website redesign

Marketers have always relied on testing.

But let’s be honest. It’s probably only in the last few years that they’ve begun discussing conversion rates rather than golf scores over a beer.

The Austin #CRO community is a dedicated bunch: @peeplaja @jtrondeau @mercertweets @bmassey

The Austin #CRO community is a dedicated bunch:
@peeplaja @jtrondeau @mercertweets @bmassey

The level of measurement and testing that we now have wasn’t even possible in the “old” days. Now that it is, CRO (conversion rate optimization) is clearly a “thing”.

And yes, I’ve got the data to back that up!

According to Econsultancy, in the last five years, the number of companies using A/B testing has more than doubled. Two-thirds (67%) of the companies surveyed use A/B testing, making it the top optimization method used today. Compare that to five years ago, when only a third of businesses were testing.

You might say it’s the golden age of conversion optimization.

Cool thought, I know. And it sounds like it should benefit businesses across the board. But that’s not what we’re seeing.

Whenever any tactic becomes a “thing,” it gets adopted by newbies and wanna-bes as well as the pros. So beware! You could be paying good money for “website optimization” services from an agency who just learned last month how to run a test.

The Truth About Testing and Website Optimization

The truth is, CRO is hard. You can’t learn it in a month, and you won’t be an expert until you’ve done it for years.

Let me say that again: It takes YEARS to become a pro.

What does that mean? It means lots of agencies are making mistakes without even knowing it because they’re so new to the game. Here are some of the mistakes we see most often.

1. “Best practices” landing pages

Best practice is NOT the same thing as conversion rate optimization. PPC agencies, SEO agencies, UX and UI people—they’re all claiming to do CRO.

But calling it CRO doesn’t make it so. Here’s what Brian Massey told me the other day:

“Here at Conversion Sciences, we’ve stopped doing best practices consulting because it’s so unreliable. Even if someone asks for it, we won’t sell it to them.

As brilliant as we are, when we implement best practices, we’ll be wrong on half of them.

Every audience is different. You have to test to know what’s working. Period. End of story.”

As an example, best practice says videos are good and sliders, or carousels, are bad. It says that sliders distract visitors, are hard to read, ya-da-ya-da-ya-da. Not so, according to a DeviceMagic case study, published by VWO.

This test pitted a slider against a video to see which would work better on the home page.

DeviceMagic was pretty sure the sliders were a better fit for their purposes, but they knew better than to make the change without testing

DeviceMagic was pretty sure the sliders were a better fit for their purposes, but they knew better than to make the change without testing. Interestingly, the video seemed to be an early winner. But after reaching statistical significance, the slider was the clear winner.

  • Conversions from homepage to signup page increased 35%.
  • Subsequent signups increased 31%.

Another example comes from one of my own projects. I had been commissioned to rewrite a collection of landing pages and, sadly, discovered that we were using best practice as our guide. The results? We nailed SEO, but conversions dropped to half of what they were before the rewrite.

Agencies fall into the same trap. They hear that something is working on another website, and they adopt it, no questions asked.

A landing page built on best practices rather than a solid testing strategy isn’t going to get you the results you’re looking for.

If it does, it was just dumb luck!

2. Testing the Wrong Things

When you rely on hearsay rather than data, it’s easy to make another mistake as well—testing the wrong things.

Experienced Conversion Scientists™ know which data gives them the insights they need. And they know which tools will give it to them.

Some agencies try to shave expenses by cutting out the data-collection tools—things like click testing, heatmaps and user-session recording tools. As a result, they don’t have the data to make smart decisions about what to test. These agencies pick something out of the blue to test instead of using analytics to figure it out.

In other words, they’re testing for the sake of testing.

Science should be based on hypotheses, not guesses or busy work. So you gotta ask, if your tests aren’t based on data, what’s the point?

Honestly, that’s the case for a lot of tests. Alex Turnbull, Groove’s founder, gives a great example of this. He lists some tests that are often considered “easy wins.” But for Groove, he says, they were pointless.

Pointless easy win email sign up

Pointless easy win email borrowed trust

Pointless easy win prices

Typically, these tests are the first tests newbies try to run, not because they’re relevant to the website or the audience, but because they seem like easy wins. Remember: trust the data, not someone else’s results.

3. Reliance on Self-Reported Data

Data is important. But you can’t depend on just any data.

