One of the most common question people ask about conversion optimization (besides – what’s the most high-converting button color….🙄), is what are the must-have tools everyone should be using.
With SO many tools out there, the options are endless and it’s tough to know what tools work and how to use them in the best possible way.
This very same question came up in GetUplift’s Facebook group ‘We Optimize’ a while ago and sparked a really interesting conversation about the most important part of conversion optimization; the research and the tools it requires.
Over the past 4 years, I’ve tested just about every conversion optimization tool I could lay my hands on, any tool that could help me conduct meaningful conversion optimization research, gain more insights about my audience and understand user behavior.
A few of these tools were complete failures,
Some were (hmmm, let’s just say…) nice to have…
And a select few made it into my A-list.
This A-list is the one I use for every research my team and I conduct and it contains the best, most in-depth and reliable tools that are right for my optimization process. For the way, I review websites and gain actionable insights.
This doesn’t necessarily mean these tools would be perfect for you, but they can definitely take your research skills to the next level and help you optimize your websites.
The thing about tools though, is that they are only as good as the process you’re using them with.
It’s not enough to just have the best tools out there, you need to know how to use them, what to look for and most importantly, how to gain the best in-depth insights from them.
Today, I’d like to share this A-list with you along with how I use them, what I look for and how to get the most out of them.
Below is a snapshot of my entire conversion optimization research process, the one I’m going to walk you through.
You’ll notice it’s broken into 2 parts:
- Quantitative Research – focused on collecting data on user behavior, understanding your numbers and being able to truly analyze your data.
- Qualitative Research – focused on understanding the audience better, your competitors and business.
In the first part of this guide we’ll cover quantitative research.
- Google Analytics health check
- Advanced segmentation
- Data extraction [Sampled and Un-sampled Data]
- Behavioral patterns [converters vs non-converters]
- Come up with optimization ideas/ growth tactics
1. Feedback, attitudes & opinions assessment:
- Chat transcripts analysis
- Open-ended surveys
- Customer, Sales, Customer Support interviews
2. UX analysis
- Heuristics Analysis
- User testing
3. Competitor research
- Competitors assessment framework
4. Business persuasive assets
- Awards, Milestones, Endorsements, Social Proof
5. Customer profiles design (Pains, gains, jobs definition)
- Strategyzer: The Value Proposition Canvas
(You’ll get this part of the research next week)
Ok, let’s get cracking:
Quantitative Conversion Optimization Research
Quantitative research helps identify certain behavioral patterns and leaks in your funnel. It refers to hard facts and measurable data you can use to quantify and uncover your visitors’ online behavior patterns.
Quantitative data, like Google Analytics data, helps you understand the big picture of what’s happening in “every room” of your website.
Data shows you the ‘WHAT’ – what’s happening and where.
- What visitors are doing on your site
- Which visitors convert the most or the least
- Where the leaks are in the funnel
And much more.
Our quantitative research is divided into 3 parts:
- Data accuracy check
- Data analysis strategy
- Actionable data insights
However, before you start analyzing, we need to make sure the data you’re collecting is accurate.
Here’s how we do that:
Believe me, I know.
Most people don’t like this part… But how would you feel if you spent dozens of hours extracting and analyzing data, only to discover later that it’s incorrect data?
To me, that would be like a doctor confidently prescribing you medication under the wrong diagnosis. So to avoid that, we’ll need to audit our analytics setup. You can start by using the following checklist, simply make sure each of the points below is setup correctly in your Google Analytics account:
Google Analytics Health Check Checklist
- Check if your Google Analytics account gathers Personally Identifiable Information (PII)
- Enable the remarketing & advertising reporting features
- Set the correct time zone (consider updating the timezone to your customers’ timezone vs. your own)
- Structure the GA property in 3 views: Unfiltered view, Master View, Test View
- Make sure you exclude all staff and agency IP’s
- Integrate Search Console with Google Analytics
- Exclude self-referrals
- Integrate Adwords with Google Analytics data
- Rewrite subdomains URLs
- Set default page correctly
- Add lowercase filter if needed
- Exclude URL query parameters
- Turn bots filtering setting on
- Implement goals and events tracking meaningful to measure
- Setup content grouping if helpful
- Check if the GA tracking code is properly placed and updated
- Make sure that the tracking code is working on all pages
- Track 404 pages
- Check if there any page has duplicated code
- Implement cross-domain tracking if needed
- Track what visitors interact with the site search
- Activate enhanced e-commerce if needed
- Check if you are using UTM tagging properly
- Is there any meaningful external data to import in Google Analytics
- Is it possible to track visitors across devices (cross-device implementation)
- Is the traffic correctly assigned to each channel (channel grouping)
- Do you make use of annotations
- Is the bounce rate adjusted on your blog
- Check if the default session duration works for your website, if not adjusted properly
(Depending on the website type, you may want to review all of the above or just a few).
