Every time I hear someone say, “Let’s test if the red button will be better for conversions than green (yellow, blue, etc.)”, my eyes get all bloodshot. I mentally curse whoever it was that wrote the first articles telling you that you can simply test different button colors and get extremely high lifts in conversion.
And yet thousands of webmasters and digital marketers spend their time testing button colors. Some of them succeed in making conversions higher, some – not. The problem is, most of them just test color blindly without thinking about why this should work.
The funny thing is, these articles don’t lie. The changes described in them really resulted in the indicated lifts in conversions. What most authors of such articles don’t tell you is what their assumptions were based on. They rarely lay down the preliminary analysis that led them to find that insight to test.
And all of these webmasters and digital marketers just take the winning result (test red color vs. whatever color) and implement them. The dangerous result is that they are learning the wrong things about their test (“red color buttons work better than green color buttons”).
I call this the “I’m lucky” approach
This approach is still widely used, especially when companies avoid hiring CRO specialists, and try A/B testing themselves. If you’ve ever conducted A/B testing using this approach and have received conversion lifts, think again about the results. Going back to the button color example; it is not that a specific color works better for conversions, it is because it worked better in your specific case.
On the other hand, we have the data-driven approach. This is where you painstakingly analyze all the data you have collected and come up with a hypothesis in which you A/B test. This approach is extremely effective for sites that have tens of thousands of visitors per day.
So, do we only have these two options in CRO: the completely blind “roll-the-dice” vs. the expensive, data-driven approach? Happily, the answer is no; there is one more, which exists between these two. This approach will help you get some big results without big expenses. It is called mental shortcuts, or heuristics.
What are heuristic and how can it be used for conversion rate optimization?
Heuristics is a technique used for problem-solving and learning. It might not seem optimal at first, but you’ll quickly learn it can be the optimal way to achieve goals in a timely manner, without the need to spend hours for pre-test analysis and research.
You might have already heard of some of the most famous heuristics: rule of thumb, an educated guess, an intuitive judgment, or common sense.
Heuristics work because they are mental shortcuts that ease the cognitive load of making a decision.
A Simple Example
Here is an example of the most famous heuristic applied to optimizing conversions on an ecommerce website: There shouldn’t be more than 2 different call to actions on any given page (which are far upstream from the main conversion point), and only 1 on the pages which belong to the conversion “last mile”.
Pages where you can put 2 calls to actions will be the index and catalogue pages. The primary CTA on them will be “More info” or “Add to cart” and perhaps a secondary “Sign-up” (say in the header menu). Of course, you should use another heuristic here: your main CTA should be more prominent to increase the probability of a visitor clicking on it.
The conversion “last mile pages” for your typical ecommerce website will be product, cart and checkout pages. On each of them you should have only one prominent call to action: “Add to cart”, “Buy” (or “Checkout”).
Comparing the heuristic approach vs. the data-driven approach
No matter which approach you are using, you have to start by defining a problem. Here is a simple example of a well-defined problem statement: “The conversion rate of our homepage visitors into account sign-ups is 0.65%. We believe we can do better.” Pretty simple right? 🙂
So let’s start with the data-driven approach:
Data-driven approach process
- Be sure you have your target audience described (in the form of marketing personas), so that when you analyze you can draw better conclusions. My favorite approach to this is the one invented by Alan Cooper. Read about this approach in this article.
- Be sure your Google Analytics (or any other similar tool) is setup in the right way (meaning it collects data and does it correctly). Here’s a list of 29 common Google Analytics data errors, which you should check and fix if they are present.
- Do the same for other analytics services you are using (heatmapping, calltracking, etc.).
- Install live chat if you haven’t done so, as visitors’ questions give the most actionable insights. Here are 5 reasons why you need live chat (pay special attention to page 5) and which live-chat to choose.
- Get some data from surveying your visitors. Here’s a simple and actionable guide for choosing questions which will unlock your customers’ deepest desires.
- Perform a UX analysis of your site. Here’s a list of 7 usability mistakes which kill conversion. Check if you’re violating any of them.
Obviously, more tools and techniques can be used, but the ones discussed above are the most common to gain insights for conversion rate optimization. But most importantly, from these insights you can start to form hypotheses for A/B tests. Once you determine which hypothesis would be the most beneficial to your overall business strategy, prepare an A/B test for that specific one.
The next step is to determine how much traffic you need to run a statistically significant test. This will require you to calculate the necessary sample size beforehand and do a few other mathematical housekeeping tasks. Be sure to read the Crazy Egg article, When Are You Ready For A/B Testing? by Daniel Sims.
Finally, you’ll perform your test for the necessary duration, collect your data and perform a careful analysis to see if your hypothesis was indeed correct.
As you can see, the process of data-driven approach is quite involved (and each of the steps in it is actually a topic for a separate article). Let’s see if heuristics approach is any better.
Heuristics approach process
The heuristics testing approach is much more simple in terms of qualifications and time required. Though, be sure you are using it wisely and don’t slip to the “I’m lucky” approach.
In the heuristics approach you are using mostly qualitative data and there is no need to a dive deep – just conduct a short analysis and use your common sense and understanding of the market and target audience to make a hypothesis to test.
So how do you get insights from qualitative data?
Speak to your sales team and executives
It’s very common that your sales team holds the keys to unlocking your best testing hypotheses. These come in the form of frequent questions which tend to be your user’s biggest pain points. They can be used as a source for super-actionable heuristics-based insights.
