Headlines vs buttons

“When you have written your headline, you have spent eighty cents out of your dollar.”

Are headlines dead?

In less dramatic terms: If you were allowed to optimize just one element of a page, would you optimize your headline… or something else?

Thanks to the great teachings of the father of advertising David Ogilvy, we’ve all come to believe that headlines are where it’s at. They’ve got two turntables and a microphone. They’re like Gstaad – they’re the best.

Now, I’m a copywriter. So I happen to love me some headlines.


I’m also in conversion-rate optimization. So I do a lot of A/B testing. And I’ve gotta say: I’ve seen some shit lately that is shaking the way I think about everything we write. (Watch for more about that in coming weeks.) And I want to show you a few test results that might challenge the power of the headline… and compel us to stop treating the button like some sort of shrunken, impotent and second-rate little square on a page.

Check This Out:
A Very Surprising Headline Test on Dressipi.com

This is the control version of the home page of Dressipi.com, a cool UK-based service that lets you find clothes you’ll love no matter your size.

Buttons vs headlines

Pretty standard page, right? Looks good. Is easy to read. …But it doesn’t exactly grab you, does it?

The headline, we hypothesized, left something to be desired. In forums and otherwise online, Dressipi users and prospects talked about their bodies (and the clothes they struggled to find) in a more casual, colloquial way than this headline does. They were jokey about it, self-deprecating. They used words like “big bum”. So we decided to test a riskier headline – one written the way Dressipi prospects talk. This is it:

Dressipi treatment weak button strong headline

It’s riskier, right? It’s doing more than the control was doing…

When we tested it, we saw a lift of about 15% that held throughout the test… but that never reached confidence. Visitors were essentially telling us, “It’s better, sure, but… meh.” So, dejected and dismayed, feeling like pathetic little hacks, we paused the test and did the usual post-mortem.

Was the headline too risky?

Were the question marks too much?

Was it too long?

Should we not have used an ellipsis before the third question mark?–maybe all that punctuation was messing with people’s eyes the way these italics are messing with your eyes now!!!

Admittedly, there are a hundred thousand ways you can write a headline to communicate the same message. There’s the uber-awesome Caples-esque Upworthy way. There are simpler, more timid “2.0” ways. There are flashy direct response-style ways. And then there’s everything in between. Yes, this headline could have been wrong. (That’s why we test!) But it didn’t hurt conversion; it just didn’t nudge enough people over the Game of Thrones-sized Wall keeping them from converting.

So here’s what we did…

We Decided Not to Put This Headline to Bed  –
But Rather to Test the Headline with New Button Copy

I am not one to knock the masters of copywriting – the Ogilvies, Bernbachs, Schwartzes, Capleses. (Don’t hate me for how I pluralized those names.) But I am one to challenge shizzle… and to follow a strong hunch. And my hunch told me this: That headline rocks, but it doesn’t matter because that f***ing button sucks!

So we tested this new treatment against the control:

Buttons vs headlines

Same attention-grabbing headline. New button copy.

We went from “Sign up now” to “Show me outfits I’ll love”.

The results: We got a 123.9% lift in clicks on that very button, with 100% confidence (as reported by Optimizely and confirmed with 2 other calculators).

Using the exact same headline but a different button, we went from a wishy-washy, non-significant <15% lift… to a strong-like-bull 124% lift.

That’s One Strike Against the Almighty Headline…
And Here’s Strike Two

VueScan is a solution by Hamrick that essentially keeps your outdated scanner software going strong. Check out the headline Hamrick was running on its home page:


Not exactly the most inspired line of copy ever written, is it?

Like any copywriter would, I looked at that and went, “Come on! We can slaughter that thing without even trying.” But we did try. We developed these new headline-subhead combos to test against the above Control:

Headline testing

As before, say what you will about the headlines, but this much is true: we tested a variety of headlines, some negative, some not so negative. We tested after learning about VueScan’s prospects. And we tested against a headline that anyone would pinpoint as in need of optimization.

The result: nadda.

Zip. Squat. Nothing to see here, folks. Even after 36 days of running and tons of traffic through all variations, the result was flat and non-significant. Like in the Dressipi case, even if they had reached significance, the trending lift we saw was never better than a very, very modest 4 or 5%.

So Then We Tested the Button

Again, like in the Dressipi case, we ran a new test that paired an attention-grabbing headline that had done relatively well with an optimized button. We also paired the ho-hum but as-yet-unbeatable headline with the new button copy. Here’s how it all shook out:

button test

By changing the button, we saw a statistically significant lift of clicks on the Purchase button. (Clicks on the red button fell in keeping with the lift on the blue. Why blue beat red here is the subject of a future post.)

So here’s what I’m starting to take away… Changing the headline may not ‘guarantee’ a valid, measurable and meaningful lift the way changing a button seems to. So if the goal is to move people through our funnels, which might we start to think is more important: the headline, or the button?

Two Strikes, Headline!
Can We Get One More?

Before I wrap this post up with this cool test, let me make something clear: I don’t believe headlines are dead. That’s not a cop out. 🙂 Headlines aren’t dead; they’re just not operating in isolation the way, perhaps, they might once have been. We can boost our conversion rates when we optimize our headlines and buttons together.

There’s always a better headline; there’s always a better button. If you optimize one without optimizing the other, you may not see the kind of lift you would if you created new treatments that optimized both together. Of course, that new treatment wouldn’t make for the world’s cleanest test… but it could make for a much better lift, as you’re going to see in this final study.

We tested 3 treatments against the Control on a landing page for Health Insurance Comparisons in Australia. Only the headline and/or button were changed. Check out the creative… and the results:

Headlines vs buttons

This is interesting.

Take a look at V3 there at the bottom. It’s the Control headline with the optimized button. On its own, the optimized button didn’t perform as well as it clearly could have. It still beat the Control, yeah, but it didn’t outperform the two variations where the headlines were also optimized.

So… is the seemingly humble button more powerful than the headline?

No. (At least, we don’t have data to let us say “yes.”)

But it is powerful. It’s directly connected to conversions, and that makes The Button worthy of at least as much attention as The Headline gets. If Ogilvy were around to see the web and mobile worlds, he might even agree. Now, the question is, do you agree?


PS: Thanks to Jen Havice for her excellent assistance running the first 2 tests.