Call-to-action optimisation that converts. Contrast, copy and the mistakes that quietly cost clicks.

The short answer. What call-to-action optimisation actually changes.

Call-to-action optimisation is the process of testing a button or link's copy, colour, placement and surrounding context to increase the rate at which visitors click it, then measuring whether that click actually turns into the outcome you wanted, not just a higher click count. In one of the most cited tests in conversion research, HubSpot changed nothing but the colour of a single button on its Performable landing page, red against a page dominated by green, and saw a 21% lift in clicks. The lesson researchers drew from it was not "use red buttons": it was that contrast against the surrounding page, not any one colour, is what reliably moves a click-through rate.

A CTA sits at the end of a chain of decisions a visitor has already half made, and optimising it in isolation from the page around it rarely holds up. Our guide to what conversion rate optimisation actually is covers the wider discipline this sits inside; this guide is about the one element most pages get wrong first.

Step by step. How to optimise a call to action properly.

1. Write the copy around the visitor's next step, not your product. "Start your free trial" tells a visitor what happens next. "Get started" does not, and forces them to guess. HubSpot's own analysis of its CTA data found personalised, specific calls to action performed 202% better than generic, one-size-fits-all versions across its customer base.
2. Check contrast before you touch colour choice. Measure the button's contrast ratio against the page background it sits on, not against your brand palette. A button that matches your brand's plum tone perfectly but blends into a similarly toned page will underperform a plainer button that stands out.
3. Give it room. A CTA crowded by competing links, navigation, or a second button of equal visual weight forces a decision the visitor did not come to make. One primary action per section, with any secondary link visually smaller and lower on the page.
4. Match the CTA to where the visitor actually is. A first-time visitor on a blog post is not ready for "Book a call"; a returning visitor on a pricing page usually is. Sending both the same CTA wastes the specificity that drives the biggest gains.
5. Test one variable at a time, for long enough to trust it. Our guide to A/B testing mistakes covers the sample-size and duration errors that make a CTA test's result look real when it is just noise.
6. Track the outcome, not just the click. A CTA that gets more clicks but a lower completion rate on the step after it has not actually improved anything, it has just moved the drop-off one step further down the funnel.

A worked example. A CTA that looked fine and was not.

I worked with a B2B services client whose homepage CTA read "Learn more" in a button coloured to match their brand's soft plum, the same tone as most of the page around it. Click-through was low, and the founder's first instinct was to rewrite the copy. The actual problem showed up once we measured contrast rather than guessed at it: the button's contrast ratio against its background sat close to 2:1, well under the readability threshold most accessibility guidance recommends, so a meaningful share of visitors were likely not registering it as clickable at all rather than reading it and choosing not to click.

We changed two things only: the button's background to a shade with a contrast ratio above 4.5:1 against the page, and the copy from "Learn more" to "See how it works", specific enough to set an expectation. Click-through on that section rose within the first two weeks of the test, and the completion rate on the page it led to held steady rather than dropping, confirming the extra clicks were genuine interest rather than confused visitors clicking blind. The brand-consistency instinct, keep everything in the same plum family, was the thing actually costing clicks. Contrast has to win that argument, not lose it to a style guide.

Copy patterns that work, and the ones that quietly don't. Specificity beats cleverness.

First-person phrasing, "Start my trial" rather than "Start your trial", is worth testing rather than assuming: it performs better on some audiences and makes no difference on others, and the only way to know which applies to your visitors is to run the test rather than copy a blanket rule from a blog post. The pattern that holds consistently, across every test we have run, is specificity beating vagueness, not any one phrasing trick.

Placement and repetition. Where on the page a CTA earns its click.

A single CTA repeated at the top and bottom of a long page tends to outperform one placed only at the very end, simply because a meaningful share of visitors decide before they finish reading and leave if nothing is there to click. That does not mean scattering the same button every few paragraphs: a CTA that appears too often starts reading as pressure rather than a natural next step, and visitors learn to scroll past it.

On a page with more than one logical action, a comparison, a case study, a direct enquiry, resist the instinct to give each equal visual weight. Pick the one action that matters most for that page's stage in the funnel, make it the primary button, and let anything else sit as a smaller text link. A page offering a visitor three equally prominent choices converts worse on all three than a page offering one clear one, because the extra choice itself becomes a reason to leave and decide later.

Mobile CTAs need their own pass. Thumb reach and tap targets, not just wording.

A CTA that reads well and sits in a strong position on desktop can still underperform on mobile if it is not sized and placed for a thumb rather than a cursor. A tap target under roughly 44 by 44 pixels forces visitors to aim carefully, which adds friction that shows up as a lower tap rate even when the copy and contrast are both right. On longer mobile pages, a CTA pinned to the bottom of the screen as visitors scroll tends to outperform one that only appears once at the very end of the content, for the same reason repetition helps on desktop: a meaningful share of visitors decide to act before they finish reading.

Button copy that works on desktop sometimes needs shortening for mobile, not because mobile visitors read less carefully, but because a long label wrapped across two lines inside a narrow button looks broken rather than considered. Test the mobile version separately rather than assuming a desktop win carries over; the two audiences behave differently enough that they often do not.

How to run the test properly. Avoiding a false result.

A CTA test needs a large enough sample and a long enough run to separate a real effect from normal day-to-day noise, and the two most common failure modes are the same ones covered in our guide to A/B testing mistakes: calling a winner after a few days because the numbers looked good, and changing more than one variable at once so you cannot tell which change actually did the work. Change the button's colour, or its copy, or its placement, and test that one change through to a full, pre-agreed sample size before drawing a conclusion.

It also helps to decide in advance what "better" means beyond the click. A CTA that lifts click-through but leads to a form with a lower completion rate has not improved the page, it has just moved friction one step further along. Track the full path from click to the outcome that actually matters, a booked call, a completed signup, a form submission that reaches your CRM, before calling any CTA change a genuine win.

Common mistakes. Where CTA optimisation goes wrong.

• Optimising colour before checking contrast. A colour change that still fails contrast against its background will not fix the underlying visibility problem, whatever hue you pick.
• Testing a CTA in isolation from the page's intent. A visitor who has not yet been told why they should care will not click a better-worded button; the copy above the CTA usually needs fixing first.
• Using the same CTA everywhere. A single generic "Contact us" across every page ignores how differently ready a first-time reader and a returning prospect actually are.
• Declaring a winner too early. A CTA test called after a few days, or a few dozen clicks, on a low-traffic page is usually reading noise, not a real signal.
• Chasing clicks instead of outcomes. More clicks on a CTA that leads to a form nobody finishes is not a win. Measure the full path, not the first step of it.

Start with contrast and copy specificity, since both are free to fix and tend to move the number furthest. Then move to placement and audience-matching once those two are settled, and only then to a formal split test of anything more subtle.