Is Google’s No Captcha reCaptcha A Conversion Killer?

Tweet about this on TwitterShare on LinkedInShare on FacebookShare on Google+Buffer this pageEmail this to someone



Spam is a pain in the arse. That’s pretty much something we all agree on. Some of it’s destructive; some malicious; most of it pointless; clogging up email; bloating CRM systems and messing up tracking and reporting.

reCAPTCHA, the most common counter measure, has long been a frustrating user experience, resulting in many an abandoned attempt to submit a web form. So when I heard about Google’s re-imagining of the reCAPTCHA I had to take a look.

So what is Google’s No CAPTCHA reCAPTCHA?

In announcing the release on their online security blog, Google claim that their new API, which ‘radically simplifies the reCAPTCHA experience‘, will enable users ‘to securely and easily verify they’re human without actually having to solve a CAPTCHA. Instead, with just a single click, they’ll confirm they’re not a robot‘.

Sounds great. Check out their video, it’s worth it for context:

Sounds a little geeky, but I was genuinely excited to give this a try.

So does Google’s No CAPTCHA reCAPTCHA reduce spam?

To test it, we implemented the reCAPTCHA on a lead generation form on our online accounting website.

Google reCAPTCHA on lead generation form

And the impact? Did it reduce spam?

Yes. It did, but it also had a rather painful unintended consequence – it killed genuine conversions.

In the interest of fairness, it’s worth saying this isn’t the most robust of evaluations – ideally tests should run longer, with more data. It ran for a full 24 hour period before we removed it. Given the numbers, I wasn’t prepared to keep it running.

So what happened?

If we compare the 24 hours with the reCAPTCHA versus the 24 hours prior without the reCAPTCHA, I can confirm it had the desired effect on the spam. Gone. Happy days.

Regarding conversions, lets take a look at some observations from studying the page and form data:

  • The pageviews for the two days were very similar
  • The percentages of unique users clicking the form submit button (‘Get started’) were practically identical over the two days

This suggests that the presence of the reCAPTCHA wasn’t off-putting to visitors. Those who wanted to sign up, would still try and do so regardless.

So far, so good.

The problem comes when you dig further into the conversion data. The number of successful form submissions FELL by 73% from one day to the next.

If we look at the number of unique users clicking on the form submit button and compare against the total number of clicks on the button we find our problem. Once the Google reCAPTCHA was added to the page the number of times the button was clicked DOUBLED – from an average of 1.7 times to 3.4 times.

Why? I can’t be sure, but I have a theory.

3 reasons why No CAPTCHA reCAPTCHA kills conversions

1. The appearance isn’t customisable

As far as I can tell, the appearance of the reCAPTCHA is not customisable. It can only look like this:


As such, I don’t think it looks like part of the form. It looks more like a banner.

2. The checkbox is unconventional

I’m not convinced the ‘check box’ looks like a typical check box. It just looks like a square and as such a user may not be aware that they have to click it.

3. The copy is ambiguous

I have a concern about the ‘I’m not a robot‘ text. Does the average person know what a ‘robot’ actually is in this context? Considering you’re reading this, I’m assuming you’re a web marketer or pretty tech-savvy. You know what a robot is, but are you an ‘average’ user?

The text also doesn’t have any instruction. Nothing tells you to check the box, so you have to work it out. In the words of Steve Krug, don’t make me think.

As such, my suspicion is that the ‘banner’ is ignored. The user believes they have completed the form and clicks the submit button. When nothing happens, the user continues to click the submit button thinking something is wrong with the website. This causes the number of form button clicks to double.

After a few clicks with nothing happening, they abandon the form and leave the site. Reduced spam, reduced customers.

Good effort, must try harder

I think Google’s intentions here are good. Everyone hates those old reCAPTCHAs. I’d happily never have to squint and cock my head to try and figure out another blurry photo of a house number just to access an online account. The new one does provide a better experience. I love the idea that they’re looking at the user’s entire engagement with the process – especially the mapping of a human’s jerky mouse movements that a robot struggles to recreate – but I think the execution needs more thought.

Personally, I’d rather any solution completely takes the onus off the user and is dealt with in the background by the technology. Less friction is not as good as no friction.

So, to quote their headline, ‘Tough on bots, easy on humans‘… yeah not so much Google. There’s a little more work to be done yet.

Tweet about this on TwitterShare on LinkedInShare on FacebookShare on Google+Buffer this pageEmail this to someone

Gary Robinson is a UK digital marketer, who fell into this tech world by accident and decided to stay and play. That was 1999. He's still here. His current loves are conversion optimisation, mobile and tinkering with new technology. He also has a fondness for coffee.

