Ecommerce KPIs — how to calculate GMV, AOV, conversion, CAC and LTV

Running an online store comes down to five numbers: GMV, AOV, conversion rate, CAC and LTV. The trouble starts when someone calculates them their own way — treating GMV taken from a marketplace's annual report as revenue, calling conversion rate by the name it had in Universal Analytics, or calculating LTV with a formula from a SaaS article even though the store has no subscription. Ecommerce KPIs aren't hard to calculate — they're hard to calculate the same way the report you're reading them off calculates them. This article goes through each of the five numbers with a definition taken directly from GA4's and Shopify's documentation and from how marketplaces define GMV for themselves, and ends with a sixth thing none of them talks about on its own: margin per order.

GMV — how much actually moved through your store, not how much you earned

GMV (Gross Merchandise Value) is a term that came to ecommerce from marketplaces, and it's worth knowing a marketplace's own definition before you calculate your own version. Allegro, the marketplace on which many Polish stores sell and the group behind allegro.pl (it also operates allegro.cz, allegro.sk and allegro.hu), defines GMV in its annual report for 2025, under "Alternative Performance Measures", as: "'GMV' means gross merchandise value, which represents the total gross value of goods and tickets sold on the following platforms (including value added taxes)" — that is, the gross value, including VAT, not the company's revenue. The same report states that alternative performance measures, GMV included, "are not accounting measures within the scope of IFRS", that they "may not be comparable to similarly titled measures of other companies", and that the assumptions underlying them have not been audited in accordance with IFRS or any other generally accepted accounting standards (read 2026-10-05).

Treat this as the definition of the marketplace you most likely sell on, not as the definition for the whole market: each marketplace that reports GMV writes its own definition, and the details — VAT, shipping, cancelled or returned orders — need not match, so a number called "GMV" on one platform doesn't automatically mean the same thing as "GMV" on another.

For your own store, that has one practical consequence: if you calculate "your GMV", calculate it with the same logic — the total gross value of orders placed (or paid, if that's how you define the period), before deducting returns, after-the-fact discounts and costs. GMV isn't what ends up in the till. If someone at your company says "GMV is growing" and means net revenue after returns in the same sentence, those are two different numbers — worth separating before either one reaches a board report or an investor.

This also matters when you compare GMV over time within a single company: one release might report growth as a quarter-on-quarter percentage, another might report the full-year total as an absolute figure — treat those as two separate statements, not one trend line, unless you've checked they cover the same period on the same basis.

AOV — average order value

AOV (Average Order Value) has a simpler, more standardised definition. Shopify gives the formula without ambiguity: "You can calculate average order value (AOV) by dividing your total revenue by the number of orders in the same time period: AOV = Total Revenue ÷ Number of Orders" — revenue for a given period divided by the number of orders in that same period.

GA4 doesn't have a metric literally called "AOV", but it has an equivalent one — "Average purchase revenue", defined as "The average purchase revenue over the selected time frame" (support.google.com/analytics/answer/9143382, read 2026-10-01). Watch out for a near-identical name: GA4 also has "Average purchase revenue per active user", defined as "The sum of the purchase revenue per active user" — a different number (revenue divided by the number of active users, not by the number of orders), easy to confuse by name alone in the interface.

The second trap is the base itself. GA4's documentation defines "Purchase revenue" as the sum of revenue from purchases made on your site or app after subtracting any refunds — so it deducts the value of refund events, if your store sends them. On the same page, "Item revenue" covers revenue from the items alone, excluding tax and shipping. Revenue in GA4 is therefore not GMV: it isn't counted "before returns", and whether it includes VAT and shipping depends on the value your store passes in the purchase event. Before you compare AOV in GA4 with AOV in your store's admin, check what goes into that value.

A purely illustrative example, not a real store's data: if purchase revenue in a given month is X, and the number of orders in that same month is N, Shopify's formula gives AOV = X ÷ N. Plugging in illustrative values: at X = PLN 120,000 and N = 400 orders, AOV = PLN 300. The same company, with a different definition of "order" in the denominator (say, including or excluding cancelled orders), produces a different N — and therefore a different AOV for the same month, which is why what matters is keeping one written-down definition of "order" over time, not just one formula.

