yodai Academy Analytics

Academy › Publishing › Analytics

One change a month: the one that loses the most people, not the one with the highest percentage

By the end, you'll know which stage to fix this month, the one that loses the most people in raw numbers, not percentage, and you'll have written down the one change you're going to test and the date you'll check it.

By Yodai · Published September 6, 2026 · Updated October 12, 2026

4 min read

When you measure the stages of your site (how many people come in, how many reach the product page, how many add something to the cart, how many pay), every stage has its own drop-off, and you end up with four numbers of loss staring back at you and just one month to fix something. If your analytics tool doesn't track one of those steps, work with the ones it does track: the approach is the same.

The question isn't what's broken, several things probably are. The question is what you fix first. And that's where people waste the most time: staring at the stage that looks worst on the chart, not the one that's actually costing you the most real customers.

There's one test that settles this in five minutes, and it doesn't take any more data than you already have.

Percentage misleads, the number doesn't

Picture a month for a store with 6,000 visits. From the product page to the cart, 2,700 people move forward, 1,800 drop off along the way, 40%. From starting checkout to actually paying, which is where shipping and other costs show up, 1,100 people make it through, 1,200 drop off, 52%.

52% sounds worse than 40%. But in actual number of people, the cart stage loses more: 1,800 versus 1,200. That's a 600-customer difference every month, not some statistical nuance.

If you can only fix one stage this month, fix the cart. Not because the percentage says so: the number of real people you're losing does.

People who drop off at each stage every month (out of 6,000 visits)
Visit → product page1.500
Product page → cart1.800
Cart → start checkout400
Start checkout → pay1.200
Why the percentage alone isn't enough
Sort by percentage
  • The stage that looks worst on the chart (here, actually paying: 52%).
  • It can lead you to fix a step with fewer real people behind it.
  • It's useful for comparing stages with a similar amount of traffic.
Sort by number of people
  • It tells you exactly how many customers you lose at that step every month.
  • It's the number that moves your revenue, not the one that moves the chart.
  • It's useful for deciding between stages with very different volumes.

Which one to pick: Go by the number of people almost every time, because funnel stages rarely carry the same volume. Percentage only breaks the tie when two stages move the same number of people and the fix costs you the same effort on both.

One change, not five

You already know which stage to fix. The temptation, once you have the whole list in front of you, is to ride the momentum: fix the shipping cost too, change the text on the checkout button, try a new photo on the product page. All in the same month, because why not while you're at it.

If you do, next month sales will go up, go down, or stay the same, and you won't know why any of those three things happened. You won't be able to repeat what worked somewhere else in your business, or stop wasting time on what didn't do anything at all.

One change, one month, one number you check at the end: it's the only way to learn something that'll actually help you next month, and the month after that.

Your monthly routine
  1. 1Write down the drop-off for each stage in people
    Use last month's actual numbers, not the percentages.
  2. 2Rank them from biggest to smallest loss
    The one at the top of the list is the stage you fix this month. The rest wait.
  3. 3Pick a single change for that stage
    With a clear reason why you think it'll cut the drop-off.
  4. 4Set the date and the number you'll check
    Usually 30 days out, looking only at that stage, nothing else.
  5. 5Don't touch another stage until that date
    Even if you spot another big drop-off, wait. If you change two things at once, you won't know which one worked.
Changing five things in the same month
It's the most expensive mistake, and the most tempting one: changing the cart, the shipping, the button, the photo, and the email copy all at once because you're already in there. If sales go up, you won't know which of the five changes did it, so you'll end up keeping all five, some for no real reason, instead of keeping only the one that worked and applying that same thinking somewhere else.
The note you write down before you touch anything
Month: October. Stage: product page to cart (loses 1,800 people a month, the biggest drop of the four). Change: I'm adding the exact measurements to the product page and swapping the photo for one next to a familiar object, so people can see its real size. Hypothesis: some people aren't sure it'll work for them and won't risk adding it to the cart. Check back: November 1, looking only at how many product page visits turn into a cart.
What you'll walk away with
  • You know which stage loses the most people in raw numbers, not percentage, and it's the only one you're touching this month.
  • You've written down the change, the reason you think it'll work, and the date you'll check it.
  • You don't touch any other stage until that date, even with the whole list staring back at you.
  • You've covered the whole journey, from visit to order: you know exactly where people are slipping away and what you're doing about it every month.
NextThe Price-and-Stars Card Doesn't Come From Your Google Business Profile