What Is a Good Conversion Rate? Start With What You Can Afford
A good conversion rate is the one that beats your cost per click divided by contribution per order. The formula, real benchmark ranges, and the traps.
Someone converts at 1.1 percent and is making money. Someone else converts at 4.2 percent and is quietly losing it on every order. Both will find an article telling them their number is wrong. The averages published online are built from other people's traffic, other people's prices and other people's cost structures, so they cannot tell you whether yours is good.
There is a version of the question that does have an answer. Not "is 2.5 percent good", but "what rate do I need at the price I am paying for clicks". That one takes two inputs you already have and about five minutes.
The averages you are comparing against are not your traffic
Conversion rate is far more a property of who arrived than of what they landed on. Traffic from a branded search, where the visitor typed your name, converts several times better than a cold display impression, and no amount of button testing closes that gap. When you read a single sitewide average, you are reading a weighted blend of very different audiences, and the weighting is usually the thing that changed.
| Traffic source | Commonly reported range | Why it sits there |
|---|---|---|
| Branded search | 8% to 15% | Intent was already formed before the click |
| Non-brand search | 2% to 4% | Real demand, but comparison shopping is live |
| Google Shopping | 1.5% to 3% | Price is visible before the click, so the click is pre-qualified |
| Email to existing buyers | 3% to 6% | Prior relationship, near-zero acquisition cost |
| Paid social retargeting | 2.5% to 6% | Return visits from people already deciding |
| Paid social prospecting | 0.7% to 2% | Interruption, not search, so most visits are browsing |
| Display and programmatic | 0.2% to 0.8% | Attention is borrowed and shallow |
Read those as postcodes, not targets. The practical use is diagnostic. If your Meta prospecting sits at 0.9 percent, that is ordinary and the fix is probably in the offer or the price, not the landing page. If your branded search sits at 3 percent, something is broken, because people who typed your name should not be leaving at that rate.
The uncomfortable version of this: when a sitewide conversion rate improves after a budget shift, check the mix before you take credit. Moving spend from prospecting into retargeting raises the average almost mechanically. It also raises very little else.
The rate you need is a division, not an opinion
Here is the number that belongs to you. Every visit costs what you paid for the click. Every order returns what is left after the costs that move with it. The conversion rate that turns one into the other is your break-even.
Break-even conversion rate = cost per click / contribution per order
Contribution per order = average order value x contribution marginTake a store with an average order value of $86 and a contribution margin of 42 percent after goods, shipping, packaging, processing fees and a returns provision. Each order contributes $36.12. Paying $1.40 a click, the break-even conversion rate is 1.40 / 36.12, which is 3.88 percent.
Now put the actual site next to it. At 2.5 percent, those $1.40 clicks produce a cost per acquisition of $56.00 against $36.12 of contribution, so every order loses $19.88 before a single fixed cost is paid. The 2.5 percent is not a mediocre number waiting for a nudge. It is a losing number, and the account was losing money in a straight line for as long as it ran. Work your own version in the conversion rate calculator, then check the result against the CPA calculator so you see the loss per order rather than a percentage.
Two things follow from that shape. Bidding cheaper is a conversion rate improvement, since it lowers the bar you have to clear, and it usually lands faster than any test. Raising contribution per order works the same way and is often easier still, which is why a bundle or a shipping threshold can rescue an account that no landing page change was going to save. Settle your true contribution first in the contribution margin calculator, because an optimistic margin here produces a break-even target that is too low by exactly the amount you flattered yourself.
Price and consideration set the ceiling
Ecommerce sites selling a $30 impulse item and a $3,000 considered purchase are not playing the same game, and comparing their conversion rates is close to meaningless. As price rises, the buyer takes longer, involves other people and returns several times before deciding. A 0.4 percent rate on a $2,400 average order can be a stronger business than 6 percent on a $19 one.
Lead generation shifts the meaning again. A form completion is cheap for the visitor, so rates of 5 to 15 percent are normal on paid search, and the number stops meaning much on its own. What matters is the rate through to a qualified lead and then to a closed deal, because a form that converts at 14 percent and qualifies at 8 percent is worse than one converting at 6 percent that qualifies at half. If you are running a longer funnel, judge the entry rate only alongside cost per lead and what the leads are actually worth.
Before you call a change real, check the sample
Most reported conversion rate movements are noise. This is where accounts waste the most time, because a percentage looks solid in a dashboard while resting on a handful of orders.
A page at 1,000 sessions a week and 2.5 percent produces 25 orders. Twenty-five. A week with 31 orders reads as 3.1 percent and looks like a 24 percent lift, and it is well inside the range you would expect from doing nothing at all. To detect a genuine move from 2.5 to 3.0 percent at 95 percent confidence and 80 percent power, you need roughly 16,800 sessions per variant, about 33,600 in total. At 1,000 sessions a week that test runs for eight months.
That arithmetic is why most small sites should not be running A/B tests on button colour. They do not have the traffic to resolve the effects they are hoping for, and they will keep declaring winners that are coin flips. Size the test before you start it in the sample size calculator, and if the answer is longer than your season, test something big enough to show up instead: the offer, the price, the shipping terms. Larger effects need dramatically less traffic to prove, so a 50 percent lift from 2.5 to 3.75 percent needs about 3,000 sessions per variant rather than 16,800.
Two ways the number lies
The first is the denominator. Sessions, users and unique visitors give three different conversion rates from identical sales, and analytics platforms have quietly changed their defaults more than once. A site with heavy repeat browsing can show a rate 30 to 40 percent higher measured on users than on sessions, and nothing happened except the definition. Whichever you pick, write it down and never compare across two tools that disagree.
The second is micro-conversions. Counting newsletter signups, add-to-carts and video views as conversions inflates the headline into something that no longer connects to revenue. Track them by all means, but keep the primary rate tied to the event that produces money, and check it against average order value in the same period. A rising conversion rate with a falling average order value often just means a discount, and the two together decide whether the month was any good.
Start with one number this week
Pull your cost per click and your contribution per order for the last 30 days and divide the first by the second. That single figure is your break-even conversion rate, and it tells you immediately whether you are running a site that needs fixing or an account that is bidding above what the site can support. Almost every advertiser who does this for the first time discovers the problem was on the side they had not been looking at.