Your Checkout Conversion Rate Looks Fine. Your Revenue Per Session Says Otherwise

Blended conversion rate hides the mix

Conversion rate is orders divided by sessions, and it is the number most stores watch first. The problem is the word blended. Sessions from search, from social, from email, from new visitors and returning visitors all go into the same denominator, and the orders all go into the same numerator. A shift in the mix, more low-intent traffic, fewer returning buyers, a device split that moved toward mobile, changes the number without changing anything about the store.

That is how a store can report a flat conversion rate for a quarter while revenue quietly falls. The order rate held because the traffic that converts kept converting. The money fell because the new traffic converts at a lower value, or does not convert at all and still counts in the denominator. The blended number cannot tell you which, because it was built to hide exactly that.

The three splits that matter

Split by source first. Traffic from email and returning visitors converts at a different rate than traffic from a cold social campaign, and a month that leans harder on cold traffic will show a lower blended rate even if every channel performed the same as last month. Split by device second. Mobile converts lower than desktop on most stores, and a shift toward mobile traffic drags the blended number down without any change in the checkout. Split by new versus returning third. Returning visitors buy more often and spend more per order, so a month with fewer of them looks worse on every blended metric at once.

Each split tells a different story, and each story points at a different fix. A source problem is a targeting problem. A device problem is a checkout problem. A returning-visitor problem is a retention problem. The blended number points at none of them, which is why stores that watch only the blended number keep fixing the wrong thing.

What to check on Monday morning

Open revenue per session by segment, not conversion rate by itself. Revenue per session is total revenue divided by sessions, and it catches what conversion rate misses: a stable order rate with falling order value, or a stable order rate with a traffic mix that stopped buying. Compare each segment against its own history, same source, same device, same new-versus-returning split, and look for the segment that moved.

Then check average order value against the free-shipping threshold. A store whose AOV sits just under the threshold is leaving money on the table in a way no conversion rate will ever show. And check checkout completion by step, not by page: the step where buyers leave is the step to fix, and the blended conversion rate will never name it. The work is not a new dashboard. It is reading the three splits every week until the segment that moved becomes obvious.

Keep reading: Free Shipping Thresholds: The Breakeven Math Most Stores Skip, Checkout Field Count: Cut to Eight and Stop Losing Orders.

Frequently asked questions

What is revenue per session?

Revenue per session is total revenue divided by total sessions in the same period. It is a derived metric, not a report Google Analytics ships on its own, and it answers the question conversion rate cannot: how much money does the average visit produce, regardless of how many of them convert.

Why can conversion rate stay flat while revenue falls?

Because conversion rate only counts orders, not order value or traffic quality. A flood of low-intent sessions can hold the order rate steady while average order value falls, or while the new sessions convert at a lower value. The order count looks fine. The money does not.

Should I stop tracking conversion rate?

No. Track it split by segment, next to revenue per session. Conversion rate by source, by device, and by new versus returning visitors will each tell a different story than the blended number. The blended number is the one that hides the mix, and the mix is where the leak lives.

What is a good revenue per session?

There is no universal number, and anyone quoting one is selling something. Compare your revenue per session against your own history, split by segment: same source, same device, same new-versus-returning split, this month against last month and against the same month last year. The movement is the signal, not the level.