CRO is engineering, not magic
"Conversion rate optimisation" sometimes gets pitched as button-color tests and urgency timers. The actual work is closer to product engineering: identify friction, hypothesis, test, learn.Where the actual conversion is lost
The funnel, in order of typical impact:- Page speed — 1 second extra load = ~5-10 % conversion loss
- Mobile UX — most TR e-commerce traffic is mobile
- Search and discovery — if customers can't find what they want, they don't buy
- Product page clarity — multiple images, clear specs, customer reviews
- Cart and checkout friction — every required field tests its own value
- Trust signals — reviews, returns policy, secure payment indicators
- Payment options — missing the customer's preferred method = lost sale
Page speed: still the biggest lever
- Largest Contentful Paint (LCP) under 2.5 s
- Interaction to Next Paint under 200 ms
- Cumulative Layout Shift under 0.1
Specific wins:
- Images: AVIF / WebP, responsive sizes, lazy load
- Fonts: preload, swap, subset
- CDN for static assets
- Reduce third-party scripts on the critical path
Checkout friction
- One-page vs multi-page checkout — both can work
- Guest checkout
- Auto-fill / autocomplete-friendly form fields
- Inline validation
- No surprise costs — show shipping + taxes early
- Save cart on mobile — recovery emails
Mobile-specific
- Tap targets at least 44×44 px
- Sticky add-to-cart on PDP
- Mobile wallet integration (Apple Pay, Google Pay, BKM Express)
- Don't require pinch-zoom for product images
- Auto-detect and apply discount codes from URL params
Personalisation, properly
Done well lifts conversion. Done badly, it's creepy.- Recently viewed → reasonable
- "You might also like" based on cart contents → reasonable
- Aggressive cross-channel retargeting after a cart abandon → mixed
- Personalisation that reveals knowledge of off-site behaviour → unsettling
A/B testing discipline
- Hypothesis-driven
- Sample size powered for the effect size
- Run for a full business cycle
- Don't peek and decide early
- Account for novelty effects
- Most tests fail to replicate at scale
Things that look like CRO but usually backfire
- Aggressive countdown timers
- Fake urgency ("Only 2 left!" when there are 200)
- Pop-ups for newsletter sign-up before reading
- Forced account creation
- "Free shipping" that's actually built into the price
Tactics that improve conversion at the cost of trust have a long-term negative ROI.
Analytics that drive decisions
- Funnel breakdown — drop-off at each step
- Cohort retention by acquisition channel
- LTV by channel, by product category
- Time-to-second-purchase
- Customer service tickets correlated with order patterns
