10 KPIs for Ecommerce: Metrics That Drive Growth

Most ecommerce founders have a dashboard. Far fewer have a decision they can trace back to it. Revenue, sessions, conversion rate, ad spend, and a dozen channel numbers land in a weekly email, get skimmed, and change nothing about what happens on Monday morning.

The uncomfortable pattern is that the fastest-growing stores usually track fewer numbers than the ones standing still. A founder running an eight-figure brand can tell you her contribution margin after returns, her new-customer acquisition cost, and the marketing efficiency floor she won't spend below without opening a laptop. A founder stuck at two million has forty tabs open and no answer to the only question that matters, which is whether last month's growth made money.

That gap is the whole game. A number earns a place on your scorecard only when a specific decision follows from it: raise the budget, cut the SKU, rewrite the page, change the supplier, kill the channel. If nothing changes based on the figure, it's trivia with a graph attached.

The ten KPIs for ecommerce operators below pass that test. Each one gets the same treatment: what it measures, how to calculate it without fooling yourself, the misreading that does the most damage, and the fix that turns it into an action. Some of them will argue with advice you've already adopted.

1. Conversion Rate

Conversion rate is orders divided by sessions, multiplied by 100. Google Analytics defines ecommerce conversion rate as transactions divided by sessions, which sounds unambiguous until you notice that sessions, users, and orders are three different units.

The denominator moves more than many realize. The U.S. Census Bureau's fourth-quarter 2025 ecommerce report put U.S. ecommerce sales at an estimated $316.13 billion, or 16.6% of total retail sales, up from 16.2% a year earlier. A rising channel tide lifts orders without any improvement in your funnel, so a flat conversion rate inside a growing category can hide a store that's getting worse.

The Misreading That Costs Money

The expensive version is the sitewide average. A store reports a healthy overall rate and feels fine, while paid social, mobile checkout, and one traffic-heavy landing page each convert far below it. Blend those together and the average tells you nothing about which one to fix.

Practical rule: Any conversion number you'd act on has a segment attached to it.

• Split by device, source, and visitor type. First-time mobile traffic from paid social and returning desktop traffic from email are two different businesses wearing one number.
• Break the funnel into steps. Add-to-cart rate, checkout-start rate, and payment completion rate localize the leak that a single rate hides.
• Pair it with revenue per session. A conversion lift bought with sitewide discounts can leave you with less money per visitor than before.
• Test changes big enough to move the number. Button-level tweaks require sample sizes most stores never reach. Rewrite the offer or the page structure instead.

The tactical playbook for improving ecommerce conversion rates covers speed, proof, and checkout sequencing in more depth.

2. Average Order Value

A shopper comes for one bottle of serum and leaves with a cleanser and a travel size, because the cart recommended the bundle that made sense at that price. That's AOV created by design, not luck.

Average order value is total revenue divided by order count. Pick one definition and write it next to the number, because the versions disagree: gross sales less discounts, excluding tax, shipping, and duties, is the one I'd standardize on, since it isolates the merchandising decision from the fulfillment decision.

Where AOV Actually Comes From

The misreading is treating AOV as a health metric. It climbs for four very different reasons: you raised prices, you bundled, your cheapest items went out of stock, or you discounted hard enough to drag bigger baskets out of people. Three of those four can raise AOV while lowering what you keep per order.

• Set the free shipping threshold just above current AOV. Most shoppers add the cheapest item that clears the bar, so make sure that item carries margin rather than giving the promotion away twice.
• Bundle by use case, not by discount. A starter set that solves one problem sells better than three products at 15% off that solve nothing together.
• Sell the add-on after checkout. Post-purchase offers don't interrupt the payment flow, and a meaningful share of buyers accept them.
• Split new and returning AOV. Repeat buyers respond to replenishment bundles and quantity breaks. First-time buyers respond to a threshold. Same metric, different levers.

3. Customer Lifetime Value

CLV isn't a number you compute once. It's a bet you place on the future, and most stores place it with revenue when profit is the only thing that pays for ads.

