For two years I managed conversion rate optimization accounts at a US performance marketing agency. My job sat between Shopify brands, designers, developers and data. I helped stores find where they were losing people and what to test next.
I wanted to write down what stuck with me before it blurs into "experience." None of this is a framework. These are just the things I'd tell a friend who runs an online store.
I didn't start in eCommerce. I started in music, and I was obsessed! diploma in music business, a very clear picture of myself as the next great artist manager.
Then I looked closely at what that industry asks of people. And the other side of me, the one rooted in wellness and softness, couldn’t keep up with that rhythm. I loved and still LOVE music. I didn't love the idea of burning out for the industry itself. So I changed my life completely. I wanted music to stay my passion, not become my boss. So I did!
What surprised me was that the work I was walking into wasn't about one industry. Understanding why people say yes, where they hesitate and what makes them leave works the same whether you're selling concert tickets or skincare. Universal knowledge, it turns out, travels well.
1. Most of the job is translation
I expected CRO to be about numbers. It's mostly about translating. Data has to become a design brief. A design has to become a developer ticket. A test result has to become a decision a founder can make in two minutes.
The best insight is worthless if the person who has to act on it doesn't understand it. I learned to write every recommendation as if the reader had 30 seconds and no context, because usually they did.
2. A losing test is still a result
Most tests don't win, and that's normal. The real failure is a test nobody learns from.
When a test lost, the useful question wasn't "what went wrong?" It was "what did we just learn about these customers?" Sometimes the answer was that the thing we thought mattered didn't matter at all. That's worth knowing before you spend money redesigning around it.
One that stuck with me: a brand was heading into a big sales period, and we tested banner copy, our version against the client's. (Banner tests are a pain, by the way. Sometimes the second version pops in a few seconds after the first, and that flicker alone adds friction.) The client's copy won. Fair enough: they know their brand and their customers better than any agency ever will, and the test proved it.
And that's the point. A test isn't a battle between the agency and the client. It's a way to take a chance on a hypothesis and let customers decide. The win isn't whose copy came out on top. It's what you learned about what matters to the people buying, and sharing that so the next decision is better.
3. Boring fixes beat big redesigns
Everyone wants the redesign. It's exciting and it looks like progress. But the changes I saw make the most difference were usually small and unglamorous: clearer product information, fewer surprises at checkout, pages that load faster on a phone, a size guide people can actually find.
A redesign changes everything at once, so when results move, you don't know why. Small fixes tell you exactly what worked.
4. Look at your store on your own phone
It sounds too simple, but most visitors are on mobile, and most store owners review their site on a laptop. Some of the most useful audits I saw started with someone holding a phone and trying to buy something, slowly, like a stranger would.
If you do one thing after reading this, do that. Notice every moment you have to pinch, scroll back or guess.
5. Data needs a question first
Analytics tools will give you hundreds of numbers. Without a question, they mostly create anxiety. "Why did conversion drop?" is a question. "Let's look at the dashboard" is not.
The habit I kept: write the question first, then decide which number answers it. Ignore the rest until you have a new question.
6. Trust comes from saying "I don't know yet"
Clients don't need you to have every answer. They need to know you'll tell them the truth, including when a test is inconclusive, when an idea didn't work, or when the honest answer is "we need more traffic before we can tell."
The relationships that lasted were the ones where nobody had to pretend.
What I'm taking with me
If I had to sum up two years in one idea, it's this: optimizing a store is like hosting a dinner at your place. Every moment of the night counts, from the invitation to the goodbye.
The invitation (your ads and emails): it's clear what the night is about, and people want to come.
The door (your homepage): guests feel welcome the second they walk in.
Finding their way (navigation): nobody has to ask where anything is.
The food (product pages): clear prices, descriptions people can actually read, photos that make them hungry.
The table (the cart): no clutter, nothing that makes them second-guess.
The bill (checkout): no surprises at the end of the night.
That's what people in my field call the funnel. Optimizing it just means walking through the night as a host, step by step, and fixing whatever gets in the way of a good time. Do it well and they'll want to come back for the next dinner. That's most of what CRO is: being a good host.
I'll keep writing about what I've learned working on online stores: what helps, what doesn't, and what I'd do differently. If that sounds useful, you can follow along on LinkedIn.