A call to action is the text that asks the reader to do something and tells them what they get by doing it. It sits on a button or a link: "Add to basket", "Request a quote", "Download the price list".
There are plenty of lists of examples in English. I looked at the eight text pages Google UK shows on the first page for "call to action examples": their titles alone promise at least 374 formulas. Above them, an AI Overview already lists fourteen of the most common ones, from "Buy Now" to "Schedule a Demo", with the advice to keep buttons to three to five words.
The examples are here too, but each one comes with the context in which it was measured and the market where that number was recorded. That is rarely your own market, and it is worth knowing before you copy someone else's button.
What people in the UK search for when they type "call to action"
Before the examples, one number that changes how the page should be written. In the UK, search demand isn't about examples, it's about meaning.
| Search | Monthly volume in the UK | Intent |
|---|---|---|
| cta | 8,100 | ambiguous, the acronym is used well outside marketing |
| cta meaning | 2,400 | informational, partly ambiguous |
| call to action | 1,300 | informational |
| call to action meaning | 480 | informational |
| what is a call to action | 390 | informational |
| call to action examples | 260 | informational |
| cta marketing | 90 | informational |
| cta website | 50 | navigational |
| call to action instagram | 10 | navigational |
People looking for the meaning ("call to action meaning" plus "what is a call to action") outnumber those looking for examples by more than three to one, and that is before counting "cta meaning". This is why the definition goes in the first line and not after three paragraphs of introduction, and why a page that opens with "in this guide we will look at" loses the reader before telling them anything. The same principle applies to the meta description: if the search asks for a definition, the snippet must contain it.
The 374 formulas on the UK first page
I ran the search on Google UK on 26 September 2026 and recorded, for each organic text result, how many examples the title promises and the date Google shows next to the result.
| Page on the first page | Examples promised in the title | Date shown in the result |
|---|---|---|
| HubSpot, 49 call-to-action examples | 49 | 8 May 2025 |
| AdEspresso, 17 call to action examples | 17 | 21 October 2022 |
| CTA Examples, library of 100+ examples | 100+ | not shown |
| Wix, 30 call-to-action examples | 30 | 25 May 2026 |
| Figma, 19+ call-to-action examples | 19+ | not shown |
| Grammarly, 50 effective CTAs | 50 | 11 December 2023 |
| Indeed, 51 call-to-action examples | 51 | 15 June 2026 |
| KlientBoost, 58 call-to-action examples | 58 | not shown |
| Total | 374+ |
The AI Overview at the top says a lot about where this kind of content is going. It answers the query with a generic list and a few tips, citing iubenda, Figma and Wix, and ends by offering to write personalised examples for your business. A list of formulas is exactly what an AI summary can produce on its own. What it can't produce is the number that tells you which one to use, and where.
How well a call to action converts, according to those who counted
First Page Sage aggregates data from 71 clients and publishes conversion rates by CTA type and position. The report was last updated on 17 April 2025 and the sample is from the US, mostly B2B and software.
| CTA type and position | Conversion rate |
|---|---|
| Full-page popup, content download | 13.6% |
| Full-page popup, offer or discount | 12.1% |
| Full-page popup, account creation | 11.5% |
| Lightbox, email sign-up | 9.8% |
| Mid-page banner, content download | 3.8% |
| Slide-in, sign-up or free trial | 1.8% |
| Sidebar | 0.4% to 1.2% |
| Bottom-of-page banner | 0.3% to 0.9% |
| Inline link within the text | 0.2% to 0.8% |
There is a difference of almost seventy times between the first and the last value. Before rewriting the text of a button, look at where the button sits, because position weighs more than words.
The second solid number comes from HubSpot, which compared more than 330,000 CTAs over six months: those personalised to the visitor's behaviour converted 202% better than those that were the same for everyone. Here too the sample is largely American and made up of companies using a marketing automation platform, so the number shows a direction, not the figure you will get.
We haven't found a public UK benchmark of conversion rates by CTA type. That is the reason a test on your own site is worth more than an imported benchmark, provided you have enough traffic to run one.
"Learn more" is the most common formula and the weakest
Katie Sherwin of the Nielsen Norman Group wrote the reference piece on the problem in December 2015: a "Learn more" link doesn't say what is on the other side, and anyone browsing with a screen reader ends up with a page of eight links with identical text. It still features in the list of common CTAs that the AI Overview shows for this very search.
Two years later Aurora Harley and Kim Flaherty published tests on an even more widespread label, "Get started". On a food delivery site six users out of eight clicked it expecting to find out how the service worked and found themselves facing a registration form. Clicks go up, sign-ups don't: it is the case where one metric improves while the result gets worse, and it is why click-through rate on its own isn't enough to judge a button.
The rule that follows is practical: the button text must describe the page that opens. "See prices" leads to prices, "Book a demo" opens a calendar. If you can't write it, the problem isn't the button, it's the landing page. We have also written about the link between expectation and landing page in the piece on user experience optimisation.
The public rules exist, and they are free to read
For interface microcopy there is a UK reference that none of the example lists on the first page cites: the GOV.UK Design System, maintained by the Government Digital Service for public services. On buttons it says four things.
- Write button text in sentence case, describing the action it performs: "Save and continue" when information is saved, "Continue" when it isn't.
- Add words when the action needs them: "Add another address" or "Accept and claim a tax refund" rather than a bare verb.
- Avoid using more than one default button on the same page.
- Don't rely only on the red colour of a warning button to communicate how serious the action is.
