Institute Conversion optimization services Blog Search Start 7-day trial for $1 Subscriptions Minidegrees Online courses Upcoming courses Resources

Adopting A Data Driven Approach To Increase Website Traffic

Adopting A Data Driven Approach To Increase Website Traffic

“Conversion Rate Optimization” principles don’t just apply to landing pages or capturing emails. That narrow brand of thinking severely limits the truly transformative impact CRO principles can have on a business.

In a big picture, can’t “Conversion” really be boiled down to getting more of the right people clicking on the things you want them to click?

What follows is the first in a series of articles that will include some of the most common areas where optimization strategies can be applied to achieve better results for your company.

These will be living documents that will include some of the best resources from here on CXL and around the web.

While this series is by no means exhaustive, we’ll make every attempt to look at the most common & critical areas in the marketing funnel, and how you can take a data-driven approach to improve each area of the funnel.

So Where Exactly Does “Conversion” Happen?

To guide the discussion, we’re going to be using Dave McClure’s Conversion Metrics as a framework to see where different conversion optimization opportunities lie.

Dave Mcclure conversion metrics

image source

For Acquisition, we’re primarily asking how do first time visitors & potential customers find you?

Specifically, how do we get more clicks from the people who might be great customers, but are completely unaware that we exist?

It is important you realize that we’re not talking about what happens after they click, or how to capture their information, or how we hold their attention after they click. All we’re talking about right now – exclusively – is how to stand out enough to grab their attention in very crowded environments.

This could cover a range of methods, but for the sake of this article, we’re going to focus on PPC, Social Media & Search.

7 Ways that Predictive Analytics Is Transforming Ecommerce

By Gagan Mehra

Predictive analytics help you understand what your customers are going to buy before they do.

  • This field is for validation purposes and should be left unchanged.

Before We Get Started – Understand Your Target Market, REALLY Understand It

Your success online is directly proportional to how well you understand your target market.

As cliche as that sounds, it needs repeating. If you don’t know anything about your best customers, how can you acquire more like them? How can you expect to get referrals or understand what prospects will click if all you’re doing is guessing?

If you have just a few customers, one of the easiest wins you can have right now are to find the areas within your funnel that are prime for direct interaction with customers and build in feedback loops.

This might include:

  • Welcome Emails
  • Purchase Receipts
  • Thank You Pages
  • Survey For Customers who’ve been with you for 3/6/12 months
  • Unsubscribe pages

The idea is to build these feedback loops to understand the people behind the order numbers and email addresses. What do they need? Where do they “hang out” online?

This feedback can be used to improve your acquisition strategies, your value proposition, and your processes on every level.

If you don’t have customers, our articles on customer development & writing compelling copy when you don’t have any customers, to help you to start getting people to buy.

Recommended Reading:

Acquisition – Getting Visitors At Top Of The Funnel

Ok, now that you have a deep understanding of your customer, I have to tell you something you’re not going to like hearing:

Most people aren’t online to see your ads.

As a general rule, people aren’t searching Google excited to click your text ad and nowbody is browsing Facebook hoping your sponsored story gets in the way of them seeing their niece.

Regardless, your job is to get them to click. The question is – how?

The simple answer is to focus on crafting quality headlines, show stopping graphics, or difficult to forget experiences and showing them to a highly targeted audience.

The simple answer is a cop-out.

To get attention, you must create something that speaks directly to my needs or is truly worthy of interrupting my usual browsing patterns.


Pringles – for example – created a banner ad that told a fun little story and encouraged visitors to click over 90 times. The result? Massive coverage on Twitter, Reddit, Fark & other relevant blogs as well as a Gold Cyber Lion advertising from Cannes .

Australian Air Force Banner

aussie airforce

The Australian Air Force ran the above banner ad on an Australian IT news website.

Upon inspecting the source code, visitors would find the following.


Even with a limited budget, this ad was able to win an advertising award, massively increase job inquires for the Australian Airforce, and reduce the average cost per enlistment.

Considering that the “ad” was hidden in the source code of the page, this means it spoke directly to the target market – because nobody else could see it!

