Sector Statistics: All Overall Client Data

It’s great to take an initial snapshot of conversion rates through the site, as they are at the moment, in order to have a base level to work from. Hopefully you will have goals set up in your Google Analytics account and/or ecommerce tracking. If not, that should be your first step and then allow some time to generate genuine data (remembering to run the data across any moments of season flux).

In addition to this and as a guide, please find our clients’ conversion rate statistics as shown below (as a top level) and then broken down by sector. Remembering the statistics are averages only, and should be viewed as such. We say this due to the fact conversion rates can be affected by your position in the market place i.e. you may be a volume-based fashion retailer and so a comparison against a premium retailer wouldn’t be valid. All the statistics represent data taken from all our clients over a 12 month period (1 March 2015 to 1 March 2016).

5 Steps to an Effective CRO Campaign

As part of your CRO process, there is a specific process that needs to be followed.

1. Know Your Audience, Market and Goals
2. Research and Insights
3. Wireframing
4. Test and Report
5. Renew Test Cycle

(1) Know Your Audience, Market and Goals

Audience

Knowing your audience seems obvious. However, defining your audience subsets in different ways to begin with will really help to organise and prioritise your CRO activities. This analysis will also help you to identify the messages and value propositions of real influence to your target audience – helping you to sort through those messages with impact vs those your audience would expect from you as a minimum service level:

• Risk. Defining your audience types against their doubts helps to clarify challenges they may encounter. If you are in an emerging sector, there could be doubts in relation to the market as a whole. If you operate within a competitive and saturated sector, those doubts will relate to your business in particular.
• Emotional triggers and trust. What will get each of your audience subsets “engaged” and attracted to your product and brand? Trust triggers can differ per demographic and so it is important to define whether they are the same for everyone or whether they should be refined.
• Knowledge. If your audience is educated and if they know your product really well, then generic industry messages will provide low engagement levels. If you have a split between the two, then you need to determine the needs of each group.

Market

Knowing your market really comes down to a competitor review, which can include traditional, bricks and mortar competitors and online peers. No business operates in a vacuum, and depending on the level of risk associated with the purchase, the competitor review (completed by your target audience) may be substantial.

Please remember that customer “risk” doesn’t always have to connect to cost. Customers may suffer doubts about your service or product if there is a hassle-factor risk (for example if they have to schedule meetings or be available to collect things). Service vs expectation is another risk – this comes down to trust i.e. do I trust this service provider to meet my expectations? Information deficit is a further risk i.e. do I have enough information in order to make an informed decision?

To review your online competitors, check your equivalent landing pages, check the messages present on their pages. What is emphasised on the page visually? What are their price points? Do they have testimonials? How do they present the products?

If your competitor is an ecommerce business, completing a dummy transaction through their site will also provide insights. Further quick cheats include reviewing your competitors meta description (shown on the search engine results) as this gives an indication of unique aspects of their service and the messages that they want to highlight in brief. Also reviewing the ad copy your competitors use in their PPC ads will help to identify the messages they want to convey to their customers.

Goal Mapping and Conversion Funnels

Technical it is easy to set up goal tracking in Google Analytics, but we always advocate goal mapping ahead of goal setup and tracking i.e. identifying what you would like to track first. Your primary goals will always form the initial starting point for goal tracking, those goals will usually either be a transaction (in the case of an ecommerce site) or form fill for lead generation sites. However, there may also be secondary goals which influence (and may lead to the completion) of the primary goals. For example if a visitor completes a calculator on site to get a quote, this activity may also be important to track as you then have the means to assess whether the completion of the calculator leads to a transaction at a later date.

Agreed naming conversions for goals is also important in this process, this avoids confusion going forwards and helps you to manage your goals. You can also set up conversion funnels in Google Analytics – this means you can map the customer journey through a predefined landing page path – this becomes a funnel through which you can view customer drop out.

You can set up more than one funnel, so an effective example would be funnels tied to different product categories i.e. a funnel for shoes, a funnel for dresses, etc – then you can chart the drop off rates for each stage of that particular landing page journey.

