Online Store Metrics in Türkiye: Conversion, Returns, Profit and Growth

سفير تركيا · 2026-10-12T00:04:16+0300

Turn store figures into clear weekly decisions
Consistent definitions, a suitable period and profit after cancellations and returns

A good dashboard links numbers to decisions the team can carry out.

Store traffic and reported sales may rise while cash availability falls or order losses increase. Sometimes performance is weak; sometimes metric definitions are mixed together: an order created but unpaid, a refund recorded in another month, or revenue including tax and shipping compared with costs excluding them. A useful dashboard starts with a short metric dictionary and links order records to payments and fulfillment. Choose a limited set of metrics you can explain and act on, rather than placing everything the tools display on one screen. All example figures are hypothetical Turkish-lira amounts; they are neither market averages nor targets for every store.

Fix the period, source and definition

Show the reporting period, time zone, currency and data update time at the top. Establish whether the dashboard shows orders created during the month, amounts received in your account during it, or orders whose final outcome is known. These are different views that can all be correct simultaneously. Maintain order-creation, payment, shipping, delivery and refund dates so you can move between views without changing a metric’s meaning.

Use the order system as the reference for order counts and statuses, the payment provider for money movements and the carrier for delivery statuses. Traffic tools help explain behavior and attribution but cannot replace financial reconciliation. Where a source is missing, mark the metric as preliminary rather than silently filling the gap with an estimate. Give each metric a name, calculation and correction owner. Identify the difference when comparing a promotional week with a normal week, and provide explanatory detail when comparing a channel selling large baskets with one selling single items.

Create a short metric dictionary

A metric’s definition provides half its value. Conversion may divide orders by sessions, customers or product-page visits; these are not interchangeable percentages. Average order value may be before or after discounts and include or exclude tax and shipping. Choose a clear internal definition, match the tool’s report when comparing and record the date of changes.

In WooCommerce Analytics, average order value relates to net sales divided by included orders, with results affected by order statuses and period. Refunds may include money returned without a physical product return. Therefore, refund value alone cannot establish the percentage of items returned. Create separate operational metrics with names that prevent confusion. Display the numerator and denominator beside percentages for small volumes; ten cases out of one hundred gives clearer context than ten percent alone.

Internal metric	Suggested definition	What to clarify
Paid-order conversion	Paid orders ÷ eligible sessions	Session source and matching period
Cancellation after payment	Orders canceled before shipping ÷ paid orders	Keep separate from payment failure
Order return rate	Orders returned after shipping ÷ shipped orders	Partial or full returns and a sufficient observation window
Average value at payment	Product value after discounts ÷ paid orders	Tax and shipping excluded here
Order contribution	Net revenue minus variable costs	Before fixed expenses

Separate order stages before calculating conversion

Start with an evidence-based funnel: product view, add to cart, checkout started, order created, confirmed payment, shipping and delivery. A customer may start checkout several times or try another card; attempts are not buyers. Use unique orders and exclude test orders and duplicate events according to your definition.

Google Analytics e-commerce events require setup and verification, including purchase, refund, transaction ID and currency. A ready-made dashboard does not prove that events arrive once with correct values. Test a known order and partial refund and compare them with internal records and the tool. Privacy settings and blocking tools affect measurement coverage, so explain coverage differences rather than treating every discrepancy as lost sales. If sessions fall while orders remain stable, examine measurement and channel mix before declaring a layout change successful.

Do not substitute purchase-event counts for unique orders without checking transaction IDs, duplicates and payment status.

Read a hypothetical dashboard without mixing measures

Assume the month recorded ten thousand eligible sessions and two hundred and fifty created orders. Two hundred and twenty were paid; twenty of those were then canceled before shipping and refunded. Two hundred were shipped, and twenty later returned in full. Assume this cohort’s outcomes are complete so returns can be measured on a mature basis. The thirty unpaid created orders are excluded from cancellation after payment in this example.

