The CTA Tax

A CTA button is the smallest unit of conversion design. It’s also the unit most teams treat as cosmetic — change the color, shrink the radius, ship it. The copy on the button rarely gets the same scrutiny as the H1, even though the button is where the entire page either pays off or doesn’t.

We wanted to know which CTAs work. Not in the abstract, but in the wild — on the live pages of real companies trying to convert real visitors. So we wrote a scraper that extracted every visible CTA button from 1,200 active landing pages (8,400 buttons total), tagged each one for verb, word count, position, and prominence, and matched the button data to the page’s CTA dimension score in our analysis pipeline.

The findings break into three groups: a count problem, a verb problem, and a placement problem. Each one is independently fixable. Each one costs pages between 4 and 10 points on the CTA dimension. Stack them and you’re looking at a 12–18 point swing on overall page score — large enough to move a page from the bottom half into the top quartile without changing anything else.

The median page in our sample shows 4.2 visible CTAs above the fold. The top quartile averages 2.1. That gap is the single biggest predictor of CTA-dimension score in the dataset.

Pages with one primary CTA score an average of 6.8/10 on the CTA dimension. Pages with three or more competing CTAs average 4.7/10. The 2.1-point gap maps to a 31% difference on the underlying dimension scale, which translates to an 8–10 point overall page score swing depending on how the rest of the page is built.

Every additional button above the fold is a tax on attention. The visitor pays it before they ever click.

The mechanism is the one Jakob Nielsen has been writing about since 1995: choice friction. When a visitor sees “Get started,” “Book a demo,” and “Learn more” visible simultaneously, they hesitate. Hesitation costs you the click. Pages that pick one primary action and visually subordinate everything else (lighter weight button, smaller size, text link) keep the visitor on a single track.

What about hybrid CTAs (“Try free or Book demo”)?

Two paths can work if the visual hierarchy is clear. The best-performing pages with two CTAs use one primary button (filled, brand color, prominent) and one secondary (outlined, neutral, or even text link). When both CTAs are the same visual weight, the page reads as “four buttons” not “two paths.”

Here’s the distribution of primary CTA verbs in our sample, ranked by frequency. The right-hand column shows the average CTA dimension score for pages whose primary CTA uses that verb.

Two patterns. First, the CTAs that include both a verb and a specific noun (“Start free trial,” “Book a demo”) consistently outscore the bare verbs. Specificity wins on the button just like it wins in the headline.

Second, “Submit” is still the form button on 9% of landing pages we analyzed in 2026. The phrase tells the visitor they’re doing something for the company — submitting their information, submitting an application. Compare to “Send my message,” “Get my pricing,” or “Show me the demo” — phrases that describe what the visitor gets back. Same click. Different framing. Pages that use the second pattern score 5 points higher on the CTA dimension.

What about pages that A/B-tested their way to “Get started”?

We’re aware some sophisticated teams have tested verb changes and concluded “Get started” is fine. Our data doesn’t contradict that — it just clarifies what they were testing against. “Get started” beats “Begin” or “Continue.” It doesn’t beat “Start your 14-day trial” in the pages where we’ve seen the comparison. If “Get started” is your CTA, your A/B test shouldn’t be against another bare verb. It should be against a verb that names the noun.

Mobile traffic exceeded 60% of all web traffic in 2025 (Statista), yet a striking share of pages are still designed desktop-first and shrunk afterward. Our viewport analysis caught 41% of the sample pushing their primary CTA below the mobile fold. The cause is almost always the same vertical stack: hero image, headline, subhead, then button. On a 1280px desktop viewport that’s comfortable. On a 390px iPhone viewport the button lands around y=900 — well below the initial scroll.

Pages with CTAs above the mobile fold score 6.4/10 on average on the CTA dimension. Pages with CTAs buried below average 4.9/10. The placement effect is independent of the copy effect — even pages with the strongest CTA copy lose 1.5 points if visitors have to scroll to find it.

The fix doesn’t require a redesign. Most pages can swap the order on mobile only — CTA before subhead, or CTA before the hero image — without touching the desktop layout. Three lines of CSS in most modern frameworks.

Pull up your live landing page on a phone. Don’t scroll. Run through this list. Fix what you find.

1. Count the CTAs above the mobile fold. Include buttons, text links styled as CTAs, and any element a visitor could plausibly click as “the action.” If the count is more than two, you have too many. Visually subordinate everything except your primary.
2. Read your primary CTA aloud. Does it name the action? “Start your trial” passes. “Get started” doesn’t. “Send my message” passes. “Submit” doesn’t.
3. Add a friction-lowering qualifier. “Start your trial” becomes “Start your 14-day trial — no credit card.” The qualifier addresses the objection the visitor was about to have. Pages that add a sub-button line (“No credit card,” “Free for teams under 10,” “Cancel anytime”) score 3 points higher on the CTA dimension on average.
4. Check your form button. If it says “Submit,” change it to something that names what the visitor gets back. Three minutes of work.
5. Check the mobile fold. Open your page on an iPhone 13-sized viewport. Is the primary CTA visible without scrolling? If not, reorder the hero stack on mobile.

Once you’ve made the changes, score the page with the CTA analyzer or run a full page analysis. Watch the CTA dimension score move. The lift is usually visible within a single iteration.