Some of these need to be checked manually, some can be checked using the following tools:
Tool #1: GA Checker
Other than making sure we’re collecting the right data, it’s even more important to make sure that you’re collecting any data in the first place. Let’s start by making sure that the Google Analytics Tracking Code is present on all pages of your website using a freemium tool called GA Checker.
Step 1: Insert your website URL
What’s great about this tool:
- It’s quick and free to use
- It also scans for Google Tag Manager, AdWords, etc.
- And (my personal favorite), it shows you if you have duplicated code on any of your pages, like this:
GA Checker Alternatives:
- Screaming Frog is really helpful if you’re not using Google Analytics, it can crawl after any type of code snippets on all website pages.
- ObservePoint – if you’re working with really large websites and you need more complex analysis and debugging.
Tool #2: WASP
If you’re currently auditing a Google analytics account that already has some events set up, you might have a hard time figuring out what each event means. Many times these events aren’t documented anywhere or event names aren’t straightforward, this is where ‘WASP’ comes in.
WASP shows you in real time, what events are being triggered by each of your actions on the site. So using WASP’s free chrome extension you can go around clicking links and buttons on your website and see which events get triggered.
- Google Tag Assistant can do the same thing as WASP, but it doesn’t offer the comfort of seeing everything you need at the bottom of your screen, so it takes more clicks to get the same results.
This part is divided into two:
- Advanced segmentation
- Data extraction [sampled vs unsampled data]
Now that you can trust your data, let’s have a look at it.
What we’re about to do:
- Configure our segments
- Extract useful data about those segments
- Identify growth opportunities & insights
- Understand the difference between sampled and unsampled data
Step #1: Data Segmentation
This is the part I love.
During this step, we come up with different types of questions related to the behavior of our website visitors.
Most of your questions will be answered by looking at the basic analytics reports in GA, however the full functionality of Google Analytics doesn’t give us all the information we want.
The real value lies in asking complex questions and these complex questions can only be answered by building advanced segments.
Examples of complex questions:
- What’s the difference in the behavior between new and returning female visitors who reach the dresses product page?
- Do female visitors convert better when they previously read an article from the sports blog category or when they read an article from travelling blog category? Is it the same with men?
- What are the top highest converting filters for dresses category pages vs. the t-shirt category pages
- The web analytics tool you are using [Google Analytics in this case]
- Your almighty brain. I’m 100% with Angie Schottmuller here:
“Your mind is the most easily lost and forgotten as the ultimate tool. CRO isn’t rocket science. Being able to think through the experience to ask and discern “why?” is the greatest facilitator for improvement.” Angie Schottmuller, Growth Marketing Consultant
Step #2: Data Extraction
The way we handle data extraction is mostly based on data sampling.
What you need to know:
Depending on how much data your account has, and the period for which you are extracting data, Google either gives you 100% of data (that’s the case for default reports) OR chooses a sample that it thinks is a reliable representation of the population.
- Population (100% of the data) – includes all of the elements from a set of data
- Sampling (less than 100% of the data) – observations drawn randomly from the ‘Population’ and is usually used to draw inferences about the population under study. It is an inference because there will be some inaccuracy involved in drawing conclusions about the population based upon a sample.
- Standard error – measures the accuracy with which a sample represents a population. In statistics, a sample mean deviates from the actual mean of a population; this deviation is the standard error. Note that the error shouldn’t be significant given that the sampled data is randomly chosen.
Data sampling is affected by the type of analytics account, namely:
- whether it’s paid or free
- and the traffic volume in your account.
Let’s take a look at two examples:
Example #1: Google Analytics (Free)
If you’re viewing a report in Google Analytics that is based on 100 sessions which isn’t a lot of traffic, your report will reflect the data for those 100 sessions.
However, if your report is based on 2 million sessions it will take a lot of processing power to generate a report based on millions of data points. If you’re using a free tool like Google Analytics, it will only use a sample of those 2 million sessions to generate your report.
Problem is, most people aren’t aware of the sampling and simply assume that the report they’re seeing is 100% accurate though it’s actually slightly skewed.
If you’re using a Google Analytics Account and want to know the best way to go about data extraction, ask the following questions:
- Is your sampling rate higher than 50%? If it is, I wouldn’t expect a significant error in tracking. You can extract that sampled data.
- Is your sampling rate less than 50%? (most times it’s lower than 10%). In that case, I’d extract unsampled data (I’ll show you how in the next section).
*** Google Analytics Standard – General thresholds for sampling: 500k sessions at the property level for the date range you are using.
Example #2: Google Analytics 360
A very small percentage of businesses pay $150k a year to use Google Analytics Premium. Why?
Because you get the advantage of having access to 100% of your data (population).
The only exception is when you have over 100M sessions at the view level, for the date range you are using. I guess it takes a lot of processing power to generate a report with that much data.
Let’s review the best tools you can use to extract your data.