Capture Attention – But Don’t Be Irrelevant (or Offensive)

It’s tempting to do the digital equivalent of shouting from the rooftops, but it’s important to remember that what you are after is quality traffic, not just random big number of visitors.

What I mean is that when you’re focusing specifically on acquiring traffic, try to avoid sensationalist headlines or “shock” graphics, if those tactics are not in line with your overall brand messaging.


The above is a good example of what not to do from retailer Tradesy’s Facebook page. It got attention (and probably some clicks) but at what cost? It’s too bad, because Tradesy is normally on point with their brand messaging.

Controversy for controversy’s sake is rarely a good idea.

Optimizing PPC & Display Ads – Researching Your Target Market

It’s tempting for PPC managers to go after whatever will generate the highest amount of clicks, but without measuring the lifetime value of the customers that come in from those clicks, you’ll waste a lot of time & money.

  • Do the majority of people look at the page and close it right away?
  • Are you satisfied with the conversion rate from visitors to leads?
  • Of those leads, are you satisfied with the conversion from lead to customers?

It’s not enough just to generate clicks, but clicks from the right target market & people who are will move through your sales funnel with the least amount of friction are ideal.

More than half the battle is doing the market research to understand what potential visitors might be searching for, what other “interests” they might have that are synergistic with the things you sell, and where they hang out online.


If you don’t understand at a very basic level where your target market is online, how can you test anything that might bring them to you?
Recommended Reading

Split-Testing Ad Creative To Lure In The Market

Only once you understand where your market “lives”, and the segments within it (problem unaware, problem aware, expert level, etc) should you start testing the ad-creative that’s going to get their attention.

To ensure you don’t end up burning a ton of money, start testing your ad creative on a sample group that is large enough to get reliable data, but not so large it’s going to cost you a fortune to run the campaign.

Once you’ve found some ads convert reasonably well given the whole picture(impressions, CTR, Conversions, profitability), increase the budget & roll them out to more people to see.

Hubspot brought out in one of it’s articles 6 split-testing tips for Facebook ads. In general these are great rules of thumb when starting your ad campaigns.

1. Change One Variable at a Time

Your main variables are the title, the picture (if any), the copy, and the targeting.

Given your research, you should test radically different approaches for each segment of your market as well as gradual iterations that test one variable at a time.

Split Testing Ads

image source

In the above example on Facebook, the only variable being changed was the picture of the ad – the headline and copy stayed exactly the same.

Split Testing Google Ads

image source

In this one, changed the body copy in their Ads ad to focus more on “saving money”, while the second one uses language that targets beginners.

2. Keep Similar Ad Conditions

Same time of the day, same bid (although bid prices vary), same length of time, etc.

Pro tip: only compare ads that ran in similar conditions (location, devices, start and end dates and the like), if you change them, it’s impossible to know for sure which of the variables did the trick.


3. Gleen Insight From Your Reports

On the surface it might look like that one of the variations is winning, it doesn’t mean that it’s actually the best one.

What matters is not just the number of people that clicked on your ads, but of those visitors, how many responded to your call to action on the destination page.

Report Insights

image source

Long term, you might wonder the lifetime value of customers who come from specific ads, or if visitors from one ad convert faster than others.

While that’s a conversation for another day, it’s important to re-iterate that, even here, “optimization” doesn’t just occur once on a single page, but can have a dramatic impact on the entire customer lifecycle.

On the above picture you can see circled how your ads are converting, how much it costs to acquire that conversion and finally the conversion rate.

On the surface, the second ad group seems to be converting better but it also costs double to get those conversions. While it’s also a little blurry, you may notice the second ad group is for “Discounts” so even if the conversions are good, the product is not being sold at full profit margin.

If these ads do not bring customers who continue to buy at full price, you are literally spending money that you will not get back.

4. Realize When It’s Time To Create a New Ad

It’s tempting make endless tweaks to an under-performing ad & sometimes new body copy or adding a symbol can make all the difference in the world.