Customer lifetime value is the gross profit a customer generates across the whole relationship. Average order value multiplied by purchase frequency multiplied by expected lifespan, then multiplied by gross margin, is the version I trust. Revenue-based CLV runs higher and flatters every decision built on top of it.

Build It on Cohorts, Not Averages

The damaging misreading is a blended CLV computed across five years of customers and then used as permission to raise acquisition spend today. That number describes history. It says nothing about whether the customers you bought last quarter will behave the same way.

• Use gross margin, not revenue. A customer worth twice as much revenue at half the margin isn't worth more.
• Group customers by acquisition month and channel. Compare cohorts at the same age, not in the same calendar month.
• Watch the second purchase. The share of first-time buyers who order again within 30 or 60 days tells you where a cohort is heading long before a multi-year CLV exists.
• Set a CLV-to-CAC ratio and enforce it. The right ratio depends on your margin structure and how long you can fund payback. Write it down anyway, because an unwritten target is a target nobody checks.

4. Customer Acquisition Cost

Blended CAC is a comfort blanket. It's also the reason founders are surprised when a channel that "worked" turns out to have been subsidized by organic search the whole time.

Customer acquisition cost is total acquisition spend divided by new customers acquired in the same window. The numerator includes ad spend, agency retainers, affiliate commissions, creative production, and influencer fees. The denominator counts new customers only, never repeat orders.

Make CAC Tell You Where to Spend

The classic error is dividing all marketing spend by all orders and calling the result acquisition cost. That ratio always looks better than reality because returning customers carry it. The second error is comparing that inflated number to a revenue-based lifetime value, which makes almost any spend look defensible.

• Report blended and channel-level CAC together, reconciled monthly. If the channel numbers don't add up to the blended figure, you're missing spend somewhere.
• Judge CAC against first-order contribution margin. A customer who returns money on the first order has to be justified by retention, and that's a different argument.
• Use a blended efficiency ratio as a cross-check. When attribution is incomplete, total revenue against total spend keeps you honest.
• Set a payback target in orders. Many operators aim to recover acquisition cost within the first two or three orders. Whatever you choose, tie it to your cash position, not to a blog post.

Cutting CAC isn't only a media-buying exercise. The guide to lowering customer acquisition cost covers the offer, landing page, and retention changes that do more than bid adjustments ever will.

5. Return on Ad Spend

ROAS is the most quoted and most misleading number in paid media. Fixed marketing costs rose 20% to 40% while ROAS declined by at least 10% across every revenue cohort analyzed in Q2 2025, according to the Ecom CFO benchmark report preview, and EBITDA fell across all cohorts, including a decline from 11.33% to 6.15% for brands below $10 million in revenue.

Return on ad spend is attributed revenue divided by ad spend. That definition is where the trouble lives. Attributed revenue is whatever a platform took credit for, not what the ads caused, and it arrives before returns, discounts, and fulfillment costs take their bite.

Set a Break-Even and Enforce It

The misreading that sinks brands is treating a strong ROAS as proof of profit. A campaign can return several dollars per dollar spent and still lose money on a product with thin contribution margin. Break-even ROAS is one divided by your contribution margin before ad spend. Below that line, volume just accelerates the loss.

• Separate branded from non-branded. Branded search campaigns usually show inflated ROAS because they intercept demand you already created.
• Judge on profit after returns. A category with heavy return exposure needs a higher ROAS target than the same margin in a category that ships clean.
• Compare platform ROAS against blended efficiency. When the two diverge for more than a couple of weeks, over-attribution is the likeliest explanation.
• Extend the window for considered purchases. A seven-day window flatters fast sellers and starves products that take three weeks to close.

Paid social efficiency also leaks in places nobody attributes. An ad post with an unmoderated comment section full of complaints converts worse than the same creative with clean comments, so Facebook ad comment moderation rules belong on the operating checklist alongside bids and budgets.

6. Cart Abandonment Rate

Abandonment isn't a leak you fix once. It's a structural feature of online shopping, and the benchmark proves it. Baymard Institute's cart abandonment research, a compilation of 50 studies published between 2006 and 2025, found an average abandonment rate of 70.22%, with individual studies ranging from 55% to 84%. Roughly seven in ten carts never produce a transaction.

