Conversion rate optimisation strategies: the ones worth actually running.

I'm Lauren Pearson, and most conversion rate optimisation strategies lists read the same: run more tests, add urgency, add social proof, repeat. Some of that is right. What these lists usually skip is the order the fixes need to happen in, and that sequencing is usually the difference between a CRO programme that moves revenue and one that produces a folder of inconclusive test reports. Here are the eight I actually use with clients, roughly in the order I'd run them, and the mistakes that quietly waste the effort.

Why this matters. Fixing the wrong thing first wastes traffic you have already paid for.

The Baymard Institute's meta-analysis of more than fifty independent studies, covering over 4,000 ecommerce checkouts, puts the average cart abandonment rate at 70.22%. That figure is not really a persuasion problem. Most of that loss happens because a form asked for too much, a cost appeared too late, or a page took too long to load, not because the copy failed to create enough urgency. Teams that jump straight to persuasion tactics, countdown timers, pop-ups, extra testimonials, before checking for these structural leaks are polishing a bucket that still has a hole in it.

That is the logic behind the order below. Fix what is actively losing visitors first. Test and refine second. Personalise last, once there is enough traffic and enough of a working funnel to make it worth the engineering cost.

The list, with how to apply each. Eight conversion rate optimisation strategies, in order.

1. Find the leak before you fix the message. Pull your funnel by step, landing page to form start to form complete, or product page to cart to checkout, and look for the single biggest percentage drop between two adjacent steps. That step is where you start. Everything else on this list is more effective once you know which step you are actually fixing.
2. Treat page speed as a conversion lever, not an IT ticket. Google and Deloitte's "Milliseconds Make Millions" research, based on more than 30 million sessions across 37 leading retail, travel and luxury sites, found that even a 0.1 second improvement in mobile load time lifted progression through the purchase funnel, with retail conversions up 8.4% and average order value up 9.2% in the sites studied. Speed is not a nice-to-have sitting on a developer's backlog. It is a conversion input with a measurable return, and it is usually cheaper to fix than anything else on this list.
3. Strip friction out of checkout and lead forms. Against that 70.22% average abandonment figure, the recoverable ground is mostly mechanical: fewer required fields, a guest checkout option, upfront delivery or setup costs instead of a surprise at the final step, and autofill that actually works. None of this requires a redesign. It requires counting every field and cost you are asking a visitor to absorb and removing every one that is not strictly necessary.
4. Run one variable at a time, and let the test finish. I will not call a test before it has run a full business cycle, at least one full week including a weekend, and reached a sample large enough that the result would not flip if you reran it. A mid-market operations-software client once queued three changes into the same test window: a new headline, a repositioned call to action, and a shortened form. Isolating them cost an extra two weeks, but it was the only way to learn that the form length was the change that actually moved behaviour. The headline and the button position, on their own, made no measurable difference.
5. Match the call to action to where the visitor actually is. A first-time visitor reading a blog post is not ready for "book a demo"; a returning visitor who has opened your pricing page three times might be past ready for it. Confusing a macro conversion with a micro one, treating every page like it should ask for the sale, is one of the most common reasons a CTA underperforms. I go through the distinction properly, and how to report on both without losing the difference between them, in macro versus micro conversions.
6. Put trust signals at the exact point of hesitation, not scattered everywhere. A security badge at the top of a homepage does very little. The same badge next to a card number field, or a client logo next to a price, addresses the specific doubt a visitor has at that specific moment. Audit where visitors actually pause, not where a template happens to have space for a logo strip.
7. Personalise only once a segment earns the engineering cost. AI-driven personalisation can lift conversion meaningfully when it is aimed at a genuine, sizeable segment, returning visitors, a specific traffic source, a known industry. I do not build a personalised variant for a segment under roughly 10% of monthly traffic; below that, the build and maintenance cost is rarely repaid before the segment's buying patterns shift again anyway. Get the default page working for everyone first.
8. Fix what happens after the click, not just the page. A form that submits cleanly still fails as a conversion if the lead sits unrouted for two days. CRO does not stop at the thank-you page. The speed and structure of what happens next, lead routing, first response time, follow-up cadence, is a conversion factor in its own right, and it is one most CRO reviews never look at because it lives in the CRM rather than the website. Our approach to CRM automation and workflow covers the mechanics of getting that handoff right.

Where teams go wrong. The mistakes that undo the list above.

The first mistake is running several of these at once and calling the result "CRO improved this quarter" without knowing which change did the work. That confidence disappears the first time you need to decide what to try next.

The second is stopping a test the moment it looks like it is winning. Early results swing hard in both directions before a sample stabilises; a test that looks like a 20% lift on day two is very often close to flat by the end of week two.

The third is personalising too early, before there is enough traffic in a segment to justify it, which spends engineering time on a variant almost nobody sees.

The fourth, and the one I see most often with founder-led teams, is treating CRO as a one-off project rather than a standing programme. A funnel that was fixed properly in January will have new leaks by the third quarter, a new product line, a new traffic source, a pricing change. Revisit the funnel audit at that same regular interval you use for anything else that drifts, rather than waiting for conversion to visibly drop before looking again.