And sometimes it’s just time to know when to kill your darlings.
Fortunately, in Google Ads, you’re able to keep your targeting on both the campaign & adgroup level, so you can focus on writing ads specifically for that target group.


image source

Facebook has recently unrolled a similar structure that allows for more granular targeting on the individual ad level.

With this level of granularity, it’s possible to test multiple ads that target the exact same people, or if you feel like an ad isn’t performing properly, to switch the “likes and interests” targeting while maintaining other targeting parameters like gender & location.

create similar advert

5. Rotate Your Ads Often

Even a well-performing ad will wear out its welcome.

How often you rotate an ad will depend on a variety of factors like the medium you’re advertising on, the amount of unique impressions from the given medium, and how many variations you’re running in the same ad set.

Facebook users & popular blog readers for example, tend to visit more frequently. Therefore they’re exposed to your display ad more frequently.

TechCrunch sidebar ad

If you’re split-testing 2-3 ads in a set and notice that one of them is sliding off the radar, pause it when it’s performing around 20% worse than the others (or sooner).

However, if the ad’s that slipping in CTR is killer at converting, it might not be wise to dump it until it’s under a 1% CTR.

What’s more important than the arbitrary benchmarks I’m giving you here though, is that you measure performance against yourself.

Understand that your ad could be performing low due to season changes, events, or that you’re simply measuring against a temporary traffic spike. Just don’t approach PPC with a “set it and forget it” attitude.

6. For Display Ads – Understand What is Visually Appealing To Your Market

Why try to create display ads in the dark when you have a whole wealth of data about what your market finds visually appealing just a search away?

What do I mean?


By doing a graph search on Facebook, you can see what general visual styles are appealing to specific groups of people.

Looking at a search for “favorite bands of fans of Peep Laja” you notice a darker, grittier theme among the visuals being used. This is relevant, because the people connected to the page have already indicated:

  1. They’re interested in the band
  2. How the various bands, movies, television shows etc. represent themselves visually speak to a much larger visual preference about the audience you’re trying to target.

Let’s look at how this can be applied in a real world situation.

Imagine you had a client who sold interior design services & wanted to run a display campaign.

Instead of guessing which styles might appeal to your core (or local) market, you can do a little competitive analysis to see what resonates with the people nearby.

First, you’ll want to see who your competition is:


Once you have the list of who your Facebook Savvy competitors are, a search like “Photos posted by [Competitor] will show you all of the most well received images posted by the page:


At first glance this might not seem like much, but it looks like what gets the best reception is the fireplace, bathroom & living room designs. There’s also a more rustic feel from these photos, which gives some indication of at least this particular audience’s preferences.

Looking at the most liked photos from other competitive firms starts to reveal a theme about the area & with a little creative thinking, a possible creative direction for a display campaign.


While this is no means the only way to go about finding inspiration, it is a way to get indirect feedback from the market so you can come up with more informed ways to get your market’s attention.

Tangential Bonus: If you do a graph search for “Pages liked by fans of [competitor page]” & add those pages to a dedicated facebook list, you can get a real time feed of the images everyone is trying to use to grab your market’s attention.


Recommended Reading:

Using A Data Driven Approach To Social Media Marketing

When it comes to social media, the first things you need to figure out is where your ideal customer is most likely to be “hanging out” & who they’re already connected to.

Secondly, please realize that there is a whole world to “social media” that exists outside of Facebook Pages, or branded Twitter accounts.

  • Facebook Groups
  • Google+ Communities
  • Twitter Hashtag Chats
  • Instagram Hashtags
  • Shared Pinterest Boards
  • Specialized Forums
  • Even The Comments On Active Blogs

These are just a few places where your target market are more inclined to be hanging out & interacting with each other.

Where you’ll engage and how will depend on who your target market is & where the individuals who are in these groups are at in terms of readiness to hear your “pitch”.

Beyond that, different people conduct themselves differently on various platforms. As a by-product, they also expect different kind of content to fit within the culture of the platform.

Gary Vaynerchuk“I’ve got to do something different in Tumblr than I do in Facebook because the audience is different. If you want to win, you need to roll different. We’re in slang marketing now. You have to know the slang of the room you’re storytelling in. Otherwise you’re out. You can act really out of touch real quick.”