Paid-order conversion is two hundred and twenty divided by ten thousand, or two point two percent. Cancellation after payment is approximately nine point one percent, and returns are ten percent of shipped orders. One hundred and eighty orders remain after returns, representing one point eight percent of sessions. The last percentage describes the cohort’s final outcome; it is not a hidden replacement definition of conversion. Show both stages to distinguish payment improvements from product and delivery improvements. Do not infer a specific cause from a figure without reviewing cases.

Hypothetical cohort stage	Count	Related calculation
Eligible sessions	10000	Consistent conversion base
Created orders	250	Includes 30 unpaid orders
Paid orders	220	220÷10000=2.2%
Paid orders canceled before shipping	20	20÷220≈9.1%
Shipped orders	200	220−20
Orders returned in full	20	20÷200=10%
Orders remaining after returns	180	200−20

Order creation and payment are separate events that must be distinguished.

Measure returns in a way suited to the product

Use both order and item measures for multi-product baskets. An order returning one of five items counts as an order with a return, but its item return percentage differs. Record reasons such as sizing, description mismatch, defect, delay or change of mind, retaining the original complaint where needed. Do not let employees choose ‘other’ for every case; that hides clear improvement opportunities.

Measure cohorts whose delivery dates allow enough time for returns. Comparing this week’s undelivered orders with last month’s completed orders makes returns appear artificially low. Separately, track refund value by refund date for cash management even when it concerns an earlier-period order. Record resale suitability and shipping, inspection and discount costs. Return rates can remain stable while losses rise because a more expensive product is damaged. Link the reason to product code, size and channel, then test a specific improvement such as an additional measurement or detail photograph.

Separate revenue, settlement and average order value

In the example, assume product value of one thousand lira per paid order after discount and before tax and shipping. Paid-order product value is two hundred and twenty thousand. Refunds for twenty orders before shipping and twenty afterward reduce it by forty thousand, leaving net product value of one hundred and eighty thousand. Average value at payment is one thousand, while dividing net value by all paid orders gives approximately eight hundred and eighteen lira. Both are useful when named precisely.

The payment provider’s transfer is not total revenue; it may be net of commission, holds, refunds or withholding depending on the channel and case. Keep gross sales and settlement components separately. Where goods have different tax rates, use tax-exclusive revenue and expenses on a consistent accounting basis with the accountant. Customer-paid shipping can be a separate revenue line and carrier costs a separate expense line. Do not hide one merely because both have similar names.

Calculate contribution that supports decisions

From net revenue, deduct the cohort’s variable costs: goods consumed or lost, packaging, outbound and return shipping, collection and advertising allocated to the trial. The example assumes recorded goods costs include the necessary inventory and return effects, while additional damage is separate and not duplicated within them. Adapt this structure to your accountant’s records to avoid counting a loss twice.

The hypothetical figures produce contribution of forty thousand lira, or approximately twenty-two point two percent of net revenue. This funds rent, salaries and subscriptions; it does not equal net profit. With thirty thousand of fixed expenses allocated to the period, an estimated operating result of ten thousand remains before income or corporate tax and other omitted items. State these limits above the result so the example does not become a financial forecast. Compare contribution by channel and product: revenue growth in a weak product may strain cash and fulfillment capacity.

Hypothetical item excluding tax	Amount in lira	Treatment
Net product value	180000	After refunding 40000
Allocated cost of goods	100000	After inventory reconciliation
Shipping, packaging and returns	12000	Actual cohort cost
Collection and fees	6000	Including cancellation effects
Related advertising	18000	Cost for the specified period
Additional damage without duplication	4000	Separate documented loss
Cohort contribution	40000	180000−140000

Monitor repeat purchasing by cohort

Identify customers consistently and lawfully, then form a cohort of first-time buyers in a particular month. After a defined period, calculate how many returned to place another paid order under your definition. Do not divide this month’s returning customers by this month’s new customers and call it cohort retention: returners may come from different months. Keep that description separate from the proportion of returning customers among the period’s buyers.

