What makes a call to action work?
A call to action (CTA) is the instruction telling someone what to do next. It is one of the highest-leverage elements on a page, and most CTAs fail for identifiable reasons.
What the research and testing consistently support:
Specificity beats generic wording. "Submit", "Click here" and "Learn more" describe the mechanic rather than the outcome. "Get my free quote" or "Start the 14-day trial" describe what the person receives. Testing repeatedly finds outcome-focused wording outperforms generic.
First person sometimes wins. "Start my trial" against "Start your trial" has produced measurable differences in tests, though the effect is inconsistent and worth testing rather than assuming.
Reducing perceived cost. Every CTA carries an implicit price — time, money, effort, commitment, spam. Naming and reducing it works: "No card required", "Takes 2 minutes", "Unsubscribe anytime". This is frequently more effective than making the benefit sound bigger.
Visual prominence. Contrast against surroundings, adequate size, and whitespace around it. Note that the frequently repeated claim about a specific colour always winning is not supported — what matters is contrast with the page, not the hue itself.
One primary action per page. Multiple competing CTAs of equal weight split attention and reduce action on all of them. Secondary options should look secondary.
Placement matching readiness. A CTA before the reader understands the offer converts poorly; one only at the bottom is missed by people already convinced. Repeat it at natural decision points.
Removing friction after the click. A brilliant CTA leading to a long form loses everything it gained.
What tends not to work:
Manufactured urgency that is transparently false — countdowns that reset, fake stock counters. These now damage trust, and regulators have acted on them.
Aggressive or guilt-based wording, which measurably harms brand perception even when it lifts clicks short term.
Vague benefit claims without specifics.
Test on your own audience — CTA findings transfer poorly between contexts.