The cheapest revenue available to most brands is the visitors they already pay for who do not buy. This guide is for anyone considering more traffic spend. It covers reading the funnel, why device and channel splits matter more than the overall rate, and how Core Web Vitals connect to money.
Read the drop offs, not the total
An overall conversion rate is a single number averaged across every kind of visitor, and it hides everything useful.
The funnel breaks the journey into stages and shows where people leave. That matters because the stages have different causes and different fixes. Visitors leaving on the landing page usually means a mismatch between the ad and the page. Leaving on the product page usually means missing information, price or delivery expectations. Leaving in the cart usually means shipping cost appearing late. Leaving in the checkout usually means friction: forced accounts, too many fields, or a payment method people wanted and did not find.
Find the largest drop off relative to what is normal for that stage, rather than the largest drop off overall. Most people leave between landing and product, and that is true everywhere.
Split by device before concluding anything
Mobile and desktop convert differently for almost every brand, usually by a wide margin, and the blended figure describes neither.
If mobile is most of your traffic and converts at half the desktop rate, your blended rate is mostly a mobile rate with a flattering desktop average applied. The improvement available is on mobile, and it is often mechanical: tap targets, form length, how much scrolling separates the price from the buy button.
Split by channel as well. Paid social traffic arrives with less intent than brand search traffic and converts lower for that reason alone. Comparing them against a single site wide target makes paid social look broken when it is behaving normally.
Core Web Vitals are a revenue metric
Load performance sits in this section because it belongs to conversion rather than to engineering.
The relationship is well established: slower pages convert worse, and the effect is largest on mobile connections, which is where most traffic now arrives. It is also the one improvement that helps every visitor rather than a segment.
Largest contentful paint is the one worth watching first, because it measures when the page becomes useful rather than when it technically began loading. If it is poor on your landing pages, that is a conversion problem being reported as a technical one.
Page level leaks
The landing page report shows which pages receive traffic and which lose it.
Look for pages with substantial traffic and unusually poor conversion relative to your average. Those are leaks, and they are worth more than a general site improvement because the traffic is already arriving and already paid for.
A common finding is a page that ranks well organically for a term with weak purchase intent. It brings volume that was never going to buy, and it drags the site average down without being a problem to fix. Knowing that is worth as much as finding a real leak, because it stops you optimising something that is working as intended.
The order to work in
Fix the largest device gap first, because it usually affects the majority of your traffic. Then the largest stage drop off relative to normal. Then page level leaks. Then load performance, unless it is bad enough to be causing the first three, in which case it comes first.
Only after that does more traffic make sense. Buying visitors for a funnel with a known leak is paying full price to fill a bucket you already know has a hole.