Seek Out & Improve: How to Spot Optimization Opportunities in Your Analytics Data

Optimization opportunities analytics

You’ve set up Google Analytics on your website. You’re greeted by great-looking dashboards every time you log in (provided you’ve set everything up right). But you still don’t know how or where to start improving your metrics.

Sound familiar?

Nearly every website has Google Analytics set up to track basic metrics so they can analyze what’s going wrong (and right), and use that information to improve their website’s performance. That’s the purpose of analytics.

Yet most people stare at their Google Analytics dashboards and think, “Well, I guess that looks good. We’re up .0002 percent from last month.”

The question we get from many ecommerce owners is: “Based on all of this data, how do I improve? How do I make more money? How do I sell more?”

And it’s a valid question. KISSmetrics notes:

You can use quantitative analytics to understand how customers use your product or service. This will give you data points that back up what value you think you give to customers based on what features they use most often. You can find out not just who the habitual users are, but how often they use certain features and which features they use.

If you’ve set up basic tracking on your site, you have an ocean of quantitative data at your fingertips. It’s a lot to wade through, but in those murky waters lies treasure: Hidden optimization opportunities!

Optimization opportunities - analytics data
It’s time to ditch the floaties and dive deep

If you’re happy playing in the shallow end of the analytics pool, you might as well not track anything at all. You’ll only find the good stuff when you venture into deeper water. That’s where optimizers look for experiment ideas, and begin to formulate hypotheses.

In this post, we’ll cover how to use your quantitative data to unearth a gold mine of actionable analytics insights that can lead you to immediate conversion improvements.

The main types of quantitative data we’ll be working with are:

  • Data from web analytics (Google Analytics, for most of us)
  • Data from heat mapping and scroll mapping tools (usually separate)
  • Data from form-tracking tools (again, usually separate if you’re looking to get very granular data)
  • Data from various types of surveys, which may also qualify as quantitative data once the responses are consolidated (most surveys results are analyzed using cluster analysis and turned into numerical data)

Now, how do we derive conversion optimization opportunities from all of these data?

3 prerequisite steps for finding insights in Analytics

Step 1: Understand traffic patterns

At the most basic level of analysis, you’ll notice whether your website traffic is increasing over time, remaining stagnant, or decreasing.

But when you really begin to examine your traffic data, you can begin to understand the behavior (and even motivations) of your visitors.

We do this by drilling down to where the traffic is coming from, and determining the quality of those sources. How are people finding you? And are the people who find you through Facebook ads vs Google ads more likely to buy? These are the kinds of questions we want to answer.

To drill down, we use the Acquisition Reports in Google Analytics to identify the best-performing sources of traffic. We look not only at the amount of traffic an individual source brings to the site, but at how visitors who come to the site via that source behave and interact with the website.

Do they bounce?
Do they look around for a while, then leave?
Do they convert?

The answers to these questions indicate the quality of the leads each traffic source provides.

Analyzing traffic sources will tell you which sources lead to higher conversion rates, and which don’t. And that tells you where you should invest more in advertising, and where you can cut your losses.

In Google Analytics, you can also go to Behavior Reports and observe the ways in which visitors move around the website. Then you can start to see patterns based on the source of that traffic. For instance, if you notice that visitors coming from one source convert better than visitors from another source, you might start asking questions like, “What makes Facebook users more likely to convert on my site?” or “Why is our AdWords paid search campaign not delivering higher on-site conversions?”

Here is an example of a report that shows you individual traffic sources and how they behave
Here is an example of a report that shows you individual traffic sources and how they behave

Note that the problem might not be the sources themselves, but rather how your ad appears on those sources. For example, it’s a common problem to have a Facebook ad that doesn’t match its corresponding landing page, which can lower conversions considerably. On the other hand, some sources can simply be better at attracting your target audience than others.

Questions like these can provide you with ideas and hypotheses on which you can base your experiments. Or you might uncover a problem that needs to be solved in order to make users more likely to convert.

Just remember — every idea you have must be confirmed through data. If you’re not sure that the data is telling you the whole picture, go deeper. In statistics, correlation does not imply causation. Every insight must be confirmed by multiple data sources. (That said, if you uncover an obvious problem with an obvious solution, go head and fix it as soon as possible!)

Step 2: Track events (so you know what people are doing… and not doing)

The majority of every ecommerce website should be oriented toward persuading visitors to buy something. Inevitably, though, there will be some content that doesn’t directly convert visitors, instead serving as a persuasion or nurturing tool to help visitors proceed to the actual conversion.

This content typically requires some time of visitor interaction, whether reading copy, viewing videos, completing a quiz, signing up for a list, or any number of other actions.

Event tracking enables you to observe when, how, and for how long visitors interact with your website content — in essence, you’re measuring “visitor engagement”. Tracking events can show you how far down the visitor scrolled on a page, how long they played a video, and so much more.

In fact, you can track just about any interaction using event tracking scripts.

The event tracking overview in Google Analytics
The event tracking overview in Google Analytics

Event tracking metrics tell you whether a given piece of content offered fulfills its intended purpose: to nudge the visitor further down the conversion funnel. If you notice that engagement is low and conversions are down, you may need to reconsider your content.

Essentially, the content visitors are viewing isn’t motivating them to convert. This could be due to a weak value proposition or vague copy, or something purely technical — like a broken image or videos that take too long to load.

Alternatively, you might find that you have a relatively good conversion rate, in which case you could cut that content or experiment with improving it, with the hypothesis that your conversion rate will increase.

Once more, be careful not to ascribe causation to correlation. However, it’s safe to say that you can always improve your copy.

Step 3: Divide & conquer with segmentation

One of the most useful tools you’ll find in Google Analytics (and other analytics tools) is segmentation.

Segmentation makes it easy to group your visitors according to specific criteria, and view the behavior of the selected segment. This way, you can define your target audience and determine whether your value proposition, layout, etc. makes sense to them.

Depending on the nature of your business and the type of the product you offer, you’ll want to segment by different criteria. For example, you can segment by geographic location, age, or gender. If you notice one group has a higher conversion rate, it would make sense to deliberately target this group by posting relevant ads at their primary source, or using language in your copy that is specific to that particular group.

If you notice a specific segment or demographic that is underrepresented, but that you think could be a profitable target audience, you can create hypotheses to enhance the appeal of your site, and beef up marketing to bring that demographic in.

It’s a funnel, not a sieve — so plug those holes

One of the best ways to increase your conversion rate in the early phase of optimization is to plug the leaks in your sales funnel.

Luckily, the best tool to check your funnel for issues and leaks is Google Analytics’ own Goal Funnel Visualization report. You only need to set up goals properly on your Google Analytics account to benefit from this report.

An example of a funnel visualization report
An example of a funnel visualization report

Always bear in mind that each user who enters your sales funnel by adding a product to the cart is, most of the time, more than a potential customer. They’ve already expressed their willingness to buy a product from you in no uncertain terms. The fact that they did not proceed to actually buy indicates that you didn’t do enough to motivate them to complete the purchase.

There are many hypotheses and potential reasons for cart abandonment:

  • Maybe your visitors may lack trust in your website, product, or value proposition, which makes them fearful to buy from you
  • Maybe your checkout page has too many distracting elements
  • Maybe you didn’t give them an option to purchase as a guest
  • Maybe your account creation form is too long or requires too much personal data

Whatever the cause of friction may be, it’s worth dedicating significant time to finding it and reducing the dropout rate.

Focus on the low-hanging fruit (AKA your landing pages)

Another potential area of opportunity for conversion optimization is your landing pages. Once you initiate an ad campaign or any other type of promotional campaign, you need to ensure that ads, email links, social media posts, and top-level promotions lead to their own, dedicated landing pages.

The only way to ensure conversions is to create a relevant and clear landing page that matches the look, copy, and content of your ad, and also appeals to your target audience (because you’ve targeted your ads to specific audiences… right?).

This is your low-hanging fruit. Optimizing these pages is easy, quick, and gets results.

Why? Because the people who follow your ads are pretty much in the same category as the people who enter your conversion funnel. They have expressed their interest, and they’re so close to buying. You analytics tool can tell you how popular and engaging your landing pages are — and conversely, if you notice high bounce and exit rates from your landing pages, you’re getting something wrong.

Remember that your customers are not only deciding between buying from you and not buying at all. They’re also considering buying from someone else — and you must offer them a reason not to do that.

Enter your quantitative data! Most analytics tools, such as Google Analytics, KISSmetrics, and Adobe Marketing Cloud, offer a report on top landing pages. By analyzing the metrics in that report, you can better understand the performance of each landing page in the context of conversion or engagement.

The Landing Pages report can be used to compare individual landing page performance and identify issues
The Landing Pages report can be used to compare individual landing page performance and identify issues

Most importantly, you’ll want to check how your landing pages for ads and PPC campaigns perform — since after all, you’re paying for that traffic. You can do this by adding a secondary dimension to the report and checking out your acquisition channels.

By adding a secondary dimension (you can select it by clicking the smaller red box) to your report, you can identify which landing pages perform best (shown in the third column, highlighted in red)
By adding a secondary dimension (you can select it by clicking the smaller red box) to your report, you can identify which landing pages perform best (shown in the third column, highlighted in red)

By using this information, you can make further adjustments or spot issues that may indicate an ad-landing page mismatch. If any landing page has a higher-than-average bounce rate or a lower conversion rate for ad traffic, you can either improve the page, or downsize or discontinue the promotional campaign.

In the example above (and quite unsurprisingly), a CPC campaign that leads to the home page as its landing page results in zero transactions. This means the whole campaign is a waste of money — it should be discontinued or linked to a specific landing page instead of the home page.

There are also dedicated landing page evaluation tools, such as Wordstream Landing Page Grader. Of course, no landing page can be evaluated without using some form of heat mapping, such as Crazy Egg, MouseFlow, or Lucky Orange.

Identify potential issues at a glance with heat maps

By using heat mapping on landing pages that you’ve identified as containing issues, you can check for possible problems with the page’s layout or design. Heat maps help you analyze how far users scroll, where they move their mouse, and where they click, all of which is useful information.

For example, you might realize that the call to action is not prominent enough, or that there are distractions that lead people away from the landing page.

Click map as displayed in most mouse-tracking tools
Click map as displayed in most mouse-tracking tools

You can also use session recorders to try to spot usability issues. Most mouse-tracking tools offer session recording as an integral part of the tool. Session recordings may also serve to jumpstart your user testing program — though nothing can replace properly conducted user testing.

Ease friction by tracking form activity

Finally, most landing pages contain some sort of form — whether those are email address captures, payment forms, or surveys.

By using analytics tools that contain a form analytics module, like Formisimo or the aforementioned MouseFlow, we can quickly spot issues with forms that affect the overall performance of a landing page.

An example of a form performance report
An example of a form performance report

By adjusting the form layout, changing the sequence of the fields, or adding clearer, more helpful form field instructions, we’re able to drive considerable improvements in form submission.

Add site search (it’s a win-win)

Nothing is more frustrating to than landing on a site, knowing exactly what you want, and being unable to find it.

That’s why an internal search function is a must for decreasing bounce rates and increasing sales. Some studies even indicate that visitors who use site search are responsible for generating a larger proportion of revenue than their attributable share of total traffic.

The fact is that site search helps conversions on the visitor side. A properly implemented search capability enables visitors to search categories, products, and other content, and ideally provides suggestions and recommendations based on search results.

Plus, on the analytics side, your analytics tool can track the use of your internal search and provide you with useful insights.

The most useful insights from site search are the keywords that your visitors use to find products on your website. Once you know these keywords, you can use them for ad campaigns to drive increased visits to your most popular product pages, or pop them into your social posts.

Most analytics tools offer reports on the performance of site search. Using these, you can see how search contributed to your site’s overall performance.

The site search report overview can give you valuable insights, like the keywords visitors use
The site search report overview can give you valuable insights, like the keywords visitors use

Poring over these results can give you several valuable insights that enable you to improve your site, and thus its likelihood of converting visitors, in a few different ways:

  1. First, you can see what your users search for. If a popular item has no dedicated promotional campaign, you can start one based on your observation of visitor interest. The great thing here is that you already have a keyword on which to base that campaign!
  2. Second, by observing what people search for, you can improve your site copy and content. This boils down to providing more information on searched keywords that are currently not readily connected to relevant information or products.
  3. Finally, you can improve your SEO by observing the keywords your visitors use and incorporating them into your related products, so potential customers will also be able to find those products using search engines.

Google Analytics can automatically track site search if you use the standard Google Search query for website search. For other types of site search implementation, it may be necessary to use Google Tag Manager to get keywords reported in Analytics.

Quantitative data tells us a lot, but not why

When you need to know what your visitors are doing on your site, quantitative analytics has all the answers. In fact, the only thing analytics tracking can’t answer is the “why” — and that’s where qualitative data comes in!

Quantitative data tells us a lot, but not why
Quantitative data tells us a lot, but not why

But don’t whip out a survey just yet. The best qualitative research is based on your quantitative research findings.

Analyzing web traffic and tracking the behavior of your visitors can provide you with some pretty good indications of what to ask your visitors when you set out to do your qualitative research. Traffic patterns can point you to content that’s causing issues or problems on your pages that you wouldn’t have otherwise noticed.

It’s important to remember that quantitative research is not just about collecting rows of figures. It’s about making sense of those figures, and using them to find ways to improve the likelihood that your visitors will become customers.

Simply reporting how many visitors came to your website and how much time they spent on which pages won’t cut it. You need to know what visitors do, how they interact with the content, what route they follow as they they move about the site, and so on.

And always keep in mind, again, that correlation does not mean causation. Question every finding and try to confirm it before you embark on improving an aspect of your website that your initial source of data points out as a problem.

Correlation != causation (Image source)
Correlation != causation (Image source)

Ultimately, you’ll use all of this information to accomplish one thing: to improve the way you present your value proposition, products, and brand to your visitors.

Published by

Edin is a Senior CRO Consultant. Edin is into Google Analytics and testing (any A/B testing tool really) and likes to write about it. You can follow Edin on Twitter.