Conversion rate

How to calculate conversion rate

Divide the number of orders by the number of sessions in the same period, and multiply by 100:

Conversion rate = orders ÷ sessions × 100

A session is one visit: it starts when someone opens a page of your store and ends when they leave or stop for a while, often 30 minutes, depending on the tool.

February had the highest rate and the fewest orders. March had a lower rate than February and 90 more orders, because it had more traffic. Conversion rate and order count have to be read together: a rate can rise while sales fall.

In Tanqory, the Analytics dashboard shows today's sales, orders, visitors and conversion next to the same day last week. To apply your own counting rule, download the orders and analytics reports as CSV files and calculate the rate in a spreadsheet.

What to count: sessions, visitors and orders

Tools use different counts on both sides of the division, so two dashboards can show different rates for the same store and the same month. Here is one March counted four ways. The store had 15,000 sessions from 10,000 visitors, and 300 orders. Those orders came from 285 sessions (15 sessions placed a second order) and from 270 different buyers (some bought on two separate visits).

None of these is wrong. Each answers a slightly different question, so pick one, write it down and keep it.

• Sessions or visitors. Sessions count visits. Visitors (some tools say users) count browsers or devices, usually through a cookie. A rate per visitor is always at least as high as the rate per session, because one visitor can come several times before buying. Visitor counts are also less exact: the same person on a phone and a laptop is two visitors, and a cleared cookie makes a returning visitor look new.
• Orders or converting sessions. Counting orders lets one session count twice when a customer places two orders. Counting converting sessions caps each visit at one. The difference is small for a store whose customers rarely order twice in one visit, and large for stores where customers often place a second order straight after the first.
• Which orders. Count paid orders from the store whose traffic you are counting. Leave out test orders, orders you created by hand for a phone customer, and orders from marketplaces or other sales channels whose visits never reached your site. Adding those orders to your site's sessions makes the rate look higher than any visitor ever experienced.
• Both numbers from the same place and period. Orders usually come from your order system and sessions from your analytics tool. Use the same time zone and the same dates for both. Where visitors can decline analytics cookies, some sessions are never recorded while every order still is, so the rate reads higher than it is. Keep the consent rules steady when you compare months.

Bots and internal traffic

Bots, uptime monitors and your own team visit the store and place no real orders. They add to sessions and push the rate down. Take the March above: if 1,500 sessions came from bots and 500 from your team testing pages, and 4 of the 300 orders were test orders, the real rate is 296 ÷ 13,000 = 2.28%, not 2.00%.

A sudden drop in conversion rate with no change in orders is usually a jump in sessions, and a burst of bot traffic is the first thing to rule out. Look for sessions that last a second or less, view one page, or come from one place in a short window. Filter your own team's visits out by network address or a cookie, if your tool allows it.

Comparing rates across channels and devices

The rate for the whole store is an average of the rates of every channel and device, weighted by how much traffic each one sends. A change in the mix moves the total even when no part of it changed:

Mobile converted at 1.2% and desktop at 3.0% in both months. The store rate fell from 1.92% to 1.56% only because more of the traffic came from phones. Before you look for what broke, compare each device and each channel with itself.

Channels differ the same way. Visits from your email list come from people who already know the store; visits from a new ad reach people who have never heard of it. A low rate does not make a channel worse: what matters is what an order from it costs. Cost per order is cost per session ÷ conversion rate. Paid clicks at $0.40 converting at 1% cost $40 per order; clicks at $0.20 converting at 0.4% cost $50 per order, though the second channel looks cheaper per click.

Which channel an order belongs to also depends on the tool's attribution rule, such as the last visit before the order or the first one. Two tools with different rules give each channel a different number of orders, so compare channels inside one tool.

Conversion rate and related measures

Conversion rate says how often a visit ends in an order. It says nothing about what the order is worth, so check it against revenue per visitor:

Revenue per visitor = revenue ÷ sessions = conversion rate × average order value

Despite the name, many tools calculate it per session. A store at 2% conversion and $40 average order value takes 0.02 × $40 = $0.80 per session. A 20% discount on everything might lift conversion to 2.4% and cut average order value to $32: that is 0.024 × $32 = $0.77 per session. More orders, less money from the same traffic.

Gross margin makes the gap wider. If the $40 order cost you $20 in goods, it leaves $20 of margin, or 0.02 × $20 = $0.40 per session. At $32 the same goods leave $12, or 0.024 × $12 = $0.29 per session.

The rate is also the product of the steps before the order:

8% × 50% × 50% = 2%. The 900 carts that did not become orders are abandoned carts, a 75% cart abandonment rate (900 ÷ 1,200). Reading the steps tells you where to work: a low add-to-cart rate points at product pages and prices, a low checkout completion rate points at shipping costs and the checkout itself.

Ways stores raise conversion rate

Each of these can raise the rate. Each has a cost, so judge a change on orders, revenue per visitor and margin, not on the rate alone:

• Faster pages, mostly on phones. Lighter images and fewer scripts make pages open sooner. The cost is design and developer time, and sometimes dropping a feature that loaded slowly.
• Shipping cost and delivery date shown early. Showing them on the product page or in the cart avoids a surprise at checkout. The cost is lost orders from customers who would have accepted the charge if they had seen it later, or margin if you cut shipping prices to make them easier to accept.
• A shorter checkout. Guest checkout, address lookup and fewer fields remove reasons to stop. Adding payment methods through your payment gateway helps buyers who do not want to use a card. The cost is the fee on each method and fewer customers who create an account.
• Better product pages. Clear photos, sizes, materials and a plain returns policy answer the questions that stop a purchase. The cost is photography and writing time, and more returns if the returns policy becomes more generous.
• Discounts. A discount code lifts the rate for as long as it runs. The cost is margin on every order that would have happened anyway, and customers who learn to wait for the next one.
• Cart reminders. An email to a shopper who left a cart brings some of them back. The cost is the work of following consent rules for that email, and margin if each reminder carries a discount.
• Dropping traffic that never buys. Cutting spend on a channel whose visitors do not buy raises the rate because sessions fall. That is useful when the cost per order falls too, and harmful when orders fall with it.

Questions

What is a good conversion rate?

There is no single good number. The rate follows your prices, your products, your traffic mix and the counting rule you use, so a store selling $10 refills and one selling $900 furniture cannot be compared. The useful comparison is your own rate over time, by device and by channel, calculated the same way each time.

We do not quote an industry average here. Published averages mix stores that divide by sessions with stores that divide by visitors, and few say which rule they used, so a store cannot tell whether it sits above or below them.

Should I calculate conversion rate per session or per visitor?

Either works if you use one rule every time and say which. Per session is the easier one to count exactly. Per visitor gives a higher number for the same store, because one visitor can come several times before buying.

Why did my conversion rate fall when my sales went up?

Usually because traffic grew faster than orders. New visitors from an ad, a press mention or a social post buy less often than people who already know the store, so they add more sessions than orders. Check orders, revenue per visitor and the rate for each channel before you treat the drop as a problem.

Does a higher conversion rate always mean more profit?

No. A discount can raise the rate while revenue and margin per session fall. Check every change against revenue per visitor and margin per session, which take the size and the margin of each order into account.

How much traffic do I need before a change in conversion rate means something?

More than most small stores expect. As a rough rule, a count of orders varies by about its square root from chance alone: about 7 either way for a week of 50 orders. A move from 50 to 56 orders in the same traffic can be chance, so judge a test over several weeks, or until each version has a few hundred orders.