Case CRO for Ecommerce: Advanced Tips from the Experts
In the following article, we talk to national and international CRO professionals to provide us with advanced conversion tips for online shops. The article advises regarding how to analyze the internal searches, how to integrate a development team as well as some laws of design and interaction that will help us to highlight the value we offer and make the discovery, evaluation and acquisition process intuitive. Here we go:
Luis Alendo, CRO y UX Manager at PC Componentes
Think about repeat purchases. We often focus on achieving the conversion in a straightforward way. However, we must not forget the importance of repeat purchases for eCommerce. A good strategy can be to offer a discount applicable in the next purchase. Thus, we are increasing the number of orders and halving the cost of the purchase action. Bear in mind that, on Black Friday we can use the Christmas campaign as a claim to use that discount.
Personalize: If you add to a good segmentation of users the personalization of a campaign’s creativity, you can significantly improve the CTR. A personalized vs generic creativity can increase the CTR by 150%. Do not hesitate to generate a good collection focused on the different segments.
Do not take the user out of the purchase process: Many eCommerce includes a box on their checkout page to enter a promotional code. In the Black Friday campaign, the user is increasingly more used to this type of benefit. The fact of placing this box can cause the user to leave the checkout process looking for a coupon and letting the option of losing the sale. Ideally, it is best to display available coupons, provided that the user can benefit from them in their purchase. Around 10% of shopping cart’s abandonment is due to not finding a discount code. It should be noted that the main reason for abandonment is unexpected shipping costs, in this case the 25%.
Take care of the upselling: Sales during Black Friday are usually limited and the user is more willing to make a purchase than in normal situations. The upselling module located on the product details or on the home page can help to finalise the sale and to increase the average basket. If the offer the user is looking for is out of stock, we can offer a better model on promotion to try to convince him to finalize the purchase. The use of 1 to 1 personalization in these types of modules, can lead to improvements of 86% in comparison with personalization by segments. We should not forget to use this type of personalization in the recovery of abandoned shopping carts (whenever it is within our reach).
It is not all about selling: In this campaign, traffic is usually higher. One of the points that we must not forget is to capture leads. In the case of an offer running out, the option of “we will let you know if it is available again” asking for the email address usually works very well. The user Will be more willing to give the email in these situations. We can always impact them with a remarketing email in case there are no more units available. With a good lead generation strategy, in a campaign like the black campaign, you can achieve a 600% improvement.
Techniques named Fomo: In this campaign, the use of this type of technique works well. To mention that there are few units left or launching offers for a limited time period are strategies that accelerate sales considerably. I suggest the use of these techniques, but in a controlled way. If we abuse these notifications, they lose their essence.
Isaac Gassol, CRO at Mediamarkt España
The aim of my contribution is to tell you how you can take advantage of what we already know about the UX laws to increase the ratio of clicks of product listing or product detail websites.
We start from the principle of proximity which states that elements that are close to each other will be grouped together (Goldstein, 1999). The “law of proximity”, one of the many UX laws, draws on this principle and tells us that the proximity of objects helps the user to better understand and organize the information on the web(Changet al. 2002).
So, if we know that users group elements that are close to each other, we can take advantage of it and put together the most relevant information about your product with CTA:
With this in mind, our hypothesis is that according to the law of proximity scenario 2, you will get a higher CTR from the “add to cart” button.
So, just try it! You can either apply the change based on UX heuristics or just test it!
If CRO is hard for you because you think that you need an analytics team to get insights, to do user testing, to pay for an A/B tool, etc. This is not the case! You can start optimizing your website now. All you need is:
- Take a look at all the knowledge that is out there such as the laws of UX or Robert Cialdini’s tools of persuasion
- IInstall the free version of Google Optimize
- Test, test and test… and get results
- Implement and grow
Goldstein, E. B. (1999): Sensation and Perception. Pacific Grove, Brooks/Cole.
