There is a famous quote from Chinese philosopher Laozi: “A journey of a thousand miles begins with a single step”. 

But what if your first few steps take you in the wrong direction? Well, in that case, you may end up thousands of miles away from your desired destination. So the right beginning is as important as the journey itself.

In the case of conversion optimization, we take the first few right steps by following a process (which acts as a roadmap). The process you follow must be well defined and it must have a clear start and a clear end. Without a well-defined process in place, you will end up doing what I call ‘random optimization’.

Let us first start with industry experts and learn, how do they start the conversion optimization process.

Q. How do you start conversion optimization? What processes and framework do you follow?

Tim Ash (author of the bestselling book Landing Page Optimization, CEO of SiteTuners and & Chair of Conversion Conference):

” You have to start with a deep understanding of your audience. As online marketers we are often focused on the internal needs and goals of our business. But from the outside-in, the perspective is a lot more messy and complicated.

You have to understand the confusion, lack of knowledge and irrational nature of how your audience behaves. From this starting point you can design appropriate web experiences and properly motivate them to act.”

Stephen Pavlovich is the CEO at Conversion.com

At Conversion.com we take a strategic approach to conversion optimization. First, we need to understand the goals we’re looking to achieve and the KPIs that track them. Second, we need to gather and analyse data and insight – specifically, we’re looking to identify the motivations, abilities and triggers that drive them (using BJ Fogg’s behavior model).

Third, we develop the strategy – and this will depend on the capacity for testing. eg for a company with high resources and a high traffic website, we’ll use a more exploratory approach to testing, whereas with a low traffic website, we’ll need to be more selective in the tests we run. Finally, we run and analyse the tests.

Csaba Zajdó CEO at OptiMonk

We use a variation of the so-called “Bullseye Framework” for our optimization campaigns.

First, we spend some time checking and reviewing our data and our metrics, then we brainstorm ideas about potential optimization targets.

Q How do you decide what to test first and when?

Tim Ash: 

Prioritizing testing depends on the importance of the pages being optimized (the traffic sources, number of conversions, and the economic value), the availability of enough steady traffic, the difficulty of creating and implementing the test, as well as political considerations and the support needed inside of the company.

Stephen Pavlovich: 

Good question! A simple approach is to rank tests by impact and ease, meaning you can run the high-impact high-ease tests first. The problem with this, of course, that you don’t know which tests are going to be high impact – and certainly not at the start of a project. Instead, we look to see which customer objections are the most prominent – at its simplest, what is stopping visitors from converting? Then we look to see the most impactful way of fixing that objection. This is normally the most impactful place to start – but the right creative may take a few iterations to get right.

Csaba Zajdó: 

We rank these ideas according to their “ROI-potential”: we rate the potential win on a scale of 1 to 10, then we make a rough estimation about the execution costs (again a scale of 1 to 10, 10 being the easiest, less than one hour type of tasks, 1 being the most difficult, several month projects). We add these two values, and rank the ideas accordingly. We choose the top 3, with the highest estimated ROI-potential.

Q How do you validate your A/B test results?

Tim Ash: 

The test is the validation. In other words, if you run it properly you should be very sure that you have found something that performs better.

After that it is simply a matter of continuing to run with the winning version that you have uncovered. Some companies also keep the original version running continuously and show it to a small percentage of the visitors. This is a sort of insurance policy to ensure that the winning version continues to outperform the original over time.

Stephen Pavlovich:

At Conversion, we use a proprietary model for statistical analysis. You can, of course, use a system like Optimizely’s Stats Engine, or an approach that uses a combination of statistical significance, test duration and number of conversions.

Csaba Zajdó:  

We use Optimizely and Google Analytics for our tests, depending on the type of test we want to accomplish. We evaluate the results every two weeks for the smaller tests, or every second day for the larger ones.

Now some tips from yours truly:

#1 Avoid random optimization

Random optimization occurs when you optimize a website without any clear objective. You identify problems (based on some industry best practices) and then you rush to fix them in a hope that it will somehow improve the business bottomline.

