts | Form and payment errors |
| Sales conversion | Completed orders / sessions | Traffic and the whole flow |
| Contribution margin | Revenue − variable costs | Price and channel boundaries |
| Cancellations and returns | Cancellations/returns / orders | Product, promises and operations |
| Repeat purchase | Repeat buyers / eligible buyers | Product and experience |
In an illustrative flow, 1,000 sessions produce 80 carts, 40 checkouts and 24 completed orders: cart rate 8 percent, checkout entry 50 percent, checkout completion 60 percent, overall sales conversion 2.4 percent. These figures are not market benchmarks; they show how the sales funnel is calculated.
Do not automatically equate the problem with the biggest percentage drop. Falling from 80 carts to 40 checkouts could be the delivery price appearing late — but an analytics event being double-counted is also possible. Confirm the cause with screen recordings, order records, payment responses and five real users.
A 30-day launch plan for the online store
The 30 days are not for finishing a large catalogue but for seeing one category's core order flow and error scenarios. Keeping the ad date fixed while product and legal preparation is incomplete is calendar management, not quality.
| Period | Work to do | Acceptance evidence |
|---|---|---|
| Days 1–5 | Customer, product, model and unit economics | A positive contribution margin and a pilot threshold |
| Days 6–10 | Registration, tax, sales, privacy and returns rules | An adviser's review and visible pages |
| Days 11–16 | Platform, domain, catalogue and product data | A mobile flow from product to checkout |
| Days 17–21 | Payment, stock, delivery and notification integration | Success, failure and duplicate notification tests |
| Days 22–25 | Returns, refunds, roles, backups and security | A test order runs forwards and backwards |
| Days 26–28 | Analytics, SEO, the sitemap and search appearance | An event and index checklist |
| Days 29–30 | A limited launch, real orders and a decision | Margin, error, support and correction records |
The final pre-launch check
- The seller's name, address, contact details and TIN are visible.
- The price, tax, delivery and final amount are clear before the order.
- The product's stock, dimensions, material, photos and variants match the facts.
- A failed payment does not become an order; an out-of-stock product does not become a sale.
- Order confirmation, courier status and the support contact work.
- Cancellation, exchange, returns and refunds have been tested.
- Admin access has multi-factor authentication, role separation, logs and backups.
- Mobile speed, analytics, the sitemap, structured data and indexing have been checked.
- The first 30 days' ad, stock and error budget is set aside separately.
- The continue, fix and stop thresholds are written in advance.
Frequently asked questions about opening an online store
How much budget does an online store need?
There is no universal amount. The budget depends on product stock, the platform, payment integration, photos and copy, legal preparation, packaging, delivery, advertising and a returns reserve. Separate one-off setup from monthly and per-order costs; make the launch decision by the contribution margin left per order, not by total turnover.
Do you need a TIN (VÖEN) for online selling?
Continuous income-seeking online selling can create business and tax obligations. The legal form, TIN, tax regime, record-keeping and receipt requirements must be clarified for the product and activity. Check the State Tax Service's current guide and get a written opinion from an accountant on your specific activity before accepting the first order.
Is a ready-made platform or a custom site better?
For a small team and a standard order flow, a ready platform can be faster. As custom processes, integrations and data control grow, an open-source or custom solution may be justified. Test two candidates on the same order, returns, reporting and export scenarios; then compare the full monthly cost, not just the look.
How many products is it right to start with?
Many products guarantee neither trust nor sales. Starting with one clear category and a small number of products whose operations you can manage is more measurable. Do not grow the catalogue before each product's photos, variants, stock, margin and returns data are complete; first learn from real orders with limited stock and expand only the category that works.
How long does it take to open an online store?
A simple storefront can be built in a few days, but a working store's timeline depends on product data, registration, payment, delivery and legal preparation. A limited 30-day plan is a realistic start, but it counts as a launch only when the ordinary order, a declined payment, a stock mismatch, a cancellation and a product return have all been tested.
Conclusion: open an order system, not a website
To open an online store, first make one unit of the sale work. The customer must find the right product, see the final amount, complete the payment, track the order and get a clear way back when there is a problem. And the business must know how much money remains from that order.
The first version can be small. A version without rules, owners and measurements is not small — it is just uncontrolled. Prove the whole path with one category; then grow the catalogue and the advertising.
Sources
- State Tax Service of the Republic of Azerbaijan: tax guide
- Law of the Republic of Azerbaijan "On Electronic Commerce"
- Law of the Republic of Azerbaijan "On Protection of Consumer Rights"
- Law of the Republic of Azerbaijan "On Personal Data"
- Central Bank of Azerbaijan: register of licensed payment institutions
- Google Search Central: e-commerce site structure
- Google Search Central: Product and Offer merchant listing data
- PCI Security Standards Council: payment security resources for merchants
I'm Anar Rustamli - a strategist, entrepreneur, and AI adoption leader working at the edge of growth, technology, and human thinking. Since 2016, my work has focused on helping businesses evolve in a rapidly changing digital landscape. I design growth systems, AI-powered workflows, and strategic frameworks that align performance with purpose. I believe real growth happens when strategy, data, and human insight work together - and my mission is to help businesses adopt AI in a way that strengthens both their results and their identity.