Gary Vaynerchuk

Using Data To Find The Best Approach For Publishing Content on Facebook


Assuming you have some people that aren’t your friends and family connected to your social media accounts, there are analytic dashboards (for the major social networks anyways) that show you how each piece of content performed.

Facebook for example, gives you loads of info including how many people saw the post (Reach), what type of content – links, photos, status, etc. – got the most interaction, how much engagement each post gets and more.

With this data, you can move away from the “post everything we publish” Facebook strategy & instead try to curate high performing content based on previous post performance.

Notice in the CXL Facebook page data, the top 7 posts since February are:

  1. Asking for Interaction (1. website reviews, 5. you be the conversion expert)
  2. Giving Valuable Tool Lists (2. Steve Blank resources, 6. category page for website reviews)
  3. Related to eCommerce or Design (3. high converting ecommerce site. 4. free shipping 7. persuasive design)

Beyond that, there’s also something to be learned from how the images within the links are presented, how the post was worded, and the subject matter itself.

For example, I’ve learned that UX & Persuasive Design articles perform better on Facebook, while Analytics performs better on Twitter.


Outside of content, Facebook insights also helps you understand the times of day your fans are most active, where they are in the world, the common languages spoken, and demographic breakdown.



By actually using this data, combined with Facebook’s advanced targeting features that allow you to display posts to specified genders, relationship status, age, location, etc. You can really focus your publishing efforts in a way that directs content to the right people at the right time.


The team over at Post-Planner experimented by sending a post that said “good-morning” to their Facebook audience at 7am in local time zones & local languages.


Early reports showed an increase in organic reach between 10-20%, which is huge, considering Facebook’s efforts to drive organic reach down in the past year.

What it comes down to is this – If you understand how the channel actually operates – rather than how you’d like it to operate – you can continuously refine the way you present your content to get more clicks, comments, shares & organic reach.

Recommended Reading:

Using Data To Improve Your Headlines On Twitter

On Twitter, the headline that you attach to your article when tweeting it out can have huge implications on how it performs.

Buffer has done a lot of testing on Twitter with different headlines and this is the process they recommend:

  1. Come up with 2 different headlines for an article that you’re about to send out
  2. Tweet both of them out  at roughly the same time, at least 1 hour apart. Tweets that are either in the AM or both in PM work the best, clear “morning” and “afternoon” times are crucial.
  3. Compare the data and choose the winner.

First tweet:

Tweet one

image source

Second tweet:

Tweet two

The second tweet got much more engagement, and so was chosen as the one to recycle later – as sharing tweets more than once typically leads to better performance the second time around.

Recirculation RTs

image source

Also, you may not realize this, but if you go to, Twitter itself has a pretty robust (free) dashboard that tells you all about your top performing tweets.


What’s interesting to note here are the correlations between increased reach, @replying others & participating in specific hashtags (like #Mozinar)

It’s also interesting to see how – in some cases – the number of retweets, faves & replies don’t necessarily translate into clicks.

If you want to take Twitter Analytics a step further – and you have Twitter Cards installed –  you can also check out the Card Analytics data, which will show you the relationship between impressions & clicks, change over time, and more.




Beyond improving headlines & images on Twitter, it’s also vital you’re distributing content at the times your followers are most likely to see them.

This is where my favorite tool Followerwonk comes in. Using the time of day report, I can easily see when most of my – or other influencers –  followers are online.

I’m also able to see roughly where in the world my followers are tweeting from:


Search Twitter Bios for targeted keywords, locations, and min-max followers:

Followerwonk Bio Section

As well as segment mine – or anyone else’s – followers by social authority.


While there are a number of ways this can be helpful, as it relates to your acquisition efforts on Twitter, it’s important to build a solid network of relevant tweeters with similar audiences to help you distribute & amplify your messaging.

Twitter ListPersonally, I categorize who I follow and organize them into Twitter Lists, then use TweetDeck to keep track of each list in real time.

Using this basic technique, I’ve built specific lists around newsletters I’d like CXL to be featured on, relevant industry people, and around few other strategic goals.