Not all of that is failure. The same research found 42% of U.S. online shoppers abandoned because they were browsing or not ready to buy. Among the rest, 40% cited extra costs that were too high at checkout. That split matters, because discounting a browser teaches them to wait for a coupon, while discounting a shipping-cost objection just hides the underlying problem.

Cart abandonment rate is one minus completed orders divided by carts created, times 100. The blended figure is nearly useless on its own. Split it by device, traffic source, cart value, and checkout step before anyone proposes a fix.

Recovery rule: Time the first recovery email to the moment the shopper's intent is still warm, within an hour, then space the remaining messages across several days without escalating discounts.

The checkout itself is usually the culprit. Baymard's checkout usability benchmark of 344 leading U.S. and European ecommerce sites found 65% had checkout UX rated mediocre or worse, while only 2% were rated good. If almost everyone's checkout is mediocre, the fix is rarely a bigger discount. It's guest checkout, visible shipping costs early, fewer form fields, local payment methods, and error messages that tell people what to correct.

7. Repeat Purchase Rate

Ask two people on your team for the repeat purchase rate and you'll get two numbers that share a name and mean different things. One is a period rate: customers with two or more orders in the window, divided by customers who ordered in that window. The other is a lifetime count: every customer whose entire history includes a second order. The first moves when retention changes. The second mostly moves as your customer base ages.

Set the Window to Your Product's Rhythm

A 30-day window makes sense for consumables. It's meaningless for furniture, where a second purchase might arrive in year three. A repeat rate measured on the wrong window will either look catastrophic or look fine while new buyers churn.

• Match the window to the replenishment cycle. If a product lasts 45 days, measure at 60 and 90 days, not 30.
• Read it by cohort. A cohort holds the starting group fixed instead of mixing this year's buyers with five years of history.
• Slice by first product purchased. Some entry products create loyalists and others attract one-time bargain hunters. That finding changes your ad creative and your assortment.
• Report subscriptions separately. Recurring orders inflate retention and can mask a collapsing one-time purchase business.

The most useful companion number is the second-purchase rate inside the first 60 days. It arrives early enough to act on and correlates with where lifetime value is heading.

8. Gross Profit Margin

Margin is the ceiling on every other number here. It sets your break-even ROAS, it caps what you can pay for a customer, and it decides whether a discount is a growth tactic or a slow liquidation.

Gross profit margin is revenue minus cost of goods sold, divided by revenue, times 100. The word that gets ignored is landed. Cost of goods includes the product, inbound freight, duties, and handling. A supplier price without freight and duty is a number that flatters a SKU you may be losing money on.

Get to Landed Cost per SKU

Reviewing margin at the company level hides the product that's dragging everything down. One oversized item with high shipping cost can erase the profit of ten clean SKUs, and the blended report will never show it.

• Build landed cost per unit and update it when supplier prices or freight rates move. A quarterly refresh is the minimum in most categories.
• Review SKU-level margin monthly. Sort by contribution dollars, not percentage, so you see where the money comes from.
• Track contribution margin after shipping, platform fees, and expected returns. Gross margin is the ceiling. Contribution is what you spend.
• Set discount guardrails from the margin floor. Decide the maximum markdown per SKU before the promotion calendar is built, not during the negotiation.

Discount rule: Every percentage off is a margin decision wearing a marketing costume.

Before signing off on a launch price, run the landed cost through an Amazon FBA profit calculator so the fee stack and fulfillment costs are in the same place as the product cost.

9. Marketing Efficiency Ratio

Attribution is getting worse at the exact moment discovery is moving further from your pixels. Salesforce's ecommerce statistics report that 39% of consumers, and more than half of Gen Z, already use AI for product discovery. Those paths don't always leave a clean click trail for your last-click report to credit.

Marketing efficiency ratio is total revenue divided by total marketing spend. Because it doesn't depend on platform attribution at all, it can't be inflated by two platforms claiming the same order.

How to Use a Blended Number

The misreading is using MER to make channel decisions. It can't tell you whether to cut Meta or Google, because it has no idea who deserves credit. It's a control, not a scalpel, and the moment someone uses it to justify killing a channel you've lost the plot.