Conversion in GA4 today — a "key event rate", not a "conversion rate"

If you're using GA4 and looking for a metric called "conversion rate" the way it was named in the old Universal Analytics — you won't find it under that name. Standard Universal Analytics properties stopped processing data on 1 July 2023 (support.google.com/analytics/answer/11583528), and GA4 has renamed things since: Google now calls the events that measure actions important to your business's success "key events", while a "conversion" is an action you use to measure the performance of your ad campaigns and optimise your bidding strategy (support.google.com/analytics/answer/13965727). Per GA4's documentation, the metrics go by different names today: "Key events" is "The number of times users triggered a key event", and "conversion rate" itself comes in two variants, session-based and user-based:

• "Session key event rate" — "The percentage of sessions that converted. Session key event rate = number of sessions in which a key event happened / total number of sessions".
• "User key event rate" — "The percentage of users who converted. This metric is calculated as the number of users who performed a key event action divided by the total number of users" (support.google.com/analytics/answer/9143382, read 2026-10-01).

In other words: GA4 still talks about "conversion" when describing these metrics ("the percentage of sessions/users that converted") — only the metric's name in the interface changed, not the underlying concept. If a report or an agency gives you a "CR" without saying whether it's the session or the user variant, ask — the difference is real for customers who buy across several sessions.

The difference between the session and user variants isn't cosmetic. Take a purely illustrative case: if one user visits the store across three sessions and buys only in the third, for the session-based rate that user contributes three sessions to the denominator, of which the key event occurred in only one. For the user-based rate, the same user counts once, regardless of how many sessions they came back in before buying. For products customers look at several times before deciding (a higher price point, a longer decision process), these two variants can diverge noticeably — another reason to always state which one you're quoting, instead of saying "conversion X%" in general.

Take care when reading the number of orders in GA4: the documentation separately defines the metrics "Transactions" and "Ecommerce purchases" — described almost identically ("the number of completed purchases on your site" / "the number of times users completed a purchase") and both fed by the same purchase event. The documentation doesn't explain when the two numbers diverge — so pick one of them for reporting and stick with it, and if they do differ in your data, check first how your store sends the purchase event.

CAC and LTV — the cost of winning a customer, and their value over time

CAC (Customer Acquisition Cost), per Shopify's definition, is "total cost of acquiring a single customer", calculated as: "Divide this total [marketing and sales spend for the period] by the number of new customers who made purchases for the first time during that same period" — the total spent on acquiring customers divided by the number of new customers in that same period. Shopify's own worked example puts $500 of marketing spend against 10 new customers, giving a CAC of $50 per customer (shopify.com/blog/customer-acquisition-cost, read 2026-10-01) — an illustration from the article, not an industry benchmark.

For LTV (Lifetime Value, which Shopify calls CLV), Shopify's blog gives two formulas (shopify.com/blog/customer-lifetime-value, read 2026-10-01). Basic: "CLV = (Average Order Value × Purchase Frequency) × Average Customer Lifespan" — AOV times purchase frequency times the average length of the customer relationship. Extended, accounting for the time value of money: "CLV = Gross Margin Per Customer Lifespan × [Retention Rate / (1 + Discount Rate − Retention Rate)]" — gross margin over the whole relationship, adjusted by the retention rate and the discount rate.

Shopify's CAC article also says that if your CLV-to-CAC ratio "falls between 3:1 to 5:1, your acquisition strategy is working efficiently" (its CLV article simply calls 3:1 "a good LTV to CAC ratio") — but neither page gives a methodology or a sample size behind those figures, so we don't cite them here as a number to benchmark your store against. What's left is the construction itself: calculate CAC and LTV separately, watch the trend of each, don't chase a magic ratio without a source.

Two variables in the basic LTV formula need their own interpretation, not an industry benchmark. "Purchase frequency" is the number of orders per customer in a given period (a year, say) — calculate it from your own order history, customer by customer, not by guessing one value for the whole catalogue. A starting point is GA4's "Transactions per purchaser" metric, which the documentation describes as the average number of purchases per purchaser in the selected time frame. Set the range to a year and it gives you an annual purchase frequency — but one value for all customers; segments you still split out yourself. "Average customer lifespan" is the average length of time a customer keeps coming back before they stop — and this figure differs even within a single store, between product categories with different purchase frequencies (fast-rotating cosmetics versus furniture bought once every few years). Calculating one LTV for the whole catalogue without that breakdown usually produces a number that describes no real customer segment.