It’s still too early to tell, month over month data for CXL shows a 75.54% increase in Twitter traffic in May, when I started really committing to the technique.


Part of the reason I believe it’s been successful, is that I’ve been sharing content with influencers at times they are talking about something relevant & similar.

Again, only time will tell if this technique is going to prove sustainable, but understanding that the best acquisition is about getting the right message, in front of the right people, at the right time, I believe committing to these techniques on Twitter will prove valuable for us in the long run.

Recommended Reading:

Other Social Networks

It doesn’t make sense to go into all of the social networks analytics data, please realize that there is a lot that can be learned about your audience and the type of content that’s going to grab their attention, when you really pay attention to the analytics native to each social platform.

Our Slideshare:


Our Top Performing Youtube Content:


What to look for in data?

You should be analyzing 2 main things in your social data:

  1. How successful is that channel at driving quality traffic for you?
  2. The type of content that performs best over that channel

From the data you have gathered you’ll start to see patterns appear on what works best and what doesn’t. Armed with that knowledge, double down on the things that work.

Getting social right is all about experimenting, looking at the data and learning. Then using that knowledge to iterate the next time around.

Recommended Reading:

Re-Optimizing For Search Traffic

There are so many more things that can be improved at the top of the funnel when it comes to acquiring traffic.

For instance, re-optimizing title tags & meta descriptions to be more “click-worthy” in a busy search result page can have massive results on your bottom line as can things like rich snippets.

Using Google Webmaster Tools To Find High Impression, Low CTR Key Phrases

To find out Click Through Rates for all the different keywords you rank for, check out your Google Webmaster Tools data.

After navigating on the left hand side Search Traffic -> Search Queries section,  you will see the average Impressions, Clicks, CTR, and Position for all of the keywords you appear in search for.


Have The Data, Time To Optimize

What you need to pay attention to are Impressions, Clicks and CTR and in addition to that look at GA for things like bounce rate and time on site for the pages that correspond to those specific keywords.

High number of impressions and a decent CTR rate?

High Impressions & CTR

So far so good. Move into GA for that specific keyword. Low bounce rate,high time on site? Everything’s good here, move on to the next one.

What if I have high bounce rate (relative to the goal of the page) but impressions and clicks are decent? If they don’t bother spending time on the page, how can we expect them to convert further down the funnel?

Most likely what people are looking for and what you’re offering do not match.


In this case you have 4 Options:

  • You can try tweaking the Title and Meta tags
  • OR build more links to the page to increase rankings
  • OR go after a different keyword
  • OR rewrite the content itself

Before I determine which will be the best tactic, I have to do more research.

It could be disastrous if you just went and re-wrote all of your title tags without knowing the overall effect it might have.

In an ideal world, I’d want to start with the tactic that takes the least amount of effort, I’ll check the Search Results Page (Serp) for Customer Lifetime Value & see there’s already a ton of content around “calculating” customer lifetime value.


If I’m going to rewrite this title tag, the idea would be try to use language that speaks more closely to the user’s search intent.

Digging into the webmaster data again, I see how else visitors are arriving to this content.

CLV pt 2 CLV pt 3

This along with the related searches at the bottom of the main Customer Lifetime Value search indicate that moving away from the “calculating” language might not be a good idea.

CLV related searches

Though, I might test something more compelling like “How to Calculate & Improve Customer Lifetime Value” & play with the meta description to be more clickable. Rewriting the meta descriptions to include calls to action was one of the things Redstar did to bring 200% growth for a client.

(Note: Due to some visual changes Google’s been testing, I’d use Moz’s Title Tag Preview Tool for proper formatting)

New Title Tag Guidelines Preview Tool Moz

Considering that rewriting titles probably won’t better match user intent, my next option is to increase the number of links that point to this particular url in hopes that it might increase the page’s ranking.

To understand what my competition looks like, I use Moz’s keyword difficulty tool to get an overview of how my “competitors” in this search stack up in terms of backlinks & overall domain authority.


Looking at this analysis, it’s not too difficult to understand why Wikipedia or Harvard Business School take up the top 2 spots.