Chang, D.; Dooley, L. and Tuovinen, J. E (2002). Gestalt Theory in Visual Screen Design — A New Look at an old subject. In: ed. Selected Papers from the 7th World Conference on Computers in Education (WCCE’01), Copenhagen, Computers in Education 2001: Australian Topics, Volume 8. Melbourne: Australian Computer Society, pp. 5–12.
Georgi Georgiev, CEO at Analytics Toolkit and author ofStatistical Methods in Online A/B testing
Test everything. The “test everything” approach is something that, if adopted, can greatly improve the path of any digital business, and even more for e-commerce businesses, where measurement is much easier to get right. The idea is that A/B testing should not be limited to potential improvements suggested by your CRO consultant or your product development manager/team. All significant changes to the way the business operates should be tested before being fully adopted.
This also includes what simply “has to be done”. Just because something has to be done, does not mean that there is only one way to do it, or that the implementation is error-free. A/B testing of a change, even when it is unavoidable, can safeguard against potential losses. For example, if the payment details site needs development to include some additional disclaimers or if the fund’s flow needs to be amended to enforce more secure payments, these should be A/B tested. Remember that is the implementation itself that is being tested, as well as the idea that drives the change.
There are three main reasons why companies do not test as much as they should:
- The first one is the lack of instrumentation and infrastructure. Obviously, some investment is required and there is no way to avoid it. The bigger the business, the more profitable will be the investment.
The first one is the lack of instrumentation and infrastructure. Obviously, some investment is required and there is no way to avoid it. The bigger the business, the more profitable will be the investment.
- The second one is the fear of cross-contamination of results because of taking different tests at the same time. Such fear can lead to a severely reduced testing speed or even cause worse validity issues. However,problems caused by simultaneous testing are less common than is generally perceived..
- Finally, many experts may still be under the impression that in order to conduct a valid test they must wait for a fixed schedule, which may be several weeks or even months before running a test. The concern is that this would paralyze innovation. While this could end up being the case, depending on the situation, since many years ago new approaches are available that allow for flexible monitoring and decision making in A/B testing, while keeping the expected rigor, e.g Agile A/B test.
Combine this methodology with intelligent decisions about the level of acceptable risk for each test and you will have a lean but stable commercial operation.
Vicente Martínez, CRO at Telefónica and Professor at K School
“The good carpenter measures twice and cuts once”, is a proverb that refers to the saving of material by not having to throw away the piece, and the time involved in redoing the whole process. Applying this principle to CRO is a basic that we cannot always guarantee, but it is nevertheless highly recommended to do it.
When we land on a project either as project owners or as a consultant, we face the need to show results, and because of that, it seems that we need to start testing as soon as possible. We want, or we are asked, to share the improvements that our hypothesis is producing as soon as possible. This is not the problem itself, and it is not unreasonable to start testing at the beginning of the project, the problem is that sometimes the analysis is not trivial and we need to invest a lot of time in obtaining and unraveling all the information to understand what is happening.
To be data-driven is not just about having everything tracked, this only provides us with data. We must transform this data into information that helps us to understand what is really happening to be able to fix or improve it. Using the necessary qualitative and quantitative analytical tools and having the time to do a deep analysis and prioritization is absolutely necessary to be able to target the most successful tests.
When we launch a test prematurely, we usually encounter the following consequences. On the one hand, we may be focusing on something that does not move the needle at all and that will not make a difference to any of the segments to which it is applied.
A real-life example of a “textbook” tie produced by a test in G. Optimize
In this case, we will get a tie or the need to collect data for weeks and weeks, only to end up with inconclusive data.
Considering that the tests we can run in parallel are limited, this has led us to have a resource blocked for weeks because we have not invested sufficient time and use of the necessary tools to be able to analyze it in detail.
As a conclusion, we can summarise that sometimes by trying to save a few days in the analysis we will lose weeks in which this worthless test is in production blocking the opportunity to test something that will help our users. So, to conclude we will use another Spanish proverb, “dress me slowly that I am in a hurry”. Let’s understand analysis as an investment and try not to get caught up in the rush to run tests that will not move the needle just for the sake of testing something.
