A single site-wide conversion rate is one of the most confidently misleading numbers in analytics. It exists, it is easy to calculate, and it describes no visitor who has ever used your site.
The reason is that traffic sources differ by more than they resemble each other. Someone who typed your brand name into Google already decided to find you. Someone who saw a video ad between two other videos did not decide anything. Both are visitors. Averaging them produces a number that is true of the arithmetic and false of everybody.
- Email to existing list~6%
- Branded / direct~4.5%
- Organic search~2.8%
- Paid search, non-brand~2.2%
- Paid social, cold~0.8%
- Display~0.4%
Directional, not authoritative. Published channel benchmarks disagree substantially and vary by industry, offer, and how each dataset defines a conversion. The ordering is far more stable across sources than the values, and the ordering is the point.
Roughly where each channel tends to land. Email to existing list: ~6%. Branded / direct: ~4.5%. Organic search: ~2.8%. Paid search, non-brand: ~2.2%. Paid social, cold: ~0.8%. Display: ~0.4%.
We are deliberately hedging those figures. Every published set of channel benchmarks we looked at disagreed with every other one on the numbers, because they define conversions differently and sample different industries. What they do agree on is the shape. Email and branded traffic sit at the top, cold paid social and display sit at the bottom, and the gap between top and bottom is not 20 percent, it is closer to an order of magnitude.
Why the ordering is stable even when the numbers are not
Because the ordering is really a proxy for intent, and intent is a property of the visitor rather than of your website.
Email to an existing list reaches people who already gave you their address, which means they already converted once. Branded search reaches people who know your name. Non-brand paid search reaches people with a problem but no opinion about you. Cold paid social reaches people who were doing something else entirely and got interrupted, competently.
So the channel is not causing the conversion rate. The channel is selecting for how far along the visitor already was, and the conversion rate is mostly reporting that selection back to you.
The trap this creates
The obvious move on seeing a chart like the one above is to shift budget toward the channels at the top. It is also frequently the wrong move, and the reason is worth being precise about.
Email converts at 6 percent because the list exists. Branded search converts well because people know the brand. Something built that list and taught people the name, and it was often the cold, badly converting channels at the bottom of the chart. Cutting them to fund the top of the chart works for a quarter and then the top of the chart quietly shrinks, because you stopped feeding it.
- 01What does a conversion cost from this channel, rather than what rate does it convert at
- 02Is this channel creating demand or harvesting demand something else created
- 03What happens to branded search volume when this channel is paused for a month
- 04Does this channel's traffic reach step 2 of the funnel, or bounce at step 1
The last one is the most useful diagnostic we know and the least used. A channel that converts poorly but gets people deep into the funnel has a page problem. A channel that converts poorly and loses everyone at the first step has a targeting problem, or is sending traffic that was never going to buy. Those two situations look identical in a conversion rate and completely different in a funnel.
That is the case for segmenting the funnel itself by source rather than comparing final conversion rates. Put a price on every funnel step covers the arithmetic, and why your ad platform and your analytics disagree covers why the channel attribution feeding these comparisons is shakier than it looks.
One practical instruction
Stop reporting a blended conversion rate as a headline number. If it must exist, put it next to the per-channel breakdown so nobody mistakes it for a description of anything real. A blended figure that moves 0.4 points month over month is usually reporting a change in traffic mix rather than any change in how well your site works, and teams routinely take credit or blame for exactly that.