What CRO is, and honest benchmarks
Conversion rate optimisation is the practice of increasing the share of visitors who take the action you want, systematically, with evidence. The honest benchmark bands from my own funnels: B2B sites converting visitors to leads run anywhere from 1 to 10 percent, well-targeted commercial pages convert around 1 percent to purchase in consumer health, and content pages a fraction of that. The band is wide because conversion is mostly a targeting property, not a page property: the same page converts differently depending on who arrives.
That is the first working principle. Before optimising a page, check who is landing on it. My diagnostic split: impressions without clicks means the page angle is wrong, traffic without conversions means the offer or the audience is wrong. Each points at a different fix, and only the second one is CRO.
What actually moves the number
The biggest CRO win I have shipped was not a button colour. At ContactOut, the head of design and I rebuilt landing pages around the actual job the visitor came to do, and conversion went from 5 percent to 20 percent. The moves that did it: message-to-query match, one action per page, and proof near the ask. The same logic scaled at Superpower, where pages closely connected to the unique selling point converted at 1 percent while generic pages rounded to zero, which is why landing page structure is its own discipline.
Where the funnel leaks decides where the effort goes: on one campaign the single biggest preventable loss was 41 percent of people abandoning at the name-and-email step, a form problem, not a traffic problem. Finding those leaks is the CRO audit. Proving your fix caused the lift is A/B testing, and being honest when it did not is most of the craft. CRO compounds every other channel, halve the leak and you double what SEO and paid deliver, which is why it belongs inside growth strategy rather than bolted on after.
Where CRO sits organisationally matters too. As an SEO I want the traffic I win to convert, but a growth lead should not spend all week inside the funnel: the leverage is in ranking the leaks against every other experiment in the queue and buying the fix from whoever is best placed to ship it, design, engineering, or copy. Owning the number is the job. Personally turning every screw is not.
FAQ
What is a good conversion rate?
Depends entirely on what converts to what. Visitor-to-lead on B2B sites runs 1 to 10 percent, well-targeted commercial pages convert around 1 percent to purchase, content pages far less. Track your own trend against your own baseline rather than a global average.
What improves conversion rate fastest?
Message-to-query match, one action per page, proof near the ask, and shorter forms at the point of drop-off. A rebuild on those lines took ContactOut landing pages from 5 to 20 percent, far beyond what element tweaks produce.
Is CRO worth it for low-traffic sites?
The mindset is, the testing often is not: below meaningful traffic, A/B tests cannot reach significance. Fix the obvious leaks by judgment, and spend the remaining effort getting traffic.