Tagged with: , , , , , , ,
Posted in Business Performance, Conversion Testing, Customer Experience
  • Ouch! Thanks for testing this — I hope Google makes some changes.

    • cheers for the comment Dan. I think its a good thing they’re trying to do and no doubt, they have plenty of data to optimise the product. So hopefully the next iteration improves.

    • They never will, as long as you ask them.

  • Matt Lacey

    Very interesting.

    I wonder if you gave it a title or intro line of copy in keeping with the style of the main form it would sit more happily in the flow of the form, which might help with issue 1. Granted, this perhaps shouldn’t be necessary.

    Points 2 & 3 seems really sensible.

    I would add a note of caution, that we have seen a number of tests flip within the first week even when early results have been damning. There is often a scientist vs businessman tension when testing which is understandable. Did you reach a reasonable number of outcomes/conversions before pulling those early readings?

    Are you considering other any follow-up tests or parking it for now?

    Finally, I wonder if this was re-tested once/if people start to recognise No Captcha from other sites it might perform better. There might be an early adopter disadvantage here.

    • Cheers for sharing your thoughts Matt.

      Agree – issue 1 could perhaps be minimised by adding in copy, but as you say its not ideal. I think more customisable elements would be better.

      And I 100% agree with you on the test results flipping comment. I really wrestled with this – which is why I thought it important to make the comment about this not being the most robust of evaluations. With my #CRO head on I wanted to run the test for a adequate period of time to ensure enough outcomes; be considerate of business cycles, etc.

      With my business head on, I’m looking at losing 74% of leads through a primary lead generation channel. On a B2B site, where you’re not dealing with high traffic volume, a thorough test would take weeks, which is not something a business can afford to do. As we had an alternative plan for dealing with spam, I was comfortable removing the reCAPTCHA.

      So, I’m not saying reCAPTCHA is definitively a killer of conversion, but rather the article was to highlight the need to not blindly place it on a site, just because Google says they’ve solved the spam problem.

      Re: your final two points. We have implemented a few things behind the scenes that so far seem to be helping. As I mentioned in the post, I prefer that rather than asking the user to do something extra – that’s like a tax on someone else’s bad behaviour.

      And you’re right, it might just be early days and people don’t recognise the new reCAPTCHA and don’t know what to do with it. A few big sites have apparently started using it – like WordPress – so perhaps that adoption will come sooner rather than later.

      cheers Matt.

  • John Hughes

    Did you test is v the original reCaptcha or v no reCaptcha at all. I am not at all surprised that adding another think for a user to do lowers conversion. That seems like common sense. The real test is whether it performs better than the original reCaptcha in terms of the conversion rate.

    If spam is sufficient that you need some kind of Captcha, what is one that gets the balance right between killing spam and not killing conversion? That’s the real test.

    • Thanks for your thoughts John and good points. No, the test was not versus the old reCAPTCHA, but I still think it is valid. I wanted to know if adding the new reCAPTCHA would reduce spam whist maintaining conversions. It could have been any anti-spam device – I would still want to know the impact of adding it.

      As I mentioned in the post, the appearance of it on the page did not affect the percentage of people trying to submit the form – they just couldn’t figure out how to use it.

      And sometimes, even common sense needs testing – not everyone applies it and if they do, it’s based on assumptions, conventions and best practice. But that doesn’t mean that it works on my specific audience.

      • John Hughes

        100% agree, common sense is no arbiter of results 🙂

        My point really is noCaptcha is an evolution of reCaptcha, and (in my opinion) the real question is whether noCaptcha is an improvement on that, from a conversion point of view – you wouldn’t add a captcha element to your form if you didn’t have a spam problem to address. Which offers the best protection for the lowest opportunity cost?

        • That’s a fair comment John and it would be a good test to run. Whilst we’re not running it as an A/B test, our new changes in the background are doing just that – they don’t interfere with the sign up of genuine leads, but they’ve reduced the spam volume.

  • Wow, excellent analysis – this un-customizable design definitely DOES make you stop to think, and we are all looking at this as 12+ hour-a-day digital fiends… so for ‘normal’ users this clearly is a stumbling block and the data don’t lie…

    • Thanks for your thoughts, KPM. Good point, if you have questions about what to do with it and you spend all your waking hours online, what will your average customer think about it?