These rules were written for public services, and for exactly that reason they are the safest starting point for a commercial site.
Stock formulas produce buttons that say nothing. A few replacements that hold up:
| Stock formula | What it communicates | Version that describes the action |
|---|---|---|
| Submit | what the browser does | Send your request |
| Learn more | nothing | See prices |
| Get started | nothing | Open an account, it takes 5 minutes |
| Subscribe | the action, not the content | Get the report every Monday |
| Contact us | no promise | Request a quote for your project |
| Sign up free | free until when? | Try it for 14 days, no card needed |
Examples by context, with the criterion for choosing them
Ecommerce
Baymard Institute brings together 50 independent studies on basket abandonment: the documented average is 70.22%, updated on 22 September 2025. Seven baskets in ten never reach the checkout, so the button that decides revenue isn't the one on the product page, it's the one at checkout.
On the product page, the formulas that work are the ones that say what happens after the click: "Add to basket" keeps you on the page, "Buy now" takes you to the checkout, "Tell me when it's back in stock" catches people who would otherwise leave. At checkout, put the information that makes people abandon into the button itself: "Go to payment, delivery included" says one thing more than "Continue". We covered the other levers on the product page in the guide to ecommerce SEO.
To calibrate expectations for the UK market: according to the ONS retail sales bulletin for August 2026, 28.8% of retail sales in Great Britain were made online, up from 28.4% in July. More than a quarter of retail spending goes through a checkout button.
Lead generation and B2B
In B2B a generic CTA costs more than elsewhere, because the form comes after weeks of evaluation and whoever fills it in already has a precise question in mind. "Request information" collects the curious, "Request a quote for your installation" collects people who actually have an installation.
An example from our own files. For Macropix, a Milan-based manufacturer of custom LED screens, the nine keywords that bring the most qualified leads don't contain the company name: they are descriptions of products or applications such as "ledwall outdoor" or "totem led" (LED totem). We have told the full story of the project in the piece on SEO for manufacturing. The same logic applies to the button: someone who arrives searching for "ledwall outdoor" needs a CTA that names that object, not a "contact us" that fits every page. On building the full journey we have a dedicated guide to B2B SEO strategy and a lead generation service.
Newsletters, content and landing pages
"Subscribe to our newsletter" describes the action of the sender, not the benefit to the reader. The versions that hold up state content and frequency: "Get the SERP analysis every Monday". When there is a file to download, stating format and size lowers friction: "Download the PDF, 12 pages". For pages built to convert we have collected five changes in the piece on landing pages, and the criterion for writing the text around the button is in the guide on how to create SEO content.
Ads and social
On Meta, and in Google Ads display and video formats, the button text is chosen from a preset list: "Shop now", "Learn more", "Sign up". The free space is the line above, and that is where the specific promise goes, because the button alone doesn't carry it. The same goes for retargeting campaigns: on how they work and the minimum volumes, see our guide to remarketing.
The mistakes that come up most often
- Several different CTAs on the same screen. Every extra option is a reason to choose none of them, and on mobile they all take up the same space.
- "Submit" as the form label. It describes what the browser does, not what the person who filled in the form gets.
- A promise the next page doesn't keep. It is the mistake documented by the tests on the "Get started" button: it brings clicks and doesn't bring conversions.
- The same CTA for people who are learning and people comparing prices. On an informational article "Request a quote" is badly timed, "See how we work" isn't.
- A button that appears only once, halfway through a three-thousand-word article, and never again.
How to tell whether a call to action works
The number to watch isn't how many people click, it's how many of those who click get to the end. A button that sends 12% of visitors to a form that converts 3% of them earns less than one that sends 5% to a form that converts 20%.
In Google Analytics 4 you set up one event on the click and one on completion and look at the ratio between the two. On the calculation and benchmark values there is a separate page on what conversion rate is, and on the metrics to keep an eye on over time, the one on SEO KPIs. A warning on method: with thirty conversions a month an A/B test can't tell an improvement from a fluctuation, whatever gap the tool shows.
Frequently asked questions
CTA stands for call to action: the text on a button or link that asks the reader to do something and says what they get. In the UK the acronym is searched more than the full phrase: "cta meaning" has around 2,400 searches a month and "call to action meaning" around 480, although part of the first figure is about other uses of the acronym.
One main action, repeated several times down the page. Repeating the same button doesn't confuse anyone: offering three different ones on the same screen does, because it forces the reader to choose before they have understood.
Yes, in the imperative. The GOV.UK Design System asks for button text that describes the action it performs, and all its examples start with a verb: "Start now", "Sign in", "Save and continue". "Request a quote" works better than "Quote request".
Long enough to state verb and object, no more. The GOV.UK Design System adds words only when the action needs them, as in "Add another address" or "Accept and claim a tax refund". When a condition is needed, such as the price or the length of a trial, it sits better in the line under the button than inside it.
Not directly as a ranking factor, but link text is a signal Google reads, and a button labelled "Learn more" doesn't describe the destination page. The same text is also the only part of the button that survives when an AI system summarises the page.
One thing shifts the reasoning, and almost nobody is considering it yet. When ChatGPT, Gemini or an AI Overview summarises a page, only the button's text label survives: colour, position and size disappear. A page full of "Learn more" gets summarised without its offer, while "Request a quote for industrial solar installations" arrives intact on the other side, inside an answer the user reads without ever opening the site. The accessibility rules the Nielsen Norman Group has been repeating since 2015 have, by accident, also become rules for visibility in answer engines, and this is a piece of work that falls under answer engine optimisation long before any technical change.