Random optimization also occurs when every second or third day you ask yourself this question “What should I do next?”. For example, starting your analysis by looking at the ‘All Pages’ report or ‘Landing Pages’ report is random optimization.

Let us wait and see what will happen if we somehow reduce the bounce rate of top landing pages.

If a web page has got a bounce rate higher than the website average than surely it must be repulsive to users and need fixing.

Bounce rate is a tricky metric. It can suggest many things:

1. Your web page does not satisfy users’ query hence people bounce from the landing page.
2. Your web page fully satisfies users’ query and there is no reason to browse the website any further.
3. You are not getting the right users to your landing pages. There is nothing wrong with your landing pages, the traffic that is coming is not relevant.

So is high bounce rate good or bad? There is no single right or wrong answer. It depends upon how you interpret the data.

But no matter what bounce rate suggests, there is no direct correlation between bounce rate and sales i.e. increasing or decreasing the bounce rate does not directly result in a corresponding increase or decrease in sales.

So there is no guarantee that optimizing the bounce rate is going to improve the business bottomline. The chances of improvement are as good as flipping a coin and expecting ‘head’.

Much of the conversion optimization that exists today and that is taught is old school, is about evaluating page designs (via series of A/B Tests, multivariate tests, heatmaps, etc) to improve business bottomline.

I would have loved to tell you that the A/B test is the miracle cure to all of your conversion problems. But the reality is that it is not. You need to do a lot more than evaluate page design to improve the business bottomline.

I have nothing against A/B test or conversion rate as such. But I am not obsessed with them either. For me, as an analyst, A/B test is just another tool. It has its own place and is helpful in certain situations.

But if you are starting your optimization journey by evaluating page designs and running series of A/B test then you won’t get optimal results and may even end up wasting your time and resources.

The websites I handle are usually high traffic websites (millions of monthly sessions) and they have got tens of thousands of web pages.

There are hundreds of web pages that get thousand of sessions each and near-identical traffic volume. So if I start my optimization process by optimizing the performance of top pages (in terms of traffic), I will forever be optimizing them.

Now I am not saying that you can never get results through random optimization. You can, especially if you are optimizing a low traffic website with clear winners (clear top 10 pages).

But…

• Will you get results?
• When you will see your results?
• What kind and magnitude of results you will see?
• How long it will take to get results?

The answers to all of these questions will remain as random as the ‘random optimization’ itself if you choose not to follow a formalized process.

#2 Determine what the business is prioritizing

You get the answer to this question through the people who actually run the business and not from Google Analytics reports or the website itself.

Consequently, when you are starting out, you need to interview your client.

Ask questions, tons of questions related to:

• business objectives
• current marketing activities
• Pain points (like satisfaction with the current website, marketing campaigns, etc)
• Products
• Target Audience
• Competitors
• SWOT (Strength, Weaknesses, Opportunities, Threats)

Document all the important information provided to you.

Documentation is very important. If you don’t document the information, you will most likely forget half of the key information sooner or later, especially if you are handling many projects.

You can ask questions through Skype, phone, email, or one to one meetings. The important thing here is that you ask questions and not just once but throughout the duration of the project. I can guarantee that you will learn much more and much faster by asking questions than trying to figure out everything on your own via Google Analytics.

It is only by asking questions that you can truly embrace Agile Analytics methodologies.

WHAT’S INSIDE: My Step-By-Step Blueprint For Generating Record Sales And ROI By Leveraging Analytics Data.

Learn To Generate Record Sales And ROI By Leveraging Analytics Data. (150 pages)

Yes I want the ebook

#3 Determine what is being prioritized on the website and via campaigns

We often start our optimization process by visiting the client’s website under the assumption that the website accurately reflects business needs and wants or what the business is prioritizing.

But this not always the case. For example, if a business wants to grow blog subscribers but the blog link is not even in the top navigation menu of their website, it tells you of a gap between what the business is prioritizing and what is being prioritized on the website.

