Funnel reports are good at telling you percentages. Opt-in rate 35%. Sales page 4.5%. Upsell 22%. What they're bad at is telling you what any of it is worth.
So let's do it the slow way. Ten thousand people are about to click an ad for an online course. We'll follow every one of them, and every dollar, through each step of the funnel, and keep a running ledger as we go.
The funnel is a model, but a realistic one: a free guide, a $97 course, a $27 order bump, a $197 upsell and a $47 downsell. When you’re done reading, swap in your own numbers. The shape of the answer rarely changes.
Stop 1: The ad click
Ten thousand people click. At $1.20 a click, that's $12,000 spent before anyone has seen a page.
This is the only number in the funnel that’s guaranteed. Everything after it is a maybe.
Ledger: −$12,000
Stop 2: The opt-in page
The opt-in page asks for an email in exchange for a free guide. 35% say yes. That's 3,500 new leads.
The other 6,500 leave. You paid $7,800 for them, and most will never come back. This is the single biggest exit in the whole funnel, and it happens on a page that earns nothing directly. Hold that thought.
Ledger: −$12,000 (leads aren't money yet)
Stop 3: The sales page
Of the 3,500 leads, 60% go on to the sales page: 2,100 people. Of those, 4.5% buy the $97 course. That's about 95 buyers and $9,167 in sales.
It's the moment the funnel starts making money, and it still isn't enough. After the core offer, we're $2,833 short of what we spent.
Ledger: −$2,833
Stop 4: The order bump
On the order form, a $27 checklist sits in a checkbox. 35% of buyers tick it: $893.
It's a lovely number to see in a dashboard, and it barely dents the hole.
Ledger: −$1,940
Stop 5: The upsell
Right after the purchase, buyers see a $197 coaching call. 22% take it: $4,096.
This is where the funnel turns. The ledger crosses zero here, on a page that most funnel owners build last and test never.
Ledger: +$2,156
Stop 6: The downsell
Buyers who said no to the coaching call are offered a $47 mini course instead. 30% of them take it: $1,039.
Ledger: +$3,195
Stop 7: The refund desk
Over the next 30 days, 8% of the money comes back as refunds: $1,216.
Ledger: +$1,979
That's it. Ten thousand visitors, $12,000 in ads, $15,195 in sales, and $1,979 left over. Revenue per visitor after refunds: $1.40.

Three things the ledger shows that a funnel report doesn't
1. Profit is a thin slice, so small gains become big ones
The funnel turns $15,195 of sales into $1,979 of profit. That slice is so thin that any extra revenue, with the same ad spend, lands on it almost entirely.
Here's what a 10% improvement at each stop does, with everything else unchanged:
Improve this by 10% | Extra revenue | Profit becomes | Profit change |
|---|---|---|---|
Opt-in rate (35% to 38.5%) | +$1,398 | $3,377 | +71% |
Sales page conversion (4.5% to 4.95%) | +$1,398 | $3,377 | +71% |
Leads reaching the sales page (60% to 66%) | +$1,398 | $3,377 | +71% |
Course price ($97 to $107, same conversion) | +$843 | $2,822 | +43% |
Upsell take rate (22% to 24.2%) | +$350 | $2,329 | +18% |
Refunds (8% to 7.2%) | +$122 | $2,100 | +6% |
Order bump take rate (35% to 38.5%) | +$82 | $2,061 | +4% |
A 10% better opt-in page is worth 71% more profit. Not 10%. That's the multiplier hiding inside every paid funnel: your ad spend doesn't grow when your pages get better, so the whole gain falls through to profit.

2. Everything downstream inherits the top
Notice that the opt-in page, the click-through to the sales page and the sales page itself all produce exactly the same gain. That's not a coincidence. Every buyer, every bump, every upsell starts as a lead. Improve any step at the top and every step below it gets 10% more people to work with.
That’s why the opt-in page, which earns nothing on its own, is one of the most valuable pages you have. It’s also why judging it on opt-ins alone is dangerous: a headline that brings in 10% more leads who never buy isn’t a 10% gain. It’s a 0% gain with a nicer chart. Judge the top of the funnel on what it produces at the bottom: revenue per visitor, not opt-in rate.
3. The money is made after the "yes"
Look back at the ledger. After the core sale, the funnel is still $2,833 in the red. It's the order form and the post-purchase pages, the bump, the upsell and the downsell, that carry it into profit.
Those pages see far fewer people than the top of the funnel, so a test there takes longer to read. But they’re cheap to change, they don’t touch your ad account, and a better upsell offer can move revenue far more than a better upsell headline. If your post-purchase pages are the same ones you built at launch, that’s probably where your easiest money is.
Where to put your next test
Put the ledger and the leverage table together and a simple order falls out:
The sales page offer or the opt-in promise. The biggest multiplier, judged on revenue per visitor so a lead magnet that attracts the wrong people can't win.
The upsell offer. Not the button, the offer itself: what it is, what it costs, and how it follows from what they just bought.
The price of the core offer. A 10% higher price that keeps conversion close to flat is worth more than most page tweaks, and it's a test, not a guess.
Refunds. Rarely tested, quietly expensive. Better onboarding after the sale is a refund test too.
The order bump last. It's worth having, but in this funnel a 10% better bump is worth $82.
And whatever you test, watch the whole ledger, not just the step you changed. A sales page that sells more but sends fewer people to the upsell can lose money while “winning”. We wrote about that trap in why your A/B test winner didn't make you more money.
Run your own ledger
You don't need a tool for this, just your last month of numbers and ten minutes. Fill in each line:
Line | How to work it out | Your number |
|---|---|---|
Ad spend | Visitors × cost per click | |
Core sales | Visitors × opt-in rate × share reaching sales page × sales page conversion × price | |
Order bump | Buyers × bump take rate × bump price | |
Upsell | Buyers × upsell take rate × upsell price | |
Downsell | Buyers who declined the upsell × downsell take rate × downsell price | |
Refunds | Total sales × refund rate | |
Profit | Everything above, minus ad spend and refunds |
Then add 10% to one rate at a time and watch what happens to the last line. The step that moves it most is your next test.
If you'd rather not do it by hand, this is exactly what Splitlab's revenue reporting is built for: each variation's revenue per visitor counts the whole funnel, bumps, upsells and downsells included, so a test is judged on the bottom of the ledger, not the top.
My final thoughts
Following one funnel to the last dollar changed how I read funnel reports. Percentages tell you how a step performs. Only the ledger tells you which step pays for everything else.
In this funnel, the front end loses money and the upsell rescues it. In yours it might be the order bump, or refunds might be quietly eating the profit. You won’t know until you run your own numbers through every stop.
Do it once by hand. Then put your next test where a small lift moves the most money, not where the conversion rate looks worst.
Follow the money, not the percentages.
Built for revenue
Test for revenue, not just clicks.
Run A/B, multivariate, and multi-armed bandit experiments — with real revenue tied to every variation.
We’ll increase your conversion rate by 30% or you don’t pay. Terms apply.
✓
Revenue per variation
✓
A/B & MVT
✓
Multi-armed bandit
Test for revenue, not just clicks.
A/B, multivariate and multi-armed bandit experiments, with real revenue tied to every variation.


