Traffic went up and revenue did not. It is the most common paid-marketing complaint there is, and the cause is almost never the traffic. Somewhere between arriving and paying, people are stopping, and the number that tells you where is already in your analytics.
Why is my traffic up but my conversions flat?
One of three things. The traffic is the wrong traffic, so it was never going to convert. The page it lands on does not do what the ad or the result promised. Or the funnel after that page is losing people at a specific step you have not looked at. Only the third is what most people call conversion optimisation.
Take them in that order, because each one is cheaper to check than the one after it. Wrong traffic shows up as a bounce rate that is high on one source and normal on the others. A promise mismatch shows up as people leaving the landing page in under ten seconds. A funnel leak shows up as a step where the drop is much steeper than the steps either side of it.
- This funnel
- A funnel without the leak
Illustrative, not your numbers. The shape is the method: every step loses people, so a step is only interesting when it loses far more than its neighbours. Here the fall from starting checkout to entering payment is the whole problem, and nothing before it is worth touching.
Growth is a process. Nothing here happens overnight, and anybody promising you overnight results is lying to you.
Get a funnel teardown from $100What is conversion rate optimisation?
Finding the steps where people who intended to buy stop, working out why, changing that one thing, and checking the change helped. It is not redesigning a page because somebody senior dislikes it, and it is not a list of best practices applied without measuring anything.
The distinction matters commercially. A redesign changes fifty things and you learn nothing about which of them mattered. A conversion programme changes one thing at a time and ends with a shorter list of what actually works on your audience, which is an asset that keeps paying.
Should I fix the funnel before buying more traffic?
Yes, and the arithmetic is not close. Doubling traffic doubles the cost. Moving checkout completion from 11% to 20% nearly doubles revenue from the traffic you are already paying for, and it keeps doing it every month afterwards without being bought again.
This is the single most expensive ordering mistake in marketing, and it is the one paid channels quietly encourage, because more spend is easier to sell than a fix. If you are about to increase an ad budget on a funnel nobody has measured, you are buying more visitors to the same leak.
Do I need A/B testing?
Only if you have the traffic to reach significance in a reasonable time. Below roughly a thousand conversions a month you will wait months for a result, and most teams stop the test early and act on noise, which is worse than not testing.
Under that volume, watch people instead. Five moderated sessions with somebody trying to complete the purchase will find a blocking problem in an afternoon, and the fix is usually obvious once you have watched three people hesitate in the same place. Testing is for choosing between two good options. Watching is for finding out what is broken.
- Split the funnel by source before anything else One channel converting at a tenth of the others is not a funnel problem, it is a targeting problem, and no page change will fix it.
- Find the step with the steepest drop Not the biggest drop. Every funnel loses most of its people at the top. You want the step that loses more than the steps around it.
- Watch five people attempt that step Analytics says where. Only a person says why, and the why is the only part you can act on.
- Change one thing Changing five and seeing an improvement teaches you nothing you can repeat on the next page.
- Check it on mobile separately Desktop and mobile funnels routinely differ by a factor of two, and an average across both hides which one is failing.
What usually fixes it?
In our experience the recurring causes are boring and cheap: a form asking for information nobody has to hand, a payment step that fails silently on one card type, a price or a delivery cost that first appears at checkout, and a mobile layout where the button sits below something that looks like the end of the page.
None of those are design problems and none need a redesign. That is the good news in almost every teardown: the expensive-looking symptom usually has a cheap cause, and finding it is the work you are buying rather than the fix itself.