The most common mistake in paid media: optimising ad performance while ignoring landing page performance. A 1% improvement in your click-through rate will drive roughly 1% more traffic. A 1% improvement in your landing page conversion rate drives 1% more conversions from your existing traffic. Both matter equally, but landing page optimisation is systematically underinvested because the results are harder to attribute in most analytics setups.
Here's our full process, including the specific tests that generated the 42% CPA reduction referenced in the headline β for a B2B SaaS client with a self-serve trial funnel over a 6-month optimisation programme.
Phase 1: Diagnosis before hypothesis
The most common CRO mistake is running A/B tests based on intuition or competitor benchmarking rather than data about your specific users' behaviour on your specific pages. We never start with tests. We start with understanding what's happening.
Heatmaps and click maps
Hotjar (or Microsoft Clarity for budget-constrained projects) provides heatmaps showing where users click, scroll depth maps showing how far down the page they get, and cursor movement patterns. The scroll depth data alone often tells the story: if 60% of users don't scroll past the hero, either your hero needs to create more pull, or your CTA needs to be above the fold.
Session recordings
Watch 50-100 session recordings on your target landing page. Filter to sessions that resulted in no conversion. Look for rage clicks (repeated clicking on elements that aren't clickable), exit points, form abandonment patterns, and mobile friction. Session recordings are humbling β users experience your landing page very differently than you imagine they do when you're designing it.
Form analytics
If your conversion action is a form submission, install field-level form analytics. This shows you where in the form users abandon, which fields cause the most hesitation (measured by time-to-complete), and which fields are left blank most often. We've removed or made optional fields that were causing 15-20% abandonment rates β an easy win with no design work required.
Phase 2: Hypothesis prioritisation
From the diagnosis, you'll have a list of potential issues. Prioritise them using an impact/effort framework: high impact, low effort first. For our SaaS client, the three highest-priority hypotheses coming out of diagnosis were:
- The hero headline was feature-focused ("Automated project tracking") rather than outcome-focused ("Ship projects 40% faster"). Session recordings showed users reading the headline and immediately scrolling down, suggesting it wasn't creating conviction.
- The trial CTA ("Start free trial") was competing with a secondary CTA ("Watch demo") of equal visual weight, creating decision paralysis.
- The social proof section (logos of enterprise customers) appeared below the fold, so 45% of users never saw it.
Phase 3: A/B testing
Test 1: Headline reframing
Control: "Automated project tracking for modern teams"
Variant: "Ship your next project 40% faster. Without the status-update meetings."
Result: +18% conversion rate on trial signups. Statistically significant at 95% confidence after 3 weeks. This single change, extrapolated to our client's monthly traffic, was worth ~40 additional trials per month.
Test 2: CTA hierarchy
Control: Two equal-weight CTAs ("Start free trial" and "Watch demo")
Variant: Primary CTA "Start free trial" (accent button), secondary CTA "Watch 3-min demo" (text link, smaller)
Result: +11% increase in trial signups, with only a 4% reduction in demo views. The total conversion rate (trial + demo) increased because more users went directly to the primary action rather than to the secondary one.
Test 3: Social proof repositioning
Control: Logo grid below the fold
Variant: Company count statement ("Used by 2,400+ teams") and three specific customer logos moved immediately below the hero, above the fold
Result: +8% conversion rate. Moving social proof above the fold provided reassurance to users at the moment they were deciding whether to read further.
The compounded result
The three tests ran sequentially, each building on the previous. The combined effect: +42% improvement in trial conversion rate. CPA dropped from $127 to $74 on identical ad spend. This is the compound math of conversion rate optimisation β each 10% improvement compounds with the next.
The ongoing programme
CRO isn't a one-time project. User behaviour changes. Your product changes. Seasonal patterns shift intent. We run a monthly testing cycle: one hypothesis in test at all times, with results feeding directly into the next hypothesis backlog.
If you're running paid media and you haven't invested in systematic landing page optimisation, this is almost certainly your highest-ROI opportunity. Get in touch to discuss a CRO audit for your programme.