Oriol Farré, Data Lead at Bershka, Professor at K School and Author of the CRO Newsletter
One of my favorite pieces of advice to optimize the conversion of eCommerce is the optimization of the CTA copy. This is something that in terms of costs has a very low impact in terms of definition and development time consuming, however, it can make the difference when it comes to measuring the conversion of your website.
A copy you should remove from the buttons of your website is the “continue” button; a button is like a door, and people go through doors that are clearly marked. If you are not clear, many people may get confused and not end up clicking on the button you want because they do not understand it. It should be very clear what the person will find when they click on the button.
One of the most critical points in e-commerce is the button to confirm the purchase. Many e-commerce sites focus on what the user will do (pay) rather than what they will actually do (buy). It is worth testing the copy of this button thinking about the user: people pay a fine and buy an iPhone. What feeling do you want your customer to leave the shop with? A reward or a punishment?
Natzir Turrado, SEO Consultant and CRO freelance
One of the most important parts of eCommerce is the internal browser. More than a tip, I will give a series of tips as this is an area that is largely ignored by many eCommerce.
- Analyze internal searches using SSA techniques. This is one of my favourite analyses to do and with more quick wins out there!
- Allows advanced search (facets and filters)
- Accept misspellings (you solve the 80% with Levenshtein) and synonyms.
- Never show 0 results, offer alternatives
- The suggestion should suggest, not compel.
- There must be a results page to browse and filter.
- Offers suggest, the possibility of browsing for serendipity.
- Think about including voice search.
- Offers also result in non-transactional queries (campaigns, returns, delivery…). Amazon’s search engine does this very well.
- Match whenever possible with a specific landing page or offer all the specific filters to the search.
A good browser for eCommerce “made in Spain” is Doofinder, and one of the eCommerce that best solves the search process is the one in Home Depot (you have to enter with a US proxy to see how it works).
Eduard Mairal, Product design and CRO at Promo Farma
Overcome the intrusive syndrome. The responsibility of a CRO profile is to understand the business in-depth and for this, we constantly need data and impressions from different business areas. It is essential to build and take care of a good relationship with all of them, always based on pedagogy and empathy. When any of these pillars fail, it is easy for the CRO to be perceived as an intrusion that hinders the work of others.
When we talk about pedagogy, we talk about explaining to the teams what is a CRO and the value he brings. In most experiments, we not only draw conclusions that improve the e-commerce transaction rate. There will also emerge insights that will be very useful for one department when making product and business decisions.
Although CRO is a discipline across all business areas, it is not the center and does not live in isolation from the rest. Here is where empathy comes into play. The CRO is born from and for the business. Therefore, when it is time to deploy CRO actions at a strategic level, it is crucial to know and communicate the impact of these actions to the different business areas. On the other hand, we must also estimate the effort required by the teams involved in the implementation of these experimentation actions.
In essence, an organization “steeped” by CRO is able to detect more problems and explain them through hypotheses – the basic ingredient for experimentation.
Dámaso Curto, CEO at Grupo Billingham
I am not going to discover the wheel, but we should talk more about the browser of a website, as it is one of the cornerstones of the conversion in an eCommerce, especially in those with a very wide catalog and type of products. And the fact is that there is still a huge amount of eCommerce that does not take advantage of its potential or that their webmasters do not even realize how relevant their results pages are.
Some of the advice I would give for its improvement would be:
- Make it visible and make the box as big as you can, so you can put a typical query of your business. On mobile, if you want to hide it, make sure you have a relevant search icon.
- Insert a placeholder (copy) in the browser box encouraging the user to search.
- When scrolling, keep it fixed so that it never disappears.
- Help to complete the search when the user is typing the query. In general, users are not experts in what they are looking for.
- Display products or category results while the user completes the search term.
- Work with synonyms so that if someone searches for example for “USB drives”, they can find “pen drives” or vice versa.
- Offer first results of your TOP sales or what interests you most so show because of the business.
