2017-01-24

“It’s tough to make predictions, especially about the future.” – Yogi Berra

Digital marketing moves at a fast pace. One year something works, and the next year it is obsolete.

Similarly, conversion optimization moves quickly. Sure, there are some core skills that seem foundational and everlasting, but year to year there are also some new skills that crop up.

What are the skills someone should invest in learning if they want to be a top 1% optimization expert in 5 years? What should they learn today?

What are the industry and technology trends that are fueling the need for these skills?

I talked to a handful of experts as well as polled different forums to get people’s thoughts. We also had internal discussions as to the trends we’ve seen here at CXL, both from the skills people request at CXL Institute and the experiences we’ve had with clients.

Optimization Beyond the Landing Page

It’s sort of a trope and a frustration in the optimization space that most people new to the craft think of CRO as a simplistic practice of A/B testing button colors or CTA text.

Everyone has railed against these practices (including us, many times), but it still seems that much of the industry is focused on short term metrics and surface level changes. Just read this (terribly lacking) Wikipedia page for CRO.

The future seems to belong to those moving far past myopic CRO practices and into the strategic and universal realms.

Those working in data-centric tech organizations probably don’t find this idea out-of-the-norm. With agile development cycles and uncertain monetization models, they live and die by data and experimentation.

But this type of universal optimization is reaching more and more organizations as they see the value in controlled experiments at scale.

As Ronny Kohavi, Distinguished Engineer, General Manager, Analysis and Experimentation at Microsoft, put it:

Ronny Kohavi:

“Experimentation is getting more recognized as essential to guide product development, when it is applicable (not always, as you can see in Section 6 in this paper).

We can see folks from Statistics, Computer Science, Economics contributing to the domain, making progress on both theoretical and practical areas.”

Stephen Pavlovich, CEO of Conversion.com, echoed this as well:

Stephen Pavlovich

“Conversion is expanding beyond just the core user experience. More and more, people are using experimentation as a way to optimise not just sales – but also pricing, functionality, and product.

And that expands the skills that optimisers need. At the core is still the same balance of analysis and creativity – but you’ll get an advantage with experience in product development, marketing, and commercial strategy. “

With the overlapping areas of marketing or product with optimization, it helps to be skillful at one of the former areas. Similarly, it helps to have a unified view of conversion optimization to help you approach every avenue of optimization – from product to communications and call centers to landing pages and more.

Talia Wolf, CEO of taliagw.com CRO training and consulting, puts it this way:

Talia Wolf:

“The knowledge gained from a more in-depth approach increases retention rates and generates invaluable knowledge to every unit within the business. Using the knowledge gained from this type of CRO process can optimize your sales teams, retention teams and even your shipping methods.

Optimizers who want to lead this type of transformation in the business must be able to do a lot more in-depth research, it won’t be enough to just be analytical or creative. A profound knowledge in every aspect of marketing and business development is necessary (e.g. research, data analysis, UX, psychology, copywriting, design and many more skills).”

Matt Gershoff, CEO of Conductrics, advises those in the space to learn about reinforcement learning as a framework for optimization and targeting. Here’s a good article to read if you’re looking to brush up on such a framework.

Conversion Optimization as a Mainstream Discipline

It looks like conversion optimization as a skill will also be embedded in many other traditional disciplines.

For example, a web designer. As Maarja Käsk said in a Facebook group discussion, “everything a CRO-minded designer does & knows how to do will be common knowledge for all UX/UI designers. So there would be no point in including “CRO” in the name of the position.”

Similarly, growth marketers, digital analysts, email marketing specialists, and customer acquisition specialists all do some form of conversion optimization. As we learned in our 2016 State of the Industry report, those who do conversion optimization go by a variety of titles.

While this doesn’t mean that centralized conversion optimization teams will go away (that would likely be a bad idea), it does mean that anyone involved in a marketing function should have a good understanding of the CRO process and the skills that make up the discipline.

This trend will likely be expedited by both the increasing salience of conversion optimization by “thought leaders” and the ease of access to testing tools.