Self-reported data—such as responses from focus groups, user testing, surveys, and forum feedback—is gathered from people’s stories, not their behavior.

The problem is people lie.

They may not mean to, but they do. If you ask them how they spend their money, they give a best-case scenario or what they wish to be true. Not the absolute truth.

Compare their answers to your analytics and you get another story. The real story.

That’s why Conversion Scientists don’t put much stock into self-reported data. Qualitative data (self-reported) is great for generating hypotheses, but it needs to be validated with testing.

Here’s where you need to be careful: A lot of agencies (especially UX and UI) redesign a site using only self-reported data. A true Conversion Scientist uses analytics and split testing to verify assumptions before deciding they’re true.

The Marks & Spencer 2014 redesign proves the point: Costing £150 million (about $230 million), this redesign took two years and led to an 8% decrease in online sales.

Based on the fact that this project took two years, I’m guessing that most decisions were made from self-reported data plus the design team’s own opinions.

It’s unlikely A/B testing was involved because testing delivers incremental changes rather than one massive change. And it allows you to mitigate technical errors, because you know for certain whether the changes you’re making are helping.

DigitalTonic says it well in their analysis of the redesign:

“Drastic changes can never be monitored meaningfully and you won’t be able to separate the variables that are causing the positive or negative impact on conversions. With testing on your current site prior to redesign, you will hit a local maxima meaning that you have optimised the site as best as you can in its current incarnation. It’s at this stage that you would take the learnings and move towards the global maxima with the redesign process.”

The issue here is really about behavioral versus attitudinal data. Look at this chart, which illustrates the landscape of 20 popular research methods, and you can see why this is such a common mistake. Self-reported data looks like a scientific approach.

20 Popular Research Methods

Behavioral versus attitudinal testing

Surveys and focus groups give useful information, but since the data is both attitudinal and qualitative, it should never be the foundation for testing.

Use it to help you develop smart hypotheses. Use it to understand your users better. But alone, it’s not valid. Behavioral (or quantitative) data is your most reliable source of information.

As Christian Rohrer states, “While many user-experience research methods have their roots in scientific practice, their aims are not purely scientific and still need to be adjusted to meet stakeholder needs.”

4. Not Understanding the Scientific Process

Agencies are time and materials companies. They bill by the hour. Understandably, they want every hour of their employees’ time to be accounted for and assigned to a winning project.

The problem is, this focus on the bottom line can actually dampen results.

Scientists need time to be curious, follow their hunches and understand the reason things are happening. A successful A/B testing agency needs to give them that time, even if some of those hunches turn out to be pointless.

In the long run, it’s cheaper to eliminate hypotheses early, before testing. If experienced Conversion Scientists are given time to “play,” they can usually do that with analysis alone, saving time and money.

In other words, a few hours of analysis beats 2 weeks of testing every time.

If you're not into making mistakes, then you're not doing anything

True inspiration requires time. Time to follow dead ends. Time to dive into the data. Time to think and ask questions. If your agency doesn’t allow that, be aware, you’re probably not getting the best results.

5. Offering a Completely Done-For-You Model

This one sounds more like a premium service than a mistake. But when it comes to CRO, it reads more like a mistake.

Some agencies believe they have more “job security” if they make the client completely dependent on them. So they do it all: collect the data, make the hypotheses and, supposedly, deliver results.


There’s no collaboration with clients. Which means they’re only using half the information they should be using to create hypotheses.

Here’s the thing: The best results come when the agency and client work together. The agency has the expertise to collect the data, but the client has the intimate knowledge of the customer. It takes both.

Seriously. If your agency is doing everything for you, they may be creating issues rather than solving them.

Does your agency see you as a money tree? Collaboration, rather than DFY services get the best results.

Does your agency see you as a money tree? Collaboration, rather than DFY services get the best results.

6. Not Bothering to Influence the Client Culture

Similarly, some agencies appear to collaborate with the client, but they draw the line before influencing client culture.

In reality, there’s a huge advantage to having an agency work so closely with you that they actually change the way you do things.

True collaboration involves getting together on a frequent basis and discussing ideas. Over time, you begin to see the thought process that goes into each website optimization effort. You begin to understand how to make decisions based on data and to value the insight numbers can give you.

When that happens, whether you consider yourself a numbers person or not, you’re hooked.