- Optimizes the 0 results page, so that the user can find a guide to refine their search, offer similar alternatives or even add help from in the search.
The effectiveness of these or other tips will depend on the type of business, you should measure them with analytical tools, such as Google Analytics, and you will be able to gather very valuable information about the user behavior that will help you to solve questions such as:
- What are users looking for most?
- How many times (on average) do users use the browser?
- How many searches return 0 results?
- Are users looking for products you do not have?
In conclusion, a user with a successful browser multiplies their chances of converting.
Daniel Pinillos, Consultor UX and SEO en Funnel Punk
The interest pyramid and the LIFT model as conversion-enhancing frameworks (CRO)
There are two methods I use when I am helping to redesign a funnel and a transactional landing page:
1. The interest pyramid.
Pirámide de interés
This method consists of listing all the elements that a website will have. We add both design/content elements and SEO (as for example a markup can be part of a rich snippet) and influence the conversion in the first step of the funnel which will be the google result.
Once the elements are listed, we do a survey of at least 10 buyer people and ask them what they value most when buying that product. Then the order is shown in the layout of the page. It seems obvious but often we do not highlight what is important to our buyer.
El método LIFT
Another heuristic method I use to improve CRO is the LIFT framework. It consists in analyzing a landing base on conversion factors:
- What is the value proposition?
If the buyer detects perceived benefits greater than the cost perceived, then their motivation will be high and the impulse to buy will be triggered in their mind. This is the base of CRO.
- Relevance, is it relevant to me?
If you do not show them what they want to see or make them dizzy they will leave, it is as simple as that. Separate the wheat from the chaff and show a clear value proposition, useful and relevant content, and a context so that the landing is related to the channel it comes from.
- Clarity, is the ext action towards purchase clear?
Highlight the CTA well (Contrast, situation, gain), design it so that the next action is clear. The copy is also an interface, serving design and content to lead to the next step.
- Distraction, Is something distracting me from the path?
Reduce possible actions, secondary CTAs, unnecessary options, non-relevant elements…
- Anxiety, does anything make me anxious?
Not seeing the total price until the end, not having any comments, not knowing which company is behind the sale, not knowing when I will receive it… generates uncertainty, anxiety and are brakes to make the purchase.
- Urgency, can I increase the urgency?
The urgency is relative, it can be internal (missing stock, offer expiry) or external (e.g. Christmas is coming and I haven’t bought anything yet).
Generate elements that help to take the next step more quickly with offers/coupons with an expiry date, counters, number of products sold, surplus stock, etc.
The CRO process consists of understanding well how the user thinks and for that we need to do more research than design. The A/B tests are the roof of the house and where we will put our hypotheses but the CRO is something more methodological and before putting the icing on the cake with an A/B test we must understand the base and improve our CRO processes. But do not forget that everything that is not measured cannot be optimized.
Arnau Vendrell, Co-CEO and director atiSocialWeb
One of the points I consider most important when it comes to optimitation is the prioritization of actions and the selection of control variables. When we audit a project, we detect many improvements that can be implemented directly (i.e. we recently found a great consumer that had a “Continue shopping” button in the shopping cart that sent the user to the front page), and improvement proposals that must be tested. The growth of a Project in terms of conversion does not lie in the implementation of all of them, but in the ability to prioritize those that generate the greatest (positive) impact on the business.
Therefore, once you have the whole set of improvements, you weight their order of execution according to the most objective as possible criteria; trying to avoid criteria such as:
- Impacts it will have
If the parameters are subjective, it is possible to get into unproductive discussions. Therefore, it makes more sense to approach prioritization as objectively as possible:
- Does it affect all users?
- Is it displayed only once or does it persist during the same session?
- Is it above-the-fold?
- Is it easy to implement (less than 4h; less than 8h; between 1 and 2 days)?
- Have I corroborated the problem or improvement with any quantitative information?
The product of the values assigned to this system allows us to simply organize by result and voilá! We have the working order.