Similarly, if the business is keen to improve website sales but the majority of marketing campaigns are traffic driven instead of conversion-driven, it tells you of a gap between what the business is prioritizing and what is being prioritized via campaigns.

In order to do such a GAP analysis, you first need to know what the business is prioritizing.

If you start your analysis by visiting the website straight away, you are not able to leverage the benefits of GAP analysis (which to be honest is a secret weapon of the analytics pros).

#4 Determine what the target market is prioritizing

We often do market research under the assumption that what business is prioritizing, is being prioritized by their target audience. What the business is selling is exactly what the target market wants.

But this is not always the case. For example, a business may be keen to sell Product X, but its target market may not be interested in buying it. In such situations when a large amount of money is spent on pushing the sales of Product X, it results in high cost per acquisition.

If there is no alignment between what the business wants and what the market needs, then there will be little to no sales.

Whenever there is such conflict of interest, you should always prioritize the needs of the target market.

What that means is, you sell what is selling, what is in demand, and not waste your time and resources to try to sell something which has no demand or try to create a market where the market doesn’t exist.

Of course, if you have got only one product/service to offer then you have got no choice. But for the vast majority of online retailers, this is not the case. You have got the opportunity to focus only on selling the top revenue-generating products.

Improving the sales of all of the products should never be your top priority.

You can do target market research through Google Analytics, Omniture, surveys, feedback, etc.

#5 Do GAP Analysis

GAP Analysis is carried out to find gaps between what the business is prioritizing, what is being prioritized on the website, and what the customers are prioritizing.

The output of GAP analysis is what we call “conversion issues“.

For example, if your customers want to know the shipping cost upfront to make an informed buying decision and the shipping cost is not disclosed on the website until the checkout then this is the gap you need to identify and close to improve the business bottomline.

Similarly, if you advertised throughout the UK but the majority of buyers come only from London, then there is a gap between where you are spending your money and where the money should actually be spent. You need to identify and close such gaps.

Following is the process I follow to do GAP Analysis in Google Analytics:

#1 Find top-selling locations – drill down to city level
#2 Find top-selling product categories
#3 Find top-selling products
#4 Find top traffic sources
#5 Find the top landing pages for conversion funnel analysis
#6 Find the top-performing keywords (optional)

I have explained all of these data drill-downs in great detail in the article: 6 data drill downs for improving Ecommerce Products Sales. So instead of just repeating everything all over again, I would suggest reading this article.

Another article you will find useful: Using Cohort Analysis & Enhanced ecommerce to understand users behavior

Understand the Anatomy of Conversion Optimization

Has conversion optimisation made you a millionaire?……No

In order to understand conversion optimization, what it really is, what it can or cannot do for your business, and, most importantly, where it fits in the ‘analytics world’, you first need to look at the big picture, i.e., business analytics.

Introduction to Business Analytics

Business Analytics (BA) is a practice of repeated and systematic exploration of business data.

However, there is no standard definition for BA. 

Depending upon who you ask (vendor or consultant), different people may come up with different or even better definitions of BA.

I have outlined a very basic definition of BA. 

Business analytics is carried out to optimize the overall business performance from operational and strategic to predictive.

Business analytics (BA) and Business intelligence (BI) are not the same thing, though they are often used interchangeably.

For me, BA is more of an umbrella term that includes data engineering, data warehousing, data mining, business intelligence, predictive analytics, etc.

Following are the various stages of business analytics:

1. Exploratory analytics
2. Data engineering
3. Descriptive analytics
4. Diagnostic analytics
5. Predictive analytics
6. Prescriptive analytics

#1 Exploratory analytics

Exploratory analytics is carried out to determine what data needs to be collected, measured, and monitored.

Data scientists/analysts work on the data collection requirements.

#2 Data engineering

Data engineering is carried out to build and maintain data management systems/architectures for:

• Collecting and monitoring data
• Bringing all of the business data together (data warehousing)
• Visualizing data to understand large data sets (data visualization)
• maintaining data quality.