That’s the point at which you stop making random marketing decisions. Instead, you call your agency and ask what data needs collecting and when you can start testing. (Congratulations! You’re a conversion geek!)

As we talked about before, a done-for-you or non-collaborative service may not be giving you the best results—and they may be charging premium rates to do so.

Always remember, you’re the resource for testing. Not many agencies actually try to influence your company’s culture. Make sure yours does.

7. Not Staffing for CRO

This is a biggie. An agency that doesn’t staff for CRO shouldn’t offer CRO. You see, the best Conversion Scientists are skilled in two areas. They’re good with numbers and they’re excellent communicators.

Good with numbers. Getting high marks in high-school math isn’t enough. Conversion Scientists are masters of data and statistics.

They know when numbers are reliable and when they’re not. So they know how long to run a test and when the results are statistically valid. They know when the math is bad, which means you can be sure you’re getting positive results.

But being good with numbers isn’t everything. Great Conversion Scientists are also…

Excellent communicators. All too often, Web developers are recruited to do analytics, and sure they understand the numbers—but not much else.

It takes a conversion optimizer to turn data into stories. Frankly, that’s where the magic happens.

At Conversion Sciences, the team spends much of their time going through the numbers to tease out the stories. If there’s a hole in the plot, they design a test to figure out what’s missing. The goal is to find the story in the data—and tell that story well.

Conversion optimizers are fortune tellers

Conversion optimizers are fortune tellers

If you think about it, conversion optimizers are really fortune tellers. They predict the future based on the data your site gives them. Is your agency converting analytics to customer stories? If not, you may be dealing with Web developers rather than conversion optimizers.

8. Failing to Test Before Going Live

Pros test and validate everything before going live. That avoids costly mistakes like Finish Line’s 2012 Web redesign, which cost the chain around $3.5 million in sales and a huge hit to its reputation.

Four days before Black Friday, Finish Line launched a freshly redesigned website, supposedly planning to “reinvent the shopping experience.” Instead, customers complained about lost orders and other technology glitches, and Finish Line had to revert back to the old design just prior to the Christmas shopping season.

Granted, that level of mistake isn’t likely for smaller brands, but bad usability can still impact reputation and profitability.

My guess is a brand agency was responsible for that redesign. It would have been smarter to work with conversion optimizers, who understand how to use data to decide on incremental changes, validating each one before moving on to the next.

9. Making Rookie Mistakes

Since CRO is now a “thing,” everyone and their office cat now offer website optimization services. Most don’t know the difference between conversions and sales, which means they’re making a lot of mistakes.

Now don’t get me wrong. We all make our fair share of junior mistakes when we’re starting out—things like delivering results without statistical validity, not analyzing traffic enough, and the like.

But this is the “golden age” of conversion optimization. Don’t you want pro results?

Again, not everyone who claims to be a conversion optimizer is. Unless your team is experienced and has a structured approach for improving conversions, they’re likely making some mistakes that could be easily avoided—if they were more experienced.

Pro CROs use a structured approach to improving conversions.

Pro CROs use a structured approach to improving conversions.

Download a free copy of our eBook Four Rookie CRO Mistakes to Avoid at All Costs.

Website Optimization Mistakes Bottom Line

As you can see, mistakes are more common than not. That’s because website optimization is hard work.

If you want to get the big results CRO promises, you need an agency that has the experience and know-how to do it right. Period. (BTW, I recommend talking with Conversion Sciences about whether they can help.)

What CRO fails have you seen? And what are you doing to keep from making bonehead testing mistakes? Share in the comments below.

The knockout punch came near the end of the webinar. Who won, UX or CRO?

Watch the Webinar Replay

Listen to the Podcast

We shot this webinar because I had two things happen in the past year that made me wonder if we shouldn’t be doing more UX as a part of our CRO efforts.

First, we helped redesign a client site using conversion optimization. During the redesign, the client experienced significant increases in demos and sales of its software. To date we’ve almost tripled their demo requests.

Then, I happened across a landing page that I felt was very well done. When I asked the designer of that page how they had arrived at that design, Adam Treister told me they had done a UX process on it. And he had documented the process in a Udemy course. The page increased enrollment clicks by 246%.

Two different approaches. Two great results. I invited my UX friend Sarah Jabeen of DiscoverSTEAM to debate this with me. How are these two process different? How are they the same?

There is only one way for you to find out.