In the number 3 spot is Custora, which if we’re measuring by backlinks, makes no sense, however because most of their content is customer centric Google might place it higher due to their domain expertise. After that, there’s a fairly standard correlation between domain/page authority and ranking position.

Hubspot, Entrepreneur, KISSmetrics and CXL all write on a variety of topics related to business & online marketing & these positions likely fluctuate pretty regularly. In fact my rank tracking data confirms that.


I suspect if I were to build more links to the customer lifetime value article, and perhaps make it a little more “beginner friendly” (due to low Time on Site and high bounce data) this page, in time, would edge it’s way up in the ranking closer to the #3 or #4 position.

This shouldn’t be too difficult considering the relationships I’m building with the Twitter techniques from earlier.

Adding Rich Snippets To Improve Visibility In Crowded SERPS

In addition to Title and Meta tags that you see on the SERP, there are also other visual extras like images, authors, video, events, products etc – all these are collectively called rich snippets.

Rich snippets provide extra credibility and typically increase CTR – up to 30% more – by including extra “bits” of information in the search result page.

Implementing Rich Snippets

Implementing Rich Snippets

image source

Getting rich snippets to work involves three steps:

  1. Pick a markup format
    Google itself suggests using Microdata BUT it accepts all three kinds of structured data.The 3 supported markup formats are:Microdata

To help you with the implementation, Google has made an Structured Data Markup Helper.

In addition to author photos & reviews,  Google also recognizes markup for video content. Read all about it here. To make sure everything is setup the right way and it actually works, you can use the structured data testing tool from Google.

I’m not going to lie, implementing structured data is a lot to wrap your head around at first, but the team at Moz has put together a graphical guide into the world of microdata and rich snippets that should prove useful – check it out here.

CXL uses Yoast SEO & Video SEO to make getting video thumbnails into the search results much easier. We will also be experimenting with using new video thumbnails to test the impact on average CTR on video content.

video serp

A quick word of warning: Google has started cracking down on rich snippet spam, and doesn’t guarantee that every ranking will include rich snippet data. However, if implemented in a useful way, it can help you to improve the clicks that come from organic traffic.

Recommended Reading:

Conclusion – There Is So Much You Can Do To Optimize Acquisition

Here we are, close to 5,000 words later, and we’ve just barely scratched the surface of what’s possible when you use data to guide your acquisition strategies.

If you’ve just been throwing stuff at your Facebook wall or Twitter stream, or if you’ve been blindly publishing articles hoping that traffic will come from the search engines, stop.

Take a step back, analyze your data, and see where you can improve.

Because if you can’t get people to visit you in the first place, how are they ever going to buy from you?

featured image credit

Big thanks to our new staff writer Ott Niggulis for the help on this article.

Related Posts

Join 95,000+ analysts, optimizers, digital marketers, and UX practitioners on our list

Emails once or twice a week on growth and optimization.

  • This field is for validation purposes and should be left unchanged.

Join the Conversation Add Your Comment

  1. Amazing article! practically a manual for growth hacking :)
    I like how you suggested the use of feedback to understand better your users with surveys and polls.

    This is exactly what we do: :) check it out.
    We are also SurveyMonkey partners. We enrich survey respondents data, big data for opinions and retargeting.

    Happy to know your blunt thoughts on our product and how to be better!


  2. Awesome article !

    The “user intent” part was very beneficial to me as i’m getting started with Ads. It will help me develop better ads and landing pages that speaks to where customers are at in the buying cycle.

    Make your days great

  3. Wow, this is amazing. Can’t wait to see what’s next in the series. Thanks for posting this.

    1. Tommy Walker

      Thanks Becky!

      I’m glad you enjoyed it :-D

Comments are closed.

Current article:

Adopting A Data Driven Approach To Increase Website Traffic

How to change your marketing during and after the crisis

The world has already changed. Your marketing should too.

Check out our Live Panel series with experts in four different industries.

These leaders will discuss how to navigate uncertain times in SaaS, eLearning, Ecommerce, and the agency landscape.

Register today