Data engineers are responsible for data’s overall availability, usability, and security (data governance).

Data engineers do all the ‘building’ (software engineering) work. 

So, if you need an algorithm, predictive model, or prototype for data mining and modelling, they can build it for you. 

However, data engineers are usually not responsible for determining data collection requirements or performing data analysis. That responsibility falls to data scientists/analysts.

Data engineers work closely with data scientists/BA people to understand their requirements and build systems/architectures which meet their requirements.

#3 Descriptive analytics (business intelligence)

Descriptive analytics is carried out to find hindsight, i.e.

1. What happened
2. When it happened
3. How much/many times it happened
4. Where it happened

Under descriptive analytics, a large set of data can be visualized (through data visualization tools) to understand it better and to identify patterns in data sets (data mining)

#4 Diagnostic analytics

Diagnostic analytics is carried out to find insight, i.e. why it happened or is happening.

Here, data science and its subsets: maths, statistics, and econometrics really come into the picture.

#5 Predictive analytics

Predictive analytics is carried out to provide foresight, i.e. what will/could happen.

Correlation and regression analysis is carried out to predict future trends/outcomes.

#6 Prescriptive analytics

Prescriptive analytics is carried out to benefit from foresight.

In prescriptive analytics, BAs (business analysts) give recommendations on how to benefit from predictions and future trends and/or mitigate future business and marketing risks.

All of these BA stages are just the tip of huge icebergs. It take many many years, to master each BA stage.

Now you may ask:

“Where does ‘conversion optimization’ fit in business analytics”?

Nowadays, almost every business has a website, whether or not it sells its products/services online.

So, businesses need professionals who can use data to optimize their online performance and meet their business goals (brand awareness, traffic, conversions, sales, etc.).

Here, there is a strong emphasis on optimizing the “online performance”. 

Data can also be used to optimize a business’s “offline performance” and to make a wide range of business decisions, from operational to strategic to predictive.

For that, we use business analytics technologies like supply-chain analytics, predictive analytics, data mining, etc all of which are not web analytics or conversion optimization.

Web/digital analysts are hired to optimize the online performance of a business.

They are responsible for analyzing and optimizing any online and offline footprints of a business as long as they can be measured, classified, or categorized and are used to optimize the [online performance] of a business.

Again, there is a strong emphasis on optimizing the “online performance” here.

Data can also be used to optimize the “offline performance” of a business, but that is not necessarily web analytics.

Conversion optimization is a subset of web analytics.

Web page designs are evaluated and optimized for conversions through various tests (A/B tests, multivariate tests, usability tests, etc.) and voice of customer analysis (surveys, feedback, market research reports, etc.).

Conversion optimizer is not a business analyst.

A data scientist is usually the senior-most data analyst who has extensive knowledge and experience in applying data science to a particular industry. 

They usually hold a PhD in maths, statistics, or computer science. These are the people who work in the capacity of business analysts.

A data scientist may head a whole team of data analysts. Among these data analysts, there could be analysts, who are specialized in analyzing web/digital data. Such analysts are called ‘web/digital analysts’.

Depending upon the size of a company and organization structure, there can be one or more web analysts.

If there is a team of web analysts, then there are going to be junior and senior web analysts within the team, and there is also going to be one senior web analyst who heads the team.

The senior web analyst often reports directly to a data scientist (if there is one) or to C-level executives (CMO, CFO, CEO, the board of directors, etc.).

Conversion optimizers often work alongside web analysts and report to the head of the web analytics team.

Again, depending upon the size of a company and organization structure, conversion optimizers may include a whole team of UI and UX experts or just one ‘guy’ doing all of the conversion optimization and web analytics.

In order for a business to truly grow, you need to optimize every aspect of your conversion funnel from operational to strategic.

A business analyst is in a position to optimize every aspect of your conversion funnel. 

He is knowledgeable about and deals with supply chain analytics, predictive analytics, data mining, business process modelling, stakeholder management, and other related topics.