21 Quick and Easy CRO Copywriting Hacks to Skyrocket Conversions

21 Quick and Easy CRO Copywriting Hacks

Keep these proven copywriting hacks in mind to make your copy convert.

  • 43 Pages with Examples
  • Assumptive Phrasing
  • "We" vs. "You"
  • Pattern Interrupts
  • The Power of Three
  • This field is for validation purposes and should be left unchanged.

You’ve decided it’s time to undertake a website redesign. Should you focus on improving UX (user experience), or should CRO (conversion rate optimization) be your priority?  Are they mutually exclusive?  Is there a time when one is more important than the other?

Sarah Jabeen from DiscoverSTEAM tackles this issue with Brian the Conversion Scientist in a UX vs CRO (replay).

You’ve probably guessed that Brian has CRO in his corner; Sarah will be leading with UX.

Watch the discussion.

You’ll still walk away with valuable information you can incorporate into your site redesign including:

  • Do you have to choose one or the other between UX and CRO?
  • When should you focus on UX, and when should you focus on CRO?
  • How does CRO inform UX?
  • What do you do when tests recommend bad UX?
  • What are the similarities of the two processes?
  • What are the differences between the two processes?

Watch the replay. I hope I have you in my corner.

How a B2B eCommerce company used a stepwise strategy for their website redesign and got 250% more leads before they were done.

We recently began the split-testing process for a B2B eCommerce company. This is only remarkable because we signed this deal over a year ago. What happened to delay testing so long?

A website redesign.

The four-month redesign turned into a six month redesign and then into a 14-month redesign. This is not unusual in our experience. The new design has launched and, after all this time, the conversion rate and revenue per visit remained about the same.

We see this as good news. Too often, redesigns actually decrease site performance for a period after launch. There are storied website redesign disasters, such as FinishLine and Marks&Spencer.

Nonetheless, the conversion optimization testing was delayed. They can never recover the revenue or the lost testing time. Let’s see then how this website redesign got 250% more leads before it was finished. Step by step.

How to Make More Money During Your Website Redesign

Conversion Scientists look at websites quite differently. You may see a valuable online revenue engine. We see a laboratory for growing sales in petri dishes, and then scaling that to business-changing proportions.

When Wasp Barcode came to us, our vision fell on the ears of a brave and daring team. The approach allowed them to grow the number of live demos by more than 200% in just a few months.

We started with our Conversion Catalyst, a six-month process designed to grow revenue quickly and permanently. We started by getting Wasp setup for website optimization. This included setting up the digital lab, a set of tools that includes analytics, click-tracking, session recording and split testing.

Download and Read the entire case study: This Website Redesign Got 250% More Leads Before it Was Finished-Wasp Barcode

Then, we went to work in a very unusual way.

Wasp Barcode sells inventory and asset tracking solutions. Their most profitable offering is a complete inventory- or asset-management system that may include software, scanners, and labelers for the things businesses need to track. The most effective way to help their prospects choose the right system is with a live demo. During this demo, a sales person will walk the visitor through their software and answer any questions they have.

Our main goal was to increase the number of visitors filling out a form to request a live demo.

We did a stepwise redesign, in which all of the assumptions about the new design were tested to ensure that they had a positive or at least neutral impact on demos. Our approach was this:

Step One: Test Things that Can Be Used in the Website Redesign

Our first step was to find the calls to action that would move more visitors to request a demo. This was a series of tests to find out what language should be placed on buttons. For example, we learned that language offering “Free Live Demo” or “Free Consultation” generated more clicks to the demo landing page and more completed demos.

Step Two: Test the New Page Design on a Portion of the Site

Their design team integrated what we learned into a redesign for one of the site’s product category pages. We tested this new design against the existing category page, the control.

Our tests showed that the new design did a great job of getting more visitors to the Demo Request page. By driving more visitors to this page, we had more resources to test lower in the funnel.

Step Three: Optimize the Demo Landing Page

We then went to work on the Demo Request Page, a page on which the prospect can complete a form requesting a Live Demo.

Our tests here revealed that removing video and adding a product shot increased form completions significantly.

This key landing page went through several tested iterations to reach a high-converting design.

This key landing page went through several tested iterations to reach a high-converting design.

The redesign was just getting started, and we had already begun generating significantly more demo requests for the business.

Step Four: Move to Another Section of the Site and Repeat

The Wasp design team designed another category page for the next section of the site. They integrated elements that visitors were clicking on frequently, such as feature lists.