  • It’s kind of weird that Google, the notorious Google of 50 shades of blue, themselves didn’t roll this in a A/B fashion to test the evidently poor default negative phrase in the captcha.

    I’m also using this on a site but the audience is ok with clicking on that phrase, others out there I’d hedge a bet would be rightly freaked out by the obtuse declaration about robots.

    • Yeah it is a little weird – maybe it has been tested and we’re not thinking the best of people, when actually they’re perfectly capable of understanding this talk of robots.

      I don’t think that is the case though.

      Out of curiosity, what is the audience type you’re showing the reCAPTCHA to on that project, if you don’t mind sharing? Are they more technically minded?

      Half the Crunch client base are in the IT industry so it’s not like this a bunch of newbies or anything

      • Target audience are high end streaming specialists. Pretty sure fait with robot talk so I don’t expect it will spook them. Infact using this is likely a kudos move in some circles.

        • That’s a good point – as I said in the post, after watching the video I was quite intrigued about using it and I’d consider myself a reasonably technical marketer. So the reaction is probably different by audience.

  • Love this Gary – we’re just investigating implementing this version so was great to find your post! We’re now investigating the alternatives again…

    • Thanks Gareth, glad it was useful. Hope you found an approach that was right for you – a balance between spam reduction and conversion

  • Nigel Stratton

    The data-theme attribute allows you to select between a dark and light theme but that doesn’t help an aqua form! On the other hand it’s probably something you could style with jQuery.

    • sagefire

      The reCAPTCHA widget is inside of an Iframe so technically you can’t do anything to it. Browsers don’t allow access to elements inside of frames that come from other domains.

  • Pingback: 7 Ways Form Accessibility Can Boost Conversions()

  • Pingback: How to create lead generation forms that convert()

  • sagefire

    “After a few clicks with nothing happening, they abandon the form and leave the site. Reduced spam, reduced customers.”

    I agree that the reCAPTCHA is poorly designed. But also the huge problem here is that your web developer did not implement an acceptable contingency for trying to submit the form without the field being checked.

    On our forms I use the JavaScript callback option so I know exactly when/if the user entered the value. Rather than use Google’s own implementation, I have the callback fill out a hidden field that follows our own paradigm/naming convention.

    If the user tries to submit the form without checking the “I am not a robot” button, they will get a notification to please do so and the area around the reCAPTCHA box will become highlighted in pink/red.

    This tells the user exactly why the form did not submit and allows them to rectify it. The worst thing in the world you can do is simply have the form do nothing or transport the user to an error page and force them to re-enter data. Those two outcomes will certainly decimate your conversion rate out of pure user dissatisfaction.

    I assume if you implement a contingency for users that didn’t know to check the box, your conversion loss will be dramatically less. Missing a required field is annoying but still only a peccadillo compared to simply not getting anywhere or having to re-fill the form—Those are big problems.

  • Pingback: Form UX – A little bit of form friction won’t hurt()

  • Pingback: Yes, I want to take care of your spam problem for you()

  • Pingback: Yes, I want to take care of your spam problem for you - Scott.Services Online Marketing NewsScott.Services Online Marketing News()

  • Pingback: Yes, I want to take care of your spam problem for you - Internet Marketing()

  • Pingback: Yes, I want to take care of your spam problem for you | Webbero()

  • Pingback: Yes, I want to take care of your spam problem for you - On Page SEO Checker()

  • Pingback: Yes, I want to take care of your spam problem for you - Synergy Capital()

  • Pingback: Yes, I want to take care of your spam problem for you – Internet Marketing How To & News()

  • Captcha or reCaptcha is a NERD language for “Dear Customer, We Are LAZY,
    so could you PLEASE do OUR work for Us?” And that means sorting out
    spam and bots for the site we are about to give a traffic credit or a
    financial gain, a purchase. No, no thank you! The road, vehicle or store identification gets so bad sometimes that I have to spend up to 2, 3 or even 5 minutes on that annoyance. You try to capture me? You ask me stupid questions such as am I a robot? I block you!

  • JoeyDoe

    Recaptcha is a sneaky way to steal people’s time in order to collect pattern recognition data for AI development. Nothing more nothing less.

  • Pingback: Marketing Land()

  • Pingback: Yes, I want to take care of your spam problem for you | Marketing Land()


Of all the places you could've ended up, you've arrived here. Might as well stay a while. Grab a coffee and have a read. And if you like it, give it a little share on your favourite social channel. My self esteem will appreciate it.

If you any questions or want just want to say hi, pick a social network or drop me a message here.