He/she often works in-house, has more control over day-to-day business operations and marketing activities, and directly deals with key stakeholders.

So, unlike conversion optimizers, the business analyst can actually optimize the whole business process, from operational to strategic to predictive.

Conversion optimizer has control over the performance of the following metrics (provided he/she controls every aspect of the online presence and the corresponding offline footprints of a business):

#1 Gross Profit from Online Sales – the profit after online marketing costs.

Gross profit = online sales – marketing cost (oversimplified definition)

#2 Return on Ad Spend / Return on investment – This metric is used to evaluate the efficiency of your investment. You spend X, you got 2X, 3X… etc in return.

#3 Cost per online lead – the average cost of generating an online lead.

#4 Cost per online acquisition – the average cost of acquiring a customer online or generating online sales or other conversions.

#5 Sales per online acquisition – the average revenue generated through an online acquisition.

#6 Per session value – the average value of a session to your website.

Per session value = total sales / total sessions.

#7 Online Conversion Rate – the percentage of sessions that resulted in goal conversions or ecommerce transactions.

#8 Average order value – the average value of an ecommerce transaction.

#9 Task completion rate – the percentage of people who visited your website and completed your desired task.

However, the conversion optimizer has little to no control over

#1 Operating Profit Margin – This metric is used to determine the effectiveness of your business in keeping operating costs in control. The Operating cost is the ongoing cost of running a business, product, or system which is beyond the control of a conversion optimizer.

#2 Net Profit Margin – This metric determines the effectiveness of your business in converting online sales into net profit.

#3 Net Promoter Score – This metric measures the likelihood of your customers to recommend your business to a friend, relative or colleague. The two factors which play a huge role in getting referrals are:

#4 After-sales service.

#5 Satisfaction with the use of purchased product/service.

Both of these factors are beyond the control of a conversion optimizer.

#6 Online Client retention rate – This metric measures how well your business retains online customers. The performance of customer support (which plays a huge role in retaining clients) is beyond the control of a conversion optimizer.

#7 Phone Call Conversion rate – This metric measures the percentage of phone call leads which resulted in sales. The performance of call center staff (which is actually responsible for converting phone calls leads into sales) is beyond the control of a conversion optimizer.

A conversion optimizer has little to no control over product pricing, packaging, positioning, merchandising, order fulfilment, management effectiveness, day-to-day business operations, after-sales service, etc., all of which play important roles in improving a business’s overall performance.

So, it is important to understand what a conversion optimizer can and can’t do for your business.

The majority of businesses, can not afford the luxury of hiring a data scientist (they are very expensive to hire, very short in supply and in great demand).

The majority of businesses also often do not understand the difference between a business analyst and a digital analyst/conversion optimizer.

This is also because of the misleading advertisements they often encounter, done by some rogue conversion optimizers who promise to make them “a ton of money” in the name of ‘conversion optimization’.

As you know by now, business analysts have entirely different skill sets than conversion optimizers and/or digital analysts, and they need to be hired and trained in-house.

Consequently, a conversion optimizer/digital analytics can not work in the capacity of a business analyst.

So next time you ask yourself this question, why conversion optimization has not made you a millionaire so far, keep the following points in mind:

You can’t just A/B test your way to the top

To truly grow your business, you need to optimize every aspect of your conversion funnel, from operational to strategic.

Just optimizing the web experience for conversions is not enough.

Don’t get misled by false advertisement

“150% improvement in conversion rate”

Sound familiar?

A 150% increase in conversion rate means nothing if there is little to no improvement in sales.

You need to ask yourself the following questions:

• How this conversion rate metric was calculated?
• Is it in aggregate form or segmented?
• Did they increase spending to improve the conversion rate?
• When was this conversion rate calculated? Was that the peak season?
• Does this conversion rate improvement really mean anything?

With so many questions and confounding variables, it is hard to measure the performance of a conversion optimizer who boasts of increasing the conversion rate but shies away from disclosing the real result, i.e. increase in online sales and gross profit.