(2) Research and Insights

The findings and insights gained in the first phase of your review creates a framework for research and insights. In this next section you are actively gathering information in order to validate and verify your initial thoughts and develop any ideas in relation to tests you may perform on the site as a part of the campaign. Conversion rate optimisation research techniques can be broken down into the following activities:

• Behavioural data taken from Google Analytics
• Walk-throughs
• Usability tests
• Heat map analysis
• Cross device usability reviews

Google Analytics

This is a detailed delve into analytics, taking data which will illuminate customer behaviour on your site. The following forms a list of key metrics and areas to review for an ecommerce site. Depending upon Google Analytics set-up and large site changes- we usually try to take the data as far back as possible to capture trends and changes over time, and compare the information by month and by year.

• Traffic by channel i.e. paid, organic, direct and referral
• Overall conversion rate by month
• Conversion rate by channel
• Overall AOV by month
• AOV by channel
• New vs returning visitor by month
• New vs returning conversion rate
• Top landing pages by traffic and bounce rate
• Traffic by device
• Conversion rate by device
• Traffic and conversion by browser
• Page and site load speed

Walk-Throughs

A walk-through simply means to start on your home page and complete your primary goal on site - completing the conversion journey as your customer would. This seems like an absolute no-brainer, however it is a surprising effective and under-utilised way to identify immediate usability issues or technical errors which may be effecting conversion rates.

Eliminating those issues creates immediate quick wins and uplift for the campaign and conversion rates. When undertaking a walk-through it is important to put yourself in the place of your customer and think about problems they may encounter – and look at your site from a different perspective rather than taking certain aspects for granted.

A walk-through also forces you to view the journey as a progressive path, rather than viewing isolated landing pages. This isolated page analysis is usually the default way in which you will view the site – whereas a customer will view the site as a natural progression of messages and actions.

Usability Tests and Interpretation

In a usability test you ask a member of your target audience to complete actions and tasks on your site, and then analyse a recording of their actions. This is a great way in which to assess whether aspects of the site appeal to your audience or present challenges.

The important point to note when you are creating a usability questionnaire (which will form the basis for the list of tasks you will ask the participant to perform) is to ensure you don’t “lead” the participant too much – i.e. it shouldn’t be a prescriptive list of actions or directions. Ideally you want to gain actual insights into natural behaviour, so don’t make it too easy for the participant – so “find a red dress in a size 10, then change the size to 12 and add to basket” is a better task than “Go to the dress category, select red on the filter options, on the product page select a size 10 etc” as the first task is actual need.

Results of usability tests can be very surprising – and interpreting the results can be fraught with pitfalls, as onsite problems can be overt and obvious in usability tests, and at the same time some challenges can be less explicit i.e. a participant may say they can find something easily, but they are actually on the wrong landing page / you’ve witnessed them getting lost in the site before this point. Knowing what you expect the participant to do aids this process.

Heat Map, Page Scroll and Click Map Analysis

Heat maps are visual overlays which give you information in relation to the behaviour of your visitors on page in the form of a “hotspot” area. Heat maps use mouse hover and click to inform those hotspot areas on the page – as mouse hover is shown to correlate with eye tracking. Please see one we prepared earlier on our homepage, as shown below.

Heat maps provide a real insight into engagement areas on your landing pages and help you to identify areas lacking engagement too. For example if you really want your visitors to view a message; a heat map can tell whether that’s happening or whether it’s being missed.

Page scroll maps work in a similar way to heat maps but instead of showing you engagement areas on the page they show you how far someone has scrolled down the page – again that’s important from a “message view” perspective.

Click map analysis (sometimes also known as a confetti map) shows you where a visitor has actually clicked on page. You can also usually filter the clicks by different choices such as new vs returning visitors and traffic source i.e. PPC, organic, direct or referral.

Cross Device Analysis and Usability

Usability helps to improve conversion rates, but it differs from CRO techniques. A good way to grasp the distinction: usability problems can be overcome by customers and a customer can “live with” usability issues, though they find them annoying.

A good analogy is to think about a stack of shelves in a physical store and the product you want is right at the back, to get it you are going to have to move other products, maybe stretch, it’s annoying, but if you want the product enough and have enough time, then you will do that to get the product you want.

In the same way we can look at a site and assess any usability issues which hinder conversions. Problems with usability hit your conversion rates hardest when the customer has little time, is doing something else, is easily distracted or if it’s an impulse buy. For those reasons, a cross device analysis (mobile, desktop and tablet) for potential usability issues can be highly effective. In addition to this the way in which a customer accesses data on mobile is different and mobile sites more often than not just collapse the site rather than reviewing the needs of the customer from a usability perspective.