One distinction matters straight away: ecommerce LTV and SaaS LTV are two different formulas, despite the identical name. The formula above calculates LTV from purchase frequency and relationship length — typical of a transactional model, where a customer buys repeatedly but without a subscription. A subscription model calculates LTV differently — from churn rate and recurring revenue, not from order frequency. If you also run a SaaS product alongside the store (or you're reading an article about SaaS metrics and trying to carry the formula over 1:1 to a store), our article on SaaS metrics — ARR, MRR, churn and retention-based LTV — covers that; it isn't the same calculation, despite sharing the metric's name.

Margin per order — the number none of the above shows you

GMV, AOV, CAC and LTV talk about value and the cost of winning a customer — none of them says whether a given order was profitable after deducting the payment provider's fee, the shipping cost that doesn't always land entirely on the customer, and packaging. The mechanism is simple and can stay invisible in the sales reports themselves: GMV and AOV can rise month over month while contribution margin per order falls, if, say, the share of free-shipping orders rises, a marketplace's commission rises, or the payment-method mix shifts towards options that cost the seller more.

Written out with variables rather than percentages: if AOV is A, the cost of goods in that order is k, the payment provider's fee is g, shipping cost not covered by the customer is d, and packaging cost is p, contribution margin per order is A − k − g − d − p. Each letter on the right changes independently of A — which is why GMV and AOV can rise month over month while this calculation breaks down, if, say, d rises faster than A (more free-shipping orders at an unchanged average basket value). We cover the exact breakdown of payment-gateway and payment-method fees that feed into g and d separately in payments and logistics — what matters here is only that without this calculation, the previous four numbers can show growth that, in practice, means nothing for profit.

Where to find these five numbers

• GA4 — AOV as "Average purchase revenue", conversion as "key event rate" (session or user — pick one and stick with it over time), order count as "Transactions" or "Ecommerce purchases" (both fed by the same purchase event — don't add them together, pick one name for reporting), and purchase frequency for LTV as "Transactions per purchaser". In the interface these metrics sit in the standard reports (the sections covering engagement and monetisation) or in Explorations if you need a custom breakdown, say AOV split by traffic channel — Explorations give you more control over dimensions, but you have to build the report yourself, while standard reports are ready immediately.
• Your store platform's dashboard — Shopify shows AOV directly in the dashboard, under Reports > Customers (shopify.com/blog/average-order-value, read 2026-10-01); other platforms have their own sales reports, but before comparing them with GA4, check how they define an "order". The platform dashboard is a more convenient day-to-day source than exploring GA4, if you don't need a breakdown by traffic channel.
• If you sell on Allegro or another marketplace — remember that the GMV you see quoted in Allegro's own investor communications describes the whole platform, not your seller account. Your own GMV (the gross value of your orders on that marketplace) is a different number; calculate it the same way, but you won't find it under a ready-made label in the seller dashboard — you gather it yourself from your own sales report.

Ecommerce KPI dashboard — a five-number checklist to get started

• GMV (or total gross order value) calculated consistently for the same period and the same definition of "sale" (orders placed vs. paid) month over month.
• AOV calculated as revenue ÷ number of orders, from the same platform as GMV — not from a different system with a different definition of "order".
• Conversion rate from GA4 — one variant chosen (session or user) and kept throughout subsequent reports, so the trend stays comparable.
• CAC calculated from actual marketing spend divided by new customers in that same period — not from the whole marketing budget divided by all orders.
• LTV calculated with the transactional formula (AOV × frequency × relationship length), not the SaaS formula, unless you run a subscription model.
• Margin per order checked at least once a quarter alongside the five numbers above — so rising GMV and AOV don't mask falling profitability.
• A definition of "order" (placed or paid, with or without cancellations) written down in one place and used consistently for GMV, AOV and the order count — not a different one in every report you happen to be reading.

For how these numbers fit into the rest of running a store — product data, automation, integrations — see our ecommerce operations section.