While the visitors in this section of the site behaved somewhat differently, we saw a positive lift in visits to the Demo Request page. This page was optimized, and delivered more demo requests to the sales team.

The step-by-step Wasp Barcode website redesign sped up.

The step-by-step Wasp Barcode website redesign sped up.

The Wasp design team then took what we had learned and redesigned the home page. This drove a significant increase in visits to the high-converting Demo Request page.

250% Increase in Demos over Six Months

Most website redesigns would still be sitting on a staging server. Wasp has enjoyed significant increases in demos during their first six months. Together, we rolled their redesign out step by step, testing along the way to ensure each change had a positive or neutral impact.


Breaking the Rules of Website Redesign

Designers and UX people may be rolling their eyes. It is an old truism that a visitor should have a consistent experience across a site, or they will feel lost.

During our stepwise rollout, we violated this rule. But when we have completed the process, providing this consistent experience, we can expect another increase in demos.

This approach also allowed us to change the design for different sections of the site. Those visitors looking for Inventory Management solutions are fundamentally different from those looking at Asset Management tools. One design would not have worked well for both.

Not everything we tried increased the conversion rate, and the Wasp team made adjustments accordingly.

Let Us Guide Your Site Redesign

Your website redesign doesn’t need to be an “all in” gamble. Find out if your website would benefit from a stepwise redesign with a free consultation.

21 Quick and Easy CRO Copywriting Hacks to Skyrocket Conversions

21 Quick and Easy CRO Copywriting Hacks

Keep these proven copywriting hacks in mind to make your copy convert.

  • 43 Pages with Examples
  • Assumptive Phrasing
  • "We" vs. "You"
  • Pattern Interrupts
  • The Power of Three
  • This field is for validation purposes and should be left unchanged.

Raise your hand if you’re considering a website redesign.  Pretty much everyone, yeah?  Well, before you undertake such a massive project, there’s a lot you should consider first…namely the effects such huge changes can have on your conversion rates.
Some businesses will pour millions of dollars into a fancy and beautiful website redesign only to discover that their customers no longer know how to interact with (ahem, buy things on) the site.  In other words, a double loss.  If you think the design of your site is keeping visitors from spending money, consider an approach that’s a bit more slow-and-steady.
Brian suggests taking a scientific approach:  he’s a scientist, after all.  It has probably been some time since you’ve had to think much about the scientific method, so here’s a recap:

  1. Research
  2. Form a hypothesis
  3. Create an experiment design
  4. Run tests
  5. Tabulate results
  6. Analyze results

Do some research then come up with some small changes you can make and measure the effects of.
It’s a cycle that often ends with a surprise. Our visitors just don’t behave the way we think they should. There are great resources out there to help understand these people we call visitors, like Crazy Egg and Google Analytics.  Brian goes into a bit more detail in the webinar, but the bottom line is, don’t fret:  there are absolutely resources out there to help you get the job done.
Here’s where I could say “You know your customers best,” so you should be able to come up with a solid list of hypotheses with which you could experiment, but I won’t.  You should still come up with a list of ideas based on research, but you should be prepared for surprises.
And remember, he’s serious when he says to keep it scientific.  Isolate a single variable as much as possible so that you know for sure what is driving changes in your site visitors’ behaviors.
Watch Brian’s webinar to get an even clearer picture of where you can start on your redesign project.
Image licensed through Creative Commons by Kevin Dooley & adapted for this post.

Get out your number 2 pencils and practice your small circles. We fill in some bubbles for you to help you redesign your home page.

What if I told you that, when a visitor reaches your homepage, to them it’s just like taking a multiple-choice test? They have a question and you offer choices.


 Full column with charts and graphics

Does your homepage design punish visitors if they make the wrong choice? This is the purpose of those standardized multiple choice tests we’ve been taking since high school: if you guess, you are likely to get it wrong.

We don’t want to punish our visitors for guessing. We can attempt to eliminate the guessing, though.

In 8 Ways Your Home Page Is Like A Multiple Choice Test we explore the rules for designing multiple choice questions provided by the Scholastic website, and then see how these apply to our home page question: “Why did you visit our website today?”

21 Quick and Easy CRO Copywriting Hacks to Skyrocket Conversions

21 Quick and Easy CRO Copywriting Hacks

Keep these proven copywriting hacks in mind to make your copy convert.