Killer Conversion Rate Quick Wins

Usability

Making actions easier for your customers on different devices represents immediate improvements in engagement and helps those customers who’ve already exhibited intent

Cart or Form Fill Updates

Focusing on your cart or form fill process can be a revelation. Customers reaching your cart area or form fill page have exhibited intent to convert through their journey through the site and in reaching those pages can be viewed as “warm leads” – so in making your cart area easy / rationalising data fields / improving the formats of your form – you can capitalise on some quick wins.

Updates on Mobile

Everyone is aware of mobile as a huge growth area for internet access. However, this device can be forgotten and the way in which mobile is being used by customers is also changing. Historically mobile was viewed as an action and execution device – i.e. the customer has already made their decision (before accessing the mobile site) so make the mobile experience short and to the point. However, now we can see that browsing, decision making and engagement time is changing on mobile and so reviewing the mobile journey can also aid conversion rates.

(3) Wireframing

Now it’s time to translate your confirmed CRO test ideas into wireframes. Wireframes are visual mock ups which represent the design of the page you’d like to change. Wireframing is really useful when you want to complete a full page transformation (more about that later) as they give you the ability to provide a visual representation of page structure, discuss and circulate those changes ahead of new page design.

The wireframe below represents a wireframe of a landing page for reference. They are usually accompanied by a rationale explaining the thinking behind the changes and the reference to the benchmark industry practice, which influences the alterations.

Not every CRO test requires a wireframe, and successful updates and alterations can be much, much smaller – the effect on conversion rate really comes down to the impact the change has upon the customer’s journey and brand perception.

(4) Test and Report

4 Easy CRO Tests

Calls to action (CTAs) and USPs. A very easy (but effective) CRO test is to try out different call to action (CTA) messages and USPs. As you’ve now completed your competitor review and assessed factors influencing customer behaviour you now have a strong basis from which to judge what makes you unique / what will matter to your audience. So trying out different USPs should form an initial test.

CTAs sit within clickable buttons and can be as simple as “click here” – not to detract from the power of “click here” as a CTA (as a button without it can really benefit from this addition depending on the demographic) – but usually qualifying the actions and adding a sense of urgency can help to approve submission i.e. “click here today to find out how we can improve your conversion rates by 10%”.

Position on page. Ideally your USPs should be easy to identify too. If they are lost in paragraphs of content, then highlighting them on page helps to reinforce them. You can use bullet points as a design device, images, etc to highlight the USPs. Also where you actually position your USPs on page affects emphasis; above the fold – which means above the screen edge, this differs dependent on screen resolution – will be the first area viewable to the visitor and so is a high value area on the page.

Colours and visual emphasis. Your landing pages operate on two levels: a text-based narrative and a visual narrative. What is often forgotten is the visual narrative and this effectively means the visual weighting and prominence you give to different messages on the page. A visitor infers the importance of the message via the visual message. Reviewing your pages from this perspective is telling and can lead to updates

Logos. Recognisable logos from known brands act as a visual endorsement and a trust trigger for your service and product. Simply including logos on a page can increase conversions. The power of a recognisable logo is amplified if the logo is relevant to the page in question and/or relevant to service or product you provide i.e. using a PayPal logo in your cart area or using “as seen in Vogue” on a fashion site.

Likewise, logos don’t work if they’re not recognisable, relevant and if they are too small to view properly. As a quirk of logo use they are also most effective when you use them in their standard brand colours. This is due to the fact they can be recognised quickly, so altering the logo and using it in mono on a site lessens impact.

Test Jargon Buster

Now you’ve identified your test ideas, completed your research and created wireframes you now have some options for the actual test process. Here’s some jargon terms and the test construct terms to help you. Most test platforms allow you to choose from the following test constructs:

Control

Control relates to your current page which remains the same throughout the test process, so you are able to capture behaviour on your site as if nothing had altered. This is great as it means that anything out of the ordinary will also affect your control page and so doesn’t skew the results of your test.

Variant

A variant page is the new page you’ve developed through your research and wireframes – this page competes with your control page through test. A test may have more than one variant page, depending on the test construct.

A/B Test or Split Test

This is the basic test construct. You simply split the traffic (coming to that landing page) 50/50 between the control and variant pages. This test is effective if you want to do some simple tests and the landing page in test doesn’t receive a great deal of traffic. When you complicate and segment tests you dilute the traffic to that landing page – which means it takes longer to hit statistically valid results (and find a winning variant).

