In short
Learning goal: by the end of this lesson you have your own store's step rates, know which step loses the most money, and have one north-star goal for every test that follows.
Who this is for: Shopify brands doing roughly $1M to $10M a year, with a growth lead or a small team who will run tests every month.
Why start with a map
Most testing programs start with a list of ideas. The trouble is that ideas cluster around the pages people look at most (the homepage, the hero banner) rather than the steps where shoppers actually leave. A funnel map fixes the order: measure every step first, then test where the loss is large and the traffic is enough to detect a change.
The funnel in this course has five steps, and every lesson follows them in order:
- Landing: the first page a visitor sees, often from an ad or an email.
- Product: the product page, where the decision is made.
- Cart: where the order is assembled and the order value is set.
- Checkout: contact, shipping and payment.
- Post-purchase: the thank-you page and the second order.
What to measure at this step
| Metric | Definition | Use it as |
|---|---|---|
| Revenue per visitor (RPV) | Total revenue in a period divided by all visitors. Equals conversion rate times average order value. | A good primary goal for most tests |
| Profit per visitor | Gross profit (revenue minus product costs, after discounts) divided by all visitors | The best primary goal when product costs are known |
| Step rate | Share of visitors who reach each step, per 100 sessions | A diagnostic: shows where shoppers leave |
| Checkout completion | Orders divided by checkouts started | A diagnostic, and a guardrail for any change near checkout |
| Average order value (AOV) | Revenue divided by orders | A diagnostic: explains an RPV change |
A step rate tells you where shoppers leave. It never tells you whether a change made more money. That's what the primary goal is for, and it is chosen before a test starts. The revenue per visitor vs conversion rate article explains why the two can move in opposite directions.
Choose profit per visitor as the north star
At your size, many of the tests worth running change what a shopper pays: an offer format, a free-shipping threshold, a bundle. A discount can raise conversion rate and lower revenue per visitor. It can lower profit per visitor even more, because the discount comes straight out of margin. If the program's north star is conversion rate, the program will slowly learn to give money away.
Profit per visitor puts every test on the same scale: what each visitor earned you, after product costs and discounts. It needs product costs for most of what you sell. Revenue per visitor is the fallback when costs aren't loaded yet.
One rule follows from this: every test in the course has one primary goal, and it is profit per visitor or RPV unless there is a stated reason. Step rates and add-to-cart rates are diagnostics. Checkout completion, refunds and page errors are guardrails.
Common problems with funnel data
- Steps that count different things. Sessions, cart adds, checkouts and orders are counted in different units. Put each against sessions ("per 100 sessions") so they're comparable, and use checkout completion as the one like-for-like step rate.
- Orders without a visit. Point of sale, draft orders, subscription renewals and orders from other channels appear in Shopify but not in a storefront funnel. That gap is expected. Part 2 of the BFCM series, the tracking audit, lists the gaps you should expect.
- Buy it now. Shoppers who use an accelerated checkout button skip the cart. A funnel that requires every step undercounts them.
- Averages that hide the mix. Mobile and desktop, new and returning, paid and email traffic often have very different step rates. Look at the split, but treat it as a description, not a result.
One store's funnel, and what each step can detect
Illustrative inputs, not data from a real store. Detectable changes computed with the same sample-size formula Benchmyrk's planner uses (two-sided, 95% confidence, 80% power, two versions, even split, 21 days).
Inputs: 8,000 visitors a day; for every 100 sessions, 62 view a product, 14 view the cart, 5.8 start checkout and 2.6 place an order.
| Step | Visitors a day who reach it | Rate of those visitors who buy | Smallest change a 21-day test on this step can detect |
|---|---|---|---|
| Landing (everyone) | 8,000 | 2.6% | 8.6% |
| Product | 4,960 | 4.2% | 8.5% |
| Cart | 1,120 | 18.6% | 7.8% |
| Checkout | 464 | 44.8% | 6.4% |
Three readings of this table:
- The biggest loss is between product and cart. 62 of 100 sessions see a product and only 14 reach the cart. That is where the next lessons start.
- Deeper steps have fewer visitors but higher rates. A test that only checkout shoppers enter can still detect a small change, because nearly half of those shoppers buy. A test on the cart page sees fewer visitors than a product-page test, but can detect a similar change.
- These are conversion-rate figures. A test judged on profit per visitor needs more visitors, because order values vary a lot. The sample size and test duration guide shows how much more.
Test ideas this map points to
You won't test anything in this lesson, but the map should already suggest where to look. Each idea comes with the mechanism behind it, because a test without a mechanism can't teach you anything when it loses:
- Product to cart is the big drop. Mechanism: shoppers lack a reason to decide now (delivery date, returns, which variant). Lesson 3 covers it.
- Cart to checkout drops on mobile only. Mechanism: a surprise cost, often shipping, revealed late. Lesson 4 covers thresholds and progress bars.
- Landing pages from paid social have a low next-page rate. Mechanism: the page doesn't continue the ad's promise. Lesson 2 covers it.
Exercise (10 minutes, on your own store)
- Pick the last 28 days, so every weekday appears four times.
- Write down sessions, product views, cart views or adds, checkouts started and orders. Shopify's analytics or your testing tool will have them.
- Turn each into "per 100 sessions" and compute checkout completion (orders divided by checkouts started).
- Mark the step with the largest drop, and write one sentence on why shoppers might leave there.
- Write down your average order value and, if you know it, your gross margin. You'll need both in lesson 4.
You don't need Benchmyrk for any of this.
Do it in Benchmyrk
Check yourself
- Add-to-cart rate rose 9% in a test and revenue per visitor didn't change. Is it a winner? No. Add-to-cart rate is a diagnostic. The primary goal decides, and it didn't move.
- Why put every step against sessions instead of against the step before? Because the steps count different things (sessions, adds, checkouts, orders). Per 100 sessions keeps them on one scale. Checkout completion is the one step rate that compares like with like.
- Your funnel shows fewer orders than Shopify. Is tracking broken? Not necessarily. Point of sale, draft orders, subscription renewals and other channels have no storefront visit. Check the gaps are the expected ones and stable from week to week.