Optimisation: one number, one change, and an honest comparison
Making something work better, one measured change at a time. The hard part isn't the testing. It's choosing the number to improve, because the easy ones lie.
- 095Optimisation
- The process of making something as effective or functional as possible, commonly used in email, SEO and website performance.
- Done properly
- One change at a time, read against a control or a fair before and after
- Judge email by
- Placed orders and revenue per recipient, not opens
- Judge the site by
- Conversion rate and average order value, read together
- Fix first
- Anything broken, then anything missing, then what already works
Optimising the wrong number is worse than not optimising
Every tool will happily help you improve a number. Pick the wrong one and you'll get better at the wrong thing. Open rates went soft once Apple started opening mail for its users, so a subject line can win on opens and lose on orders. A pop-up can win more sign-ups with a bigger discount and hand the margin away on every first order.
| Where | Tempting number | Number to judge it by |
|---|---|---|
| Campaigns | Open rate | Placed orders per recipient |
| Flows | Click rate | Revenue per recipient across the whole flow |
| Pop-ups | Submit rate | How many of those subscribers go on to buy |
| Product pages | Conversion rate alone | Conversion rate and average order value together |
| Search | Rankings | Organic visits that sign up or buy |
Fix what's broken before testing what works
There's an order to this. Testing subject lines on a list that lands in spam is rearranging the furniture in a flood.
- Fix what's broken: deliverability, flows that don't fire, a checkout that fails on some phones
- Build what's missing: the flows and segments a shop your size should already have
- Then test what exists, one change at a time
The first two steps are duller than a clever test, so they get skipped. Skip them and every test after is measured on a broken base.
How to run a fair test in Klaviyo and on the site
Change one thing. Split the audience at random. Decide before you look what result would make you switch, and let it run until both sides have enough orders to mean something. On a small list, a quick test mostly tells you about luck. Here's how big a Klaviyo A/B test needs to be.
On the site, the same rules hold with more traffic and more noise. That's the job of conversion rate optimisation: one change at a time, read against a control, kept only if it wins on the number that pays.
Related terms
FAQs
What's the difference between optimisation and A/B testing?
A/B testing is one tool for optimisation: two versions, a random split, a winner. Optimisation is the whole habit of choosing what to improve, changing it and checking the result. Plenty of it needs no test at all, like fixing a flow that stopped sending or a checkout button that fails on some phones.
How often should you review your email flows?
Check the numbers monthly, and give each core flow a proper review once or twice a year, or after any big change to your range, prices or brand. Flows age quietly: products sell out, offers expire, links break. A flow nobody's looked at in a year is a fair bet to be selling something you no longer stock.
What should an online shop optimise first?
Whatever's broken, then whatever's missing. Check that your emails reach the inbox and your core flows are live and firing, then fill the gaps: welcome, abandoned basket, post-purchase, winback. Only then test subject lines, layouts and offers. Testing the details of a programme with holes in it just measures the holes.
Rather we explained it on your own account?
Bring your Klaviyo account to a growth consultation. We'll walk through what this means for your numbers, in plain English, and what we'd fix first.
Half an hour with Theo, our founder.
We're all about relationships built on trust, mutual respect and a shared vision for success. If that sounds like your vibe, let's make some waves together 🌊