Multi-Variant Test (MVT)

MVT is the most popular form of testing. In a multi-variant test you test two or more different variant pages against the control. So you are effectively splitting the traffic into thirds. Again you need the traffic to that page to be high in order to gain statistically valid results, but it means you can test different versions of the page and gain further insights.

Page Transformation or Redirect Tests

Most test platforms allow you to input a completely new URL into test in order to make significant changes to the page, such as changing the templated layout. Once you’ve created the new URL you can then complete a split test or MVT (depending on the traffic to the landing page and the objective of your test).

Funnel Tests or Conversion Path Tests

A funnel test centres upon a conversion route through the site. You are effectively testing one conversion path against another. If a customer selects to visit a certain page, you are effectively “funnelling” that visitor through your variant test path vs the control path to check engagement levels and overall conversion rates.

Tips for Test Setup

Prior to launching your test you will also have to address test setup. Test setup relates to the metrics you will use to judge whether the control or variant has been successful i.e. what will you report upon? Primary goals are obvious – but remember that you want to know if a sale has been successful and that means tracking by the final “thank you” URL at the end of the process – rather than add to cart (though you can track a combination of the two within test.)

We also incorporate secondary set-up tracking – so those actions and pages on site identified as influencing conversion rates in the research and analysis phase will also feature in the test tracking. In addition to this we will also review universal metrics – such as bounce rates i.e. is the bounce rate higher on your control or variant page? Time on site and pages visited in order to address engagement.

Call tracking can also be incorporated into test too, through the use of vanity telephone numbers you can check to see which version of the page (the control vs the variant) led to the greatest number of phone calls.

Tips for Segmentation

Most test platform providers allow you to segment the traffic and divide it further – this is useful if you want to target new users or test visitors to the site accessing the site via a certain device. Again you are dividing and diluting the traffic through segmentation, so it is important to view this in relation to the overall objective of the test. Testing on one demographic and device also further complicates the test cycle i.e. if your test works on mobile should you then roll out the winning variant on desktop and tablet? The answer: no you should roll out your test to include those devices and compare and contrast results prior to full site application.

Segmentation Options

• New vs returning visitors (personalisation of messages and design elements)
• Geolocation
• Weather
• Browser
• Device
• IP address
• DMP data (Data Management Portals)
• Previous page
• Number of visits to the site / page

Top Tip

Remember to exclude your IP address and the IP address of your colleagues and agents from the test (home and office IPs) as their actions could skew your overall results.

Common Errors in CRO Tests

Technical updates. Updates to the control page when the test is active skews the results as it changes the base level of the test. Ideally you should delay any changes and technical updates until the test is complete

Lack of hypothesis. When you don’t know why you are testing a certain item, it makes it very difficult to determine whether the test has worked in the way in which you would expect it to do – and to know whether the test has been a success. Creating a short, general hypothesis in relation to the test helps to mitigate this problem i.e. “Research has shown the “X” is a problem for our customers, we believe the variant page will result in a greater number of visits to this landing page and increase our overall conversion rate.”

Inflexible approach to test. It’s great to have a test schedule and plan, however if tests lead you to believe that you’ve struck upon a highly lucrative change on the site, I would pursue that concept to take advantage of potential quick wins.

Testing lots of pages at once. Visitors to your site experience your site as a whole – unless you have actioned a structured “funnel” test then testing many pages and elements at once skews results as it very difficult to manage the impact of multiple changes in the process. Also accurate identification of “what’s worked” is difficult in this scenario.

(5) Renew Test Cycle

Testing should be an ongoing process and so after the first test you should complete the following actions and then schedule the new tests:

Review your hypothesis. Did the test work in the way you’d expected and hoped? If so capture what has worked in a brief document in relation to that landing page or conversion path (if it’s a funnel test). In doing this you will create a knowledge library per landing page – which will help to formulate new tests

Review your heat maps. When a test is complete remember to also review the heat maps attached to the control and variant pages – as this will give you insights into how behaviour has changed.

Capture new test ideas. A test will shed light on new ideas for test, if you are unable to action those tests at this time, create a test idea bank for future tests, you can also reinstate the research phase of a CRO campaign, if you are unsure / want to validate new ideas prior to test.