For example, if you are an ecommerce business and many prospects are asking your sales team about delivery prices and conditions, that means your site is either lacking this information or it is not easily accessible as visitors expect it to be. This is a great heuristic insight to test: “If we add information about delivery and make it more prominently displayed on pages where users expect it to be (product page, cart page), conversions will grow.”
Make it a habit of speaking to sales and executives teams (and collecting questions from live chat and support) at least once a month. That will help you kill two birds with one stone: reduce the load of repeated questions to your sales team and constantly optimize your site conversions.
Be sure to ask your visitors directly
If you have some ideas of what your visitor’s pain points are, but you’re just not sure about them enough to test them right away, then start surveying. This is a great way to justify your heuristic assumptions and make your CRO efforts more efficient.
So, how do you ask the right questions?
Zoe Uwem gives the following advice for result-focused survey questions:
- KISS (Keep It Simple, Stupid). Your prospects may be very busy, so giving them the simplest and a quick way to access and answer survey questions is the key to getting more quality answers in terms of CRO.
- Avoid asking yes/no questions. If you don’t give an opportunity for visitors to state what’s on their minds (which may be other possible answers that you haven’t thought of) and limit responses to a simple yes/no choice, that’s all you get. A simple tweak here to improve the quality of the response is to add an “Other” option where a visitor can write his variant.
- Use open-ended questions. Open-ended questions require respondents to write a comment, an essay, or other type of free-reaction text. These type of questions gives respondents more freedom to express what they really think.
As you can see, a heuristics approach is simpler in terms of qualifications and the time required, but it still gives you a great opportunity to optimize your conversions.
3 case-studies where using heuristics helped to dramatically increase ecommerce conversion rate
Here are a few case studies that show how using the heuristics approach to conversion optimization resulted in some great wins!
1. 60% increase in conversions for an online lip gloss store using educated guess heuristic
Misslipgloss.nl, a Netherlands based lip gloss store and blog, in 2014 conducted an A/B test, based on an educated guess heuristic: visitors would rather leave the site than go searching for information they need to make a decision to buy.
Before the test, information about shipping cost was missing on the product page, so the company decided to test a variation, where shipping cost was added:
Results of using this heuristic to optimize ecommerce conversion rate were astonishing – a 60% lift was achieved.
2. 69% increase in conversions for online roller skate shop using profiling heuristic
Adam Lean, the founder of RollerSkateNation.com, knows more than anyone about roller skates and his target audience. Among all of the facts he knows about roller skating, there are two which he used to dramatically increase his site’s conversions.
First is that a big part of his buyers are parents, who buy roller skates for their kids. Second – is that if you buy roller skates properly sized for a kid, you’ll have to buy a new set in 6 months.
He also realized, that when selling to parents over the phone, he regularly gave them the advice to buy skates one size larger than needed. He also would recommend them to wear extra socks to compensate for the difference in size. When this sizing conversation would transpire, the probability of a sale increased significantly. He decided to use this information on the site and conduct A/B testing to be sure he was right.
So, he added that advice on the product page and – voila – he got his 69% increase in purchases.
What Adam used when developing this experiment is called profiling heuristic.
3. 60% conversion rate lift for online apparel store using common sense heuristic
In 2015 NewYorkDress.com was trying to get some value from CRO. At the time they had a reasonable amount of daily traffic, nice volume of sales, but also a feeling they can do better. They wanted to increase their conversion rate. Particularly, they thought that making their product page better will help increase “adds to cart”.
While analyzing their product page’s look and feel, the CRO team noticed that the page lacked something crucial in terms of conversion elements: trust elements.
In this case common sense heuristic was used: adding trust elements to the page and restructuring it in a way a buyer usually requires information before making a decision (basic information about product, trust of the product, trust of the store, purchase!).
Not surprisingly, this approach gave very nice results. Firstly, it resulted in a 9.5% ecommerce conversion rate lift during the experiment:
And secondly (and this result is more important), it dramatically changed the trend of conversion rate compared to the previous year.
Here’s the ecommerce conversion rate trend for the period 2nd of March to 17st of May 2014.
The conversion grew until it went down to reach around 18% increase compared to the beginning of the period.
And here’s the trend for 1st of March to 16th of May 2015.
You can see that the conversion increased dramatically here – a 60% conversion lift at the end of the period compared to the beginning. (The marked point on the graph is when the winning variation was implemented on the site and after which there was a fall in conversions last year.)
As you can see, using heuristics can give pretty nice lifts in conversion rates even without spending hours on data analysis.
If you are interested in learning more about the heuristic approach in user interface design, I strongly recommend you read works by Jakob Nielsen (who is the pioneer in this field) and Alan Cooper (known for his books “The Inmates Are Running the Asylum: Why High Tech Products Drive Us Crazy and How to Restore the Sanity” and “About Face”).
Analysis paralysis and “heavy sifting” have a cost. The profits you may see after running a test that takes six months to prepare, implement and review – may not cover the costs of conducting the test. Time and energy are valuable. If you can test quickly, you’re much better off in the long run.
Good luck optimizing your conversions!
About the Author:
Tony Simonovsky has more than 10 years experience in digital marketing, but his story begins on 12 September 2012, when he left his office job in Moscow and completely changed his life. Since then he’s been living in India, Thailand, Sri-Lanka and Brazil, rode 30,000 km around Europe by motorcycle, practiced capoeira and have been working as an independent Google Analytics, Google Tag Manager and conversion rate optimization consultant, helping internet-companies worldwide become data driven.