The democratization of testing also accompanies and increased integration of data across tool sets. According to Krista Seiden, Analytics Advocate at Google, analytics and optimization are moving closer together, making so the data speak together more fluently and accurately:

Krista Seiden:

“Sharing data from Analytics to Optimization is one of the core tenets of Google Optimize, and this really allows CROs, marketers, developers, and many others to dig deeper into optimization and personalization.

As these specialties move closer together, it’s important to reinforce best practices.

One thing I have always been focused on, and will continue to be focused on, is educating and fostering a deep understanding of good testing practices and culture within a business, and I believe that the democratization of testing gives us the opportunity to embed this thinking even further into business.

While many are eager to take advantage of the benefits of optimization, an understanding of data, analysis, and statistical significance is key to success. As educators, it’s our job to ensure the increased interest and adoption of optimization practices is done right”

Similarly, Paul Rouke, founder of PRWD, explained in eConsultancy that the ease of access to testing tools could bring a devaluation of process and rigor, as well as more low-information users:

Paul Rouke:

“Much like the launch of Google Analytics provided a quantum leap in the amount of businesses across the world using web analytics data (caveat I am using the words “using web analytics data” loosely here), Google Optimize is also going to start bringing the concept of A/B testing to the masses.

On the one hand, this is good news for the awareness and credibility of the conversion optimisation industry. Google’s rubber stamp (and an improved tool from its last effort) will mean that more people will be developing a culture of experimentation.

On the other hand, the harsh reality is, when we get something for free, we typically place less value on its importance and the need to invest time and money into it.”

He worries that the increased salience and adoption will inevitably lead to more low-information users. That, in turn, may lower the perceived value of experimentation as people become disillusioned by poor results and ROI.

Paul Rouke:

“Will SMEs understand the different statistical models they need to use to understand whether a variant on their testing tool is truly the winner? Will all businesses be able to configure their testing tool to their analytics and ensure the data they are recording is correct?

Just as Google Optimize will help make the CRO industry visible, it will inevitably bring about poor practice and misinformation.”

It’s up to leaders in the industry, then, to educate and evangelize those new to the space on the correct ways to do things.

Michele Kiss, Senior Partner at Analytics Demystified, also mentioned the tradeoff between the convenience of cheap, easy-to-use tools, and the skills required to use them intelligently. Though, she mentioned that these tools are not entirely new:

Michele Kiss:

“Companies that want to get their feet wet with optimization are typically able to do so without a huge upfront investment. However, doing so without the appropriate process, skilled resources, and understanding (and respect) for sound statistical methodology can lead to significant business consequence.

While “ease of use” tools have their advantages, those working in optimization must possess the analysis skills to go beyond automatically calculated test results, to truly understand the significance of a particular test.”

Matt Gershoff also talked about the lack of depth and the technical debt incurred by this type of narrow-focused optimization:

Matt Gershoff:

“Because many of the popular optimization tools out in the market today fused together experimentation with easy point and click tools, we have an industry that:

has been trained to think that the scope of optimization is just presentation layer changes to websites

incurs enormous technical debt trying to manage bolt-on, web based optimization tools as they try to scale.

As the market matures, these limitations become more apparent, and with it the increased demand for APIs and ‘headless’ optimization platforms. By decoupling optimization from graphical UIs, companies can both efficiently manage these capabilities and can chose to embed them directly into any transactional marketing application, either on the client or server.”

Deeper Data Skills

We know that knowledge of statistics is important for A/B testing, and that many people get the simple things wrong. We also know intuitively that it’s important to be able to dig around in Google Analytics to get insights.

But that’s just the tip of the iceberg.

Ronny Kohavi puts it this way:

Ronny Kohavi:

“The key skills are those you see everywhere about data scientists (be able to go deep into data, ability to translate results to something that’s easy to understand) but with additional knowledge of statistics.

For example, most people don’t understand what a p-value is.

Trends include: big data tools, data manipulation (SQL and alternatives), languages like Python, and deeper analyses (e.g., heterogeneous treatment effects).”

If you’re active in the analytics community, you may have noticed a trend in 2016 towards things like R and Python, and it seems that this combination of technical abilities with traditional analyst tasks is becoming increasingly valuable.

The ability to use data is, of course, not limited to conversion optimization. Emarketer made a similar prediction that the future would belong to those with advanced data skills (check out the “in 3 years” bars below):

The ability to connect large disparate sets of data and pull actionable value is central to website optimization as well as other marketing channels.