• Write down the numerator and denominator. Include affiliates, influencers, creative production, and agency fees in spend, or MER drifts upward and lies.
• Set a MER floor at your break-even. Below that number, growth destroys cash. Above it, you have room to push.
• Track new-customer revenue share next to it. MER can look excellent while all growth comes from existing buyers and the acquisition engine stalls.
• Read it weekly, alongside channel ROAS. MER sets the direction and channel data handles the steering.

10. Return Rate

Returns are the line item most ecommerce P&Ls bury. The National Retail Federation estimates that 19.3% of online sales will be returned in 2025, against 15.8% across retail overall, as summarized in this ecommerce KPI breakdown from 2Visions. That gap is a cost online sellers carry and physical retailers mostly don't.

Return rate is returned units divided by units shipped, times 100. Track it in units as well as orders, because one customer sending back six items is a different problem from six customers sending back one each.

Make Returns a SKU-Level Signal

Refunds as a share of revenue is the wrong lens. It hides the shipping you paid both ways, the handling, the refurbishment cost, and the margin that never comes back. A blended return rate hides all of it.

• Code every return reason consistently. Sizing, fit, quality, damage, late delivery, and didn't match description point to different fixes.
• Check return rate by SKU before you scale ads. Scaling a high-return product multiplies the loss along with the revenue.
• Calculate the full cost of a return. Outbound shipping, return shipping, inspection time, and any markdown on resale.
• Prefer exchanges and store credit where the economics allow. Both keep the revenue and skip a refund.

The guide to reducing returns in ecommerce covers fit tools, size guides, and product page changes that drop return rates without shrinking the catalog.

10 Ecommerce KPIs at a Glance

A KPI is only useful when its definition stays consistent and its blind spots are visible. Use this table to scan the ten measures, then investigate the risk column before changing a budget, offer, or product page.

From Dashboards to Decisions, Putting Your KPIs to Work

These ten metrics aren't a list. They're a chain, and the links run in one direction. Gross profit margin sets your break-even ROAS. Break-even ROAS governs what you can afford to pay in customer acquisition cost. Customer lifetime value justifies paying that acquisition cost at all, and repeat purchase rate is what makes the justification true. MER sits above the whole thing as the check that nobody is double-claiming credit.

Once you see the chain, most dashboard arguments resolve themselves. A founder who wants to scale paid media and a founder who wants to protect cash are usually looking at the same evidence from two ends of the same equation.

Five Numbers, Once a Week

A weekly scorecard of five numbers beats a monthly deck of fifty. I'd build it this way: contribution margin after returns, new-customer CAC, MER, second-purchase rate inside 60 days, and return rate by top SKU. Everything else becomes a diagnostic you pull when one of the five moves.

Keep the definitions written next to the numbers and never change them mid-comparison. When a session-counting change or a platform update shifts a denominator, the target written against the old definition becomes fiction. Start a fresh baseline and say so.

Benchmark Against Operators, Not Blog Posts

Published benchmarks are good for one thing: telling you whether a number is normal or wildly off. They're built from other people's stores, measured another way. Your own cohort history is the only honest target.

The faster calibration comes from people who own a P&L at your level. Million Dollar Sellers is an invite-only group of 7-, 8-, and 9-figure ecommerce founders whose members collectively generate $15B+ in TTM revenue, and the conversations there are unusually concrete about numbers because everyone is reading their own margin reports.

If retention is your weakest link, cross-category ideas are worth stealing. These seller guides for artisan brands cover repeat-purchase tactics from small-batch sellers that translate well to larger catalogs.

Pick the three KPIs you track least well. Build a cohort view for each this week, even a rough one in a spreadsheet. Then find one operator peer and compare numbers before you spend another dollar on growth.

Million Dollar Sellers gives founders at $1M+ TTM a room where the numbers get discussed with full candor: mastermind calls, private forums, 500+ member SOPs, and events like the Inspire conference, with vendors kept out of membership. If your KPI scorecard has plateaued because you're benchmarking against blog posts instead of operators who own their P&L, that peer group is the fastest fix. Visit Million Dollar Sellers to see whether your revenue level qualifies you for a seat.