Don’t get misled by CRO case studies

“How we increased ___________ by ______”

Sound familiar?

“Technique X worked for Company Y in particular instance Z, so it is obviously a technique that will work equally well for your business”—this is what case studies communicate to the average Joe when they are used as marketing material.

Every business and industry is different, what works for one, may not necessarily work for another.

Besides, even small changes lead to big wins if your business has strong sales potential.

On the other hand, if you are a start-up or a small business, then it won’t be easy to optimize your website for sales.

So, next time you read a case study, consider the size and sales potential of the business mentioned in the case study.

One doesn’t need a lot of skills to make an extra £50k for a company whose turnover is already over £1 million. However, growing a small business requires a lot of skills and effort.

Understand statistics

” You can run A/B tests 24 hours a day, 7 days a week, 365 days a year and still won’t see any improvement in sales if you don’t understand the statistics behind such tests.”

Data sampling issues, underpowered hypotheses, statistical significance issues, underpowered tests, overpowered tests, poor data samples, confounding variables, etc., can easily skew your test results and give you imaginary lifts that will never translate into actual sales.

Related Articles:

• Bare Minimum Statistics for Web Analytics

• Understanding A/B testing statistics to get REAL Lift in Conversions

Understand econometrics

Econometrics is the application of mathematics, statistical methods, and computer science, to economic data and is described as the branch of economics that aims to give empirical content to economic relations.

– Source: Wikipedia https://en.wikipedia.org/wiki/Econometrics

” According to the law of diminishing returns, if you keep adding more of one unit of production to a productive process while keeping all other units constant, you will at some point produce lower per-unit returns.”

– source: How to allocate Budgets in Multi-Channel Marketing

So, for example, if you keep pumping more money into Google Ads campaign without changing the present form of the campaign, at some point, you will reach the point of diminishing returns, and once you cross this point, your conversion rate will go down, and cost per acquisition will go up.

Because of this reason, you can’t double your sales just by doubling your marketing budget.

It doesn’t work that way.

Similarly, in the grand scheme of things, the whole conversion optimization process is just one unit of production.

And if you keep adding more of one unit of production to a productive process while keeping all other units constant, you will, at some point, produce lower per-unit returns.

What that means is if all you are doing to improve your business bottomline is to solely rely on your service provider to optimize your website for conversions, then according to the law of diminishing returns, you won’t go very far in your business.

You need to do a lot more than just optimise your website for conversion.

You need to work on improving management effectiveness and fixing operational and strategic inefficiencies.

The conversion rate is destined to decline.

Ideally, your website traffic should increase over time. But it won’t always increase in proportion to conversion volume. 

Because of this reason, your conversion rate (which is a ratio of conversion volume to traffic) is destined to decline over time.

The ever-increasing traffic on your website will always tend to lower the conversion rate. You will always get some traffic that won’t convert, no matter what you do. 

Not every session or user can lead to conversion.

Despite all these shortcomings of the conversion rate metric, every analytics tool on the planet puts each and every visitor/session into the conversion funnel while computing the conversion rate metric.

So, this whole idea of optimizing for conversion rate is innately flawed.

You need to optimize for conversion volume (like sales, leads, etc).

Related Articles:

• Here is Why Conversion Volume Optimization is better than CRO

• What Matters More: Conversion Volume or Conversion Rate – Case Study

If you are not agile, then conversion optimization is not for you

If you take a month to add one piece of code to your website, then conversion optimization won’t benefit you much.

You need to adopt agile analytics methodologies in order to respond quickly to the ever-changing needs of your customers, search engine landscape, social media landscape, and competitive landscape.

What that means is that you need to learn to deploy solutions weekly, if not daily.

In Agile Analytics, the success doesn’t come from the level of insight you get or volume of tracking implementations you deploy but it comes from your ability to adapt rapidly and cost efficiently in response to changes in the marketing environment.

It comes from your ability to rapidly deploy solutions which solve your customers’ problems either wholly or in parts.

Related Article: How to use Agile Analytics to quickly solve your Conversion problems