  • 43 Pages with Examples
  • Assumptive Phrasing
  • "We" vs. "You"
  • Pattern Interrupts
  • The Power of Three
  • This field is for validation purposes and should be left unchanged.

Many business websites have to change with the holiday seasons, but you don’t redesign with each holiday.

We’ve been pretty busy here at Conversion Sciences. No complaints at all.
However, other parts of our lives suffer. In my case, it was decorating for the Christmas Season.
And in an unexpected way, this relates to your website.
I found myself with Christmas getting near and my Halloween decorations were still up. Before you laugh, isn’t this the same way you feel about your website? It’s needed a redesign for a while, but other priorities keep getting in the way.
That’s exactly how I felt.

There is little that can be done with Halloween decorations when preparing for Christmas... or is there?

There is little that can be done with Halloween decorations when preparing for Christmas… or is there?

This is not a traditional Christmas image.
But, since I was short on redecorating resources, I decided to redesign my decorations the way I would redesign a website: a little at a time.
I used small steps in my decoration redesign. I found some Christmas doll sweaters to put on my skeletons. Here’s the result.
This may seem a disturbing choice for Christmas decoration, but what does the data say?

This may seem a disturbing choice for Christmas decoration, but what does the data say?

Your first reaction may be that this isn’t much better. Honestly, that was my first reaction.
But I tested it against the most skeptical (and most important audience) I could: my 17-year-old daughter and her teenage friends.
The response was resoundingly positive and unprompted.
Go figger.
It’s unexpected, unique and didn’t require a decoration redesign. This is the same approach we recommend for your website.
You may think your site needs a re-do. You may feel it’s dated, familiar, or too old. As we like to say in the business, “Your opinion doesn’t matter.” It’s the opinion of your visitors that matters, and they may not see your site with as critical an eye as you. They may even love it as it is.

The Right Reasons to Redesign

There are two really good reasons for a redesign:
1. Your brand is changing completely.
2. Your site is not maintainable and needs a new foundation.
If you aren’t facing one of these two situations, consider a stepwise redesign.
We’ve been able to modify a site completely using testing tools before the business committed to the redesign.

We were able to test a complete redesign for this ecommerce site before committing it to code.

We were able to test a complete redesign for this ecommerce site before committing it to code.

We found with this test that the redesign would not be expected to have a negative effect on revenue.
Test the navigation design.
Test a revised value proposition on your home page.
Test a new layout for each of your important pages.
Test a new checkout process.
Test a new ecommerce category page layout.
Test a new mobile layout.
Piece by piece, you’ll learn what improves your bottom line and what doesn’t. We call this data-driven creative.

Faster Results than a Redesign

Doesn’t this testing process take more time that a redesign?
We’ve accompanied several clients through redesign. A website is a complicated piece of software. We’ve rarely seen a redesign come in in less than twice the time predicted.
So, no, we think our approach is faster.
And, as you find wins, you get to enjoy higher conversion rates, higher revenue and more leads as the site is redesigned.

Enjoy Your Holidays and Your Redesign

Don’t take our word for it. Author and optimizer Rich Page cited a Hubspot study that found one third of companies who implemented a redesign were unhappy with the results.
This doesn’t have to be you.
Having the resources to make your website a better place for your visitors is a great advantage over your competition. Let the competition spend months on one big shot.
Meanwhile, you can learn what your visitors want and deliver more of it month after month, until you possess a site that the competition will have trouble keeping up with.
Merry Christmas from everyone here at Conversion Sciences.

Our data-driven wish for a happy holiday.

Our data-driven wish for a happy holiday.


The folks at have put together a very complete infographic on mobile advertising.
We like articles and infographics that support their findings with research and case studies.
One thing we’d like to put a fine point on is this:

Formula: Number of mobile site visitors divided by the number of actions taken, all multiplied by 100 to give the conversion rate.

Responsive vs. Dedicated Mobile Site

We are seeing in the literature more evidence that responsive designs suppress mobile conversion rates. The primary culprit is load times. We are currently recommending the Native Mobile Website approach for phone-sized screens.
Furthermore, many sites are displaying mobile sites on tablets and phablets that have the resolution to show more. This may be suppressing conversion rates as well.
Everything You Need to Know about Mobile Ads - Via Who Is Hosting This: The Blog