The appropriate period depends on product consumption. An item bought once for years differs from a supply replenished within weeks. Monitor the profitability of subsequent purchases too, because a large discount can raise repeat buying while reducing contribution. Respect data purposes and messaging permissions; internal analysis does not provide unlimited marketing permission. If customer identity is inconsistent between a marketplace and the store, explain linking limits rather than guessing matches. Add service complaints and reasons for not returning to help explain the metric rather than relying on it alone.

Fulfillment and return data complete the picture of sales and collections.

Read attribution and advertising within their limits

Advertising, traffic tools and marketplaces may attribute the same order to different channels because of attribution rules, lookback windows and purchase timing. Do not add conversions from different reports to obtain total orders. Use unique order records for the count and channel reports to understand contributions under their own rules. Record changes to measurement windows or settings before comparing periods.

Return on ad spend compares attributed revenue with expenditure; it does not establish profit after products, shipping and returns. Start with a trial budget cap and an acquisition cost that order contribution can realistically support. Distinguish a new customer’s first order from repeat orders, and sales before refunds from sales afterward. If a campaign increases orders alongside a seasonal offer, do not attribute the entire increase to advertising without an appropriate test or comparison. Scaling also needs inventory and fulfillment capacity; a campaign profitable on paper may overwhelm the team if delays and cancellations rise.

⁦⁩ Link every signal to a testable action

Choose only three decisions in the weekly meeting, each with an owner and review date. When conversion falls, locate the funnel decline and examine actual cases: payment problems, unclear shipping or unavailable products? Do not change every store page in one day. When contribution rises, verify data completeness and inclusion of the return window before raising spend.

Set internal thresholds based on your history and capacity, rather than presenting general percentages as Turkish market rules. Alerts may use critical-case counts instead of percentages, especially for a small store. Keep ‘continue monitoring’ as an option when evidence is insufficient, and ‘pause’ when stock is missing or collection fails. After a change, monitor the intended metric and side effects: broad discounts may raise conversion while weakening profit, and expensive faster shipping may improve service while reducing contribution.

Signal	First check	Change to test
Low payment completion	Actual errors and failures	Specific payment fix
High cancellation	Stockouts and preshipping delays	Inventory and preparation-time control
Sizing returns	Reasons by SKU and size	Clearer measurements and photographs
Higher revenue, lower contribution	Discounts, shipping and advertising	Correct the offer or cost
Large reporting discrepancy	Statuses, timing and duplicate events	Consistent definitions and reconciliation

⁦⁩ Make the dashboard an operating habit

Review operations weekly: pending orders, delays, payment failures, stock and complaints. Then close monthly accounts including later-reported expenses, settlements and returns, marking cohorts with incomplete outcomes. Retain dated dashboard and dictionary copies so decisions made before figures changed remain understandable. Continuous updates support operations; closed versions support comparison and accounting.

Assign an owner to each source and a correction procedure for duplicate orders, partial refunds and currency errors. Check a small sample across order, payment and carrier records whenever a material discrepancy appears or an integration changes. Every difference does not require a large analysis project; sometimes the report simply excludes a particular status. Combine contribution, cash availability and fulfillment capacity when deciding to scale, beyond sales alone. If completed-order cost after returns is unknown, prioritize better records next month before purchasing more inventory.

Required outcome: an understandable figure traceable to an order, a decision with an owner and a review demonstrating the change’s effect.

⁦⁩ Frequently asked questions and a trustworthy dashboard checklist

Why do order counts differ between Analytics and the store?
Check measurement coverage, duplicate events, order statuses and time zone. Unique order records are the operational reference; selecting the larger number does not resolve a discrepancy.

Does higher average order value mean improved profit?
It may reflect an expensive product, basket discount or shipping charges. Review contribution, returns and cost before deciding.

When can I consider the month’s result final?
After collecting settlements and costs and establishing return completeness. State the cutoff date and open cases, and separate the cash report from the order cohort’s result.

I fixed the period, currency and time zone.

I defined every numerator, denominator and included order status.

I separated paid, shipped and returned orders from settlement.

I calculated contribution without duplicating costs or losses.

I linked each signal to an action, owner and review.

I disclosed missing data and incomplete cases.