Regarding increased data analysis skills, Michele Kiss had this to say:

Michele Kiss:

“In the coming years we are likely to see two developments:

A continuing evolution of the analytics skill set of individuals working in optimization (for example, more sophisticated use of statistics, or stitching of data sets for more informed results)

The growth of overall analytics team capabilities, via complementary teams of more business-facing analysts and deep-dive statisticians.

While it is ideal to think that one person can work with designers, UX, product managers and IT to manage tests, conduct deep statistical analysis and present easy-to-understand to executives, unicorns don’t exist, and even if they did, they can’t act as a packhorse, doing the work of five! Specialization is a likely consequence of growth of the industry, and will enhance optimization efforts for companies that invest in these larger teams.”

Ryan Urban, founder and CEO of Bounce X, puts the emphasis on repeat traffic and behavioral marketing, saying that if you can connect anonymous visitor behavior to leading indicators of purchase intent, you can unlock value you didn’t even know was there:

Ryan Urban:

“IMHO, everything is focusing on the wrong things to improve conversion. The key to 2017 conversion is evolving your traffic dynamics and that starts with behavioral marketing.

For maximum business growth + substantial revenue/session increases, you have to compel your highest intent and top converting traffic to visit way more often.

How do you do that? True Behavioral Marketing begins by identifying your anonymous visitors down to an email address and setting up people-based marketing automation to get them back. This could be as simple as identifying 50% of your visitors and delivering them a behavioral email immediately after their session ends with every product they were interested in.

If you do behavioral marketing at scale, it unlocks an enormous new free revenue channel out of thin air and drastically changes traffic dynamics.”

Of course, the technology to do this is becoming increasingly sophisticated, but it doesn’t hurt to really grok Data Management Platforms (more ad-focused) and Customer Data Platforms (more conversion/personalization/retention focused) as well as marketing automation systems.

AI and automatic optimization

We’ve had a few articles on AI and its implications for optimization, and of course, journalists are breathlessly pumping out articles on the effects AI will have on the future of work and society.

It’s a lot of noise, but what does it really mean for digital marketers on a personal level?

It’s impossible to say with precision, but from what we’ve seen with a tools like Sentient Ascend, Conductrics, or even the Stream feature by TryMyUI, it just makes our jobs a little easier.

AI is still largely misunderstood by the population at large, and that includes marketers. If one wanted to prepare for the future in which we are the slaves of our robotic overlords, one should start with a simple understanding of what AI actually is.

A lot of the skills necessary for AI and predictive targeting will revolve around the management problem. Matt Gershoff explains this in the context of a real and very much current law in Europe:

Matt Gershoff:

“For companies not paying full attention, the rush to implement machine learning automation will run smack in the European wall called the General Data Protection Regulation (GDPR). My prediction is that Article 22 of the GDPR, which gives EU citizens the right of explanation for any ‘significant’ decision, will impose the following two constraints on the use of machine learning automation:

Auditable – each time a ‘significant’ decision is made for a customer, the rule that was used will need to be logged with the decision so that the company is able to recall it if there is a customer challenge

Interpretability – the rules will need to be human interpretable so that the company can explain to consumers looking to challenge why the decision was made.

Of course, interpretability is good design regardless, as it enables the marketer to grok how and why the ML system is operating.”

But human creativity and strategic input still applies. You still need to pull your own insights.

Paul Rouke argues that this doesn’t mean the diminishing importance of human intelligence. In fact, it means it’s more important than ever to use creativity and empathy in your work:

Paul Rouke:

“Human Intelligence is more important now than ever. To match customer expectations, businesses want to create engaging and exciting online experiences and the only way to do that is through creativity and understanding. At this point, AI can’t replace these two human attributes.

To truly draw value from machine learning, you still need to have a human behind the machine, ‘feeding’ it ideas, concepts, and designs that have been built from user research and in-depth data analysis.”

While AI-based algorithms can test hundreds of combinations or serve the best variations to the most relevant traffic segments, it’s still up to humans to create those variations.

Peep Laja, founder of CXL, had this to say about the proliferation of AI in optimization:

Peep Laja:

“The are jobs to be done in