Data Extraction Tools
Tool #1 Supermetrics – sampled data
If you’re extracting data for a smaller number of sessions (below the 500k sampling threshold), we always rely on the Supermetrics and Google Sheets duo.
Supermetrics is extremely convenient. You simply launch a sidebar in any sheet, make a few selections and you have the data.
In supermetrics you can:
- Choose the account you would like to extract data from
- Extracted data directly into an excel sheet
- Select dimensions by telling Supermetrics where to place them, on rows or on columns.
- Segment anyway you like and as much as you like.
- Apply any filters to apply, such as “Device category – does not include – Tablet”
What’s great about Supermetrics:
- Ease of use
- Speed & flexibility
- Free version available
The paid version also allows you to extract unsampled data. However, it doesn’t always work.
Tool #2 Analytics Canvas – unsampled data
Analytics Canvas saves the day when the sampling rate is too high and you can’t settle for “pretty accurate” data. We’ve tested a bunch of tools that extract unsampled data from Google Analytics, but Analytics Canvas is our favorite so far.
This 2-minute tutorial created by the guys from Analytics Canvas will clarify the steps you have to go through.
The one thing missing from this video is how to work with sample settings, so here’s how to do that:
- Click “time period”
- Configure and tell “Canvas” to split your requests in multiple API calls (choose a small enough period so that at no point will there be more than 500k sessions in that timeframe). For us that meant telling Canvas that no partition should have more than 15 days.
The advantages of Analytics Canvas
- Extremely responsive, kind and patient support team
- It’s a complex tool that goes beyond data extraction, as you can connect it with other sources and automate reports
Analytics Canvas Alternatives:
- “Unsampler” is another tool that promises unsampled data but it has one major disadvantage: For now, you can only extract data for one segment at a time and that’s a deal breaker when you have to extract hundreds of them.
Now it’s time to analyze and interpret our data:
“Ultimately, data collection boils down to a simple thing: gathering meaningful data. What meaningful means is something that must be negotiated uniquely for each business case, each project, each product, each organization, and each platform.” Simo Ahava
This part is about asking the right questions:
- How are visitors behaving on our site?
- How are converters (meaning customers/signups) behaving differently than non-converters?
- And how can we increase our conversion rate by encouraging behaviours that are more likely to lead to a conversion?
Step #1: Studying Behavioral Patterns In Your Data
The most important part for identifying behavioral patterns in your data, is asking the right questions.
The goal of these questions is to help you form connections between the different patterns, build new segments and keep asking yourself more in-depth questions until you start seeing real insights.
Start by asking a general question like:
- What’s the difference in behavior between a mobile and desktop visitor on our product pages ?
Then you continue with more in-depth questions, for example:
In the example above we identified that:
- 76% of mobile visitors are interested in “buying immediately”
- And that mobile visitors who click on “Buy Immediately” button convert 6 times better than those who click on “Add to basket” button. (1.27% vs 0.23%)
So our next questions was: What’s the difference in interaction with the “add to basket” button on mobile vs. desktop?
- Visitors who save a product during their session have a 6X higher conversion rate compared to those who don’t save a product.
This was actually expected behavior, however, we also discovered that:
- The conversion rate of visitors who save a product on desktop is more than double compared to those that save a product on mobile.
Now it’s time to figure out why this is happening and come up with AB testing ideas.
Step #2: Coming up with AB testing ideas
I like to put my insights in a table (as seen below) and my hypotheses besides them.
This allows me to later go back, understand why I came up with this testing idea and what stood behind it.
|- 76% of mobile visitors are interested in “buying immediately” |
- mobile visitors who click on “Buy Immediately” button convert 6 times better than those who click on “Add to basket” button. (1.27% vs 0.23%)
|- Given that mobile shoppers have an obvious interest in “Buying Immediately” [just one product per order], we should test emphasizing this particular purchase button instead of “Add to basket”
- The same hypothesis applies for desktop, only here we’d test highlighting the “Add to basket” button.
|- The conversion rate of visitors who save a product on desktop is more than double compared to those that save a product on mobile. (14.59% vs 6.34%)||Saving a product in wishlist clearly increases the chance that the visitor will convert, on both mobile and desktop, but I’d start with desktop:
- Incentivize visitors who visited at least one product page, to save the products they are interested in.
- Incentivize those who saved products but didn’t purchase them via email marketing and app notifications.
Everything we’ve mapped out so far belongs into the quantitative research part.
The insights you find in this process alone can lead to significant improvements on your site, but there is so much more you can do with qualitative research.
As Talia explains:
“Your customers are more than just a dot on a map, a device or a browser. Behind those screens are people, people with emotional-drivers, fears, hesitations, and concerns. The only way to truly optimize and 10X your conversions is by understanding their decision-making process, understanding who they are, what they need and how we can show those on the page”
Next time, we’ll discuss exactly how to do that.
Powered by Facebook Comments