Convert Lane

Chapter 1 of 11

What Ecommerce CRO Actually Is (and Isn't)

Conversion optimization is not a list of hacks. It is a loop that finds what is broken, proves the fix with a test, and ships it. The honest version.

Contents
  1. Why conversion rate is the wrong number to worship
  2. Macro and micro conversions
  3. The three numbers worth a dashboard
  4. What CRO is not
  5. The honest math of a 10% lift
  6. What the work actually costs
  7. Benchmarks, and why to distrust them
  8. Where this sits next to everything else
  9. How this course works

Most people arrive at conversion rate optimization holding a list. Twenty-seven tricks. Nine psychological triggers. Forty things to add to your product page this week.

Throw the list away. It is the least useful artifact in this discipline, and it is the reason most stores that “do CRO” get nothing for it.

Conversion rate optimization is a research loop. You find out what is actually broken on your store, you form a hypothesis about why, you test the fix if the fix is testable, and you ship it if it wins. Then you do it again. The loop is boring, which is precisely why it works and why so few stores run it.

This chapter is the ground floor. What the number means, what it does not mean, and what you are actually signing up for.

Why conversion rate is the wrong number to worship

Conversion rate is orders divided by sessions. That is it. It is a ratio, and ratios move for reasons that have nothing to do with your store getting better.

Run a Black Friday sale and your conversion rate climbs. Not because your product page improved. Because you discounted. Buy cheap traffic from a broad interest audience and your conversion rate falls. Not because anything broke. Because the denominator filled with people who were never going to buy.

Here is the failure mode we see most often. A store’s conversion rate drops from 2.4% to 1.9% over a quarter. Panic. Somebody redesigns the product page. Meanwhile revenue is up 40%, because the marketing team found a channel that sends five times the traffic at half the intent. The conversion rate went down and the business got better.

The reverse happens too, and it is worse. A store cuts its worst-performing ad set. Traffic halves. Conversion rate jumps from 1.8% to 3.1%. Everyone congratulates the optimizer. Revenue is flat.

Conversion rate is a diagnostic, not a goal. The goal is revenue per session, or better, profit per session. When you change something on the site and revenue per session goes up, you did work. When conversion rate goes up and revenue per session goes down, you probably just made it easier to buy the cheap thing.

Track both. Always. If a test lifts conversion rate 8% and drops average order value 12%, that test lost, and every tool in this space will happily show it to you as a winner.

Macro and micro conversions

A macro conversion is the thing you get paid for. On an ecommerce store, that is a completed order.

A micro conversion is a step toward it. Add to cart. Begin checkout. Email captured. Account created. Size guide opened.

Micro conversions matter for one reason: they let you find where the leak is. A store with a 2% conversion rate and a store with a 2% conversion rate can be broken in completely different places. One is losing people on the product page. One is losing them at the shipping step. Same headline number, opposite problem, opposite fix.

The chain looks like this on most stores:

StepTypical rangeWhat a low number here usually means
Session to product page view60–85%Collection pages or search are failing
Product page to add to cart5–12%The page is not answering the buying question
Add to cart to begin checkout45–70%Cart friction, or surprise costs appearing
Begin checkout to order45–65%Checkout fields, payment options, or shipping cost

Those ranges are wide on purpose. Treat them as a smell test, not a target. A supplement store with a subscription offer and a furniture store with a $2,000 average order do not belong in the same table, and anyone who tells you otherwise is selling a benchmark report.

The point of the chain is comparison against yourself. Where does your funnel drop hardest relative to your own other steps, and relative to your own last quarter? That is the honest question. Chapter 2 is entirely about answering it.

The three numbers worth a dashboard

Most ecommerce dashboards have forty tiles and get read by nobody. Three numbers, checked weekly, split by device and by new against returning, will tell you more than any of them.

Revenue per session. Orders divided by sessions, multiplied by average order value, or simply revenue divided by sessions. This is the number every test should be judged against, because it is immune to the trade where conversion rate rises and order value falls. If you keep one number, keep this one.

Conversion rate, by step. Not the headline figure. The chain above, so a change tells you where it happened. A site-wide conversion rate moving from 2.1% to 2.3% is noise until you can say which step moved.

Average order value, watched as a guardrail. Not a goal in itself. Its job is to catch the tests that lift conversion by making the cheap thing easier to buy.

Two things people put on dashboards that are mostly theatre: bounce rate, which on a modern analytics setup measures almost nothing useful about ecommerce intent, and time on page, which is equally consistent with deep engagement and total confusion.

Check weekly. Do not check daily. Daily variance on any store below eight figures is large enough that you will see patterns which are not there, and you will act on them.

What CRO is not

Three things get sold as conversion optimization. None of them are.

It is not a list of best practices. “Add urgency.” “Use social proof.” “Reduce friction.” These are categories, not instructions. Adding a countdown timer to a store whose problem is unclear shipping cost does nothing except make the store look cheaper. Replacing an invented countdown with a real stock count often does better, and when it does, the mechanism is not urgency. It is credibility.

It is not a redesign. Redesigns change a hundred variables at once. When the number moves you learn nothing, because you cannot attribute it. When the number drops you cannot roll back the one thing that hurt, because there is no one thing. Redesigns are sometimes necessary. They are never optimization.

It is not testing everything. Testing has a cost measured in traffic and time. Most stores do not have enough of either to test the things they want to test. Knowing when not to test is a skill, and Chapter 3 spends real time on it. If your store does 300 orders a month, most of your wins will come from fixing obviously broken things and shipping them without a test. That is fine. That is the correct call.

The honest math of a 10% lift

Optimization gets sold with compounding math. Improve conversion 10% a month for a year and you have tripled your business. It is arithmetic, and it is a fantasy.

Here is what actually happens. A store doing $4M a year at 2.1% conversion runs a serious testing program for twelve months. Say twenty-four tests. Historically, across the industry, roughly one in five to one in six tests produces a real, holding win. So four wins. Winners in ecommerce cluster in the 3–8% range on the metric they targeted, not 30%.

Four wins, averaging 5%, compounding imperfectly because they overlap and interact:

  • Naive compounding: 1.05^4 = 21.6% lift
  • What you actually keep after interaction effects and regression: call it 12–15%

On $4M, a 13% lift is roughly $520,000 in additional revenue. That is a very good year of CRO work. It is not a tripling.

Two things follow from that math. First, the wins are real and worth chasing, especially since most of them cost nothing to keep once shipped. Second, anyone promising you a doubled conversion rate is either lying or planning to discount your catalog and call it optimization.

The other thing that follows: at $400k a year in revenue, a 13% lift is $52,000. Whether that pays for a testing program is a real question with a real answer, and the answer is often no. We will come back to this.

What the work actually costs

Nobody publishes this, so here it is.

Research is the cheapest part and the highest return. Two to three weeks of somebody’s attention, once, gets you a problem list that stays useful for a year. It needs no tooling beyond your analytics and a spreadsheet. If you do exactly one thing from this course, do the research in Chapter 2.

Testing is the expensive part, and the cost is mostly not money. A testing tool is $0–500 a month at the volumes most stores operate at. The real cost is traffic and calendar. A single properly-powered test occupies a page for two to six weeks. That means a store running one test at a time gets roughly eight to twelve tests a year, of which one or two win.

Implementation is where the money goes. Developer time to build variants, and more developer time to ship winners properly rather than leaving them running as permanent experiments. Budget more here than you expect, because a variant built as a quick hack becomes a permanent quick hack.

The rough shape of a year, for a store doing $4M:

TimeDirect cost
Research, done once properly2–3 weeksNear zero
Fixing what research found3–6 weeks of devDev time
Testing toolOngoing$0–6,000/yr
10–14 tests, built and analysedMost of the yearDev time
Shipping the 2–3 winners properly2–3 weeksDev time

Notice the ratio. Most of the value in year one is in the second row, not the fourth. Stores that skip research and go straight to testing invert that ratio and wonder why the programme is not paying.

Benchmarks, and why to distrust them

You will be shown a chart. Average ecommerce conversion rate is 2.5%, or 1.8%, or 3.4%, depending on whose chart it is. Your store is at 1.6%. Therefore you have a problem.

Maybe. Probably not, though, because that number is an average across a sample you know nothing about. Consider what moves conversion rate before any optimization happens at all:

  • Price point. A $18 consumable converts at multiples of a $900 mattress. Neither store is better run.
  • Traffic mix. Branded search converts 5–10x cold paid social. A store that is 60% branded search has a flattering number and a growth problem.
  • Category. Repeat-purchase categories carry returning customers in the denominator. Considered purchases do not.
  • Device split. Mobile converts at roughly half of desktop on most stores. A store that is 80% mobile looks worse than an identical store at 50%.
  • Return policy and shipping threshold. Free returns and free shipping change the number materially, and they are business decisions, not site decisions.

Use benchmarks for exactly one thing: sanity-checking whether you are in a plausible band. If your checkout completion rate is 12%, something is broken, and you do not need a benchmark report to know it. If it is 58%, stop reading charts and go look at your own funnel.

Your only useful benchmark is your own store last month, segmented by device and traffic source. Everything else is content marketing.

Where this sits next to everything else

Conversion work does not happen in isolation, and a lot of what looks like a conversion problem belongs to somebody else.

Against paid acquisition. These two are the same job viewed from different ends. A conversion improvement lowers your effective cost per acquisition on every channel simultaneously, which is why it compounds with media spend rather than competing with it. It is also why a store with a broken product page should fix the page before raising the ad budget: you are paying to send more people into the same leak.

Against SEO. Largely complementary, occasionally in tension. The tension is real when a page is ranking for a query it converts badly on, and the fix that helps conversion would hurt the ranking. When that happens, work out what the page is worth in each direction before you touch it. Usually the answer is to build a second page rather than compromise one.

Against product and pricing. This is the one people avoid. If your product is priced wrong for the market, or the offer is weak, no amount of page work fixes it. Conversion optimization can make a good offer perform to its potential. It cannot make a bad offer good, and the most useful thing research often produces is evidence that the constraint is the offer.

Against brand. Frequently framed as opposed, mostly not. The conflicts are real but narrow, and they are usually about a specific element rather than a direction. A brand team that will not put a price above the fold and a conversion team that wants nothing but price are both wrong. Where a genuine conflict exists, it is a business decision about time horizons, not something a test can settle.

The short version: conversion work multiplies whatever else is true about your business. That is why it is worth doing, and why it cannot rescue a business where the other things are not true.

How this course works

Eleven chapters, in order, because the order is the method.

The first two chapters are foundations: this one, and research, which is how you find out what is actually wrong instead of guessing.

Chapters three and four are the method itself. Turning findings into hypotheses worth spending traffic on, then running the test and reading the result honestly including the statistics, which are less complicated than they are made to sound and matter more than most operators think.

Chapters five through seven are the surfaces where ecommerce money is made and lost, in rough order of leverage: the product page, the cart and checkout, and landing pages for paid traffic.

Chapter eight is the part almost nobody writes about, which is everything that happens after the order is placed.

Every chapter has at least one section on what does not work, or what we have shipped that lost. That is deliberate. A course that only shows winners is a course written by someone who is not measuring, and you should not trust it.

The whole thing is free and there is no gate. If you would rather have it arrive one chapter a week instead of reading eight long articles in a sitting, there is a form for that below.

We run a conversion agency. Some people who read this eventually hire us to do it instead of doing it themselves. That is the entire commercial arrangement, and you can read what that engagement actually looks like if you are curious. Otherwise, ignore it and take the method.

Next: how to find out what is actually broken.

Not sure if CRO is even the right work for your store?

That is the honest question, and worth twenty minutes. We will look at your traffic and tell you if optimization pays here, or if the constraint is upstream and testing would just measure the problem more precisely.

  • 437+ stores built
  • 2,000+ tests shipped
  • No call needed to start

Claim your free audit of one page

Send us one page of your store and we’ll review it on video: what’s leaking, what to fix first, and what to leave alone. Changes you can ship this week to lift CVR, AOV and revenue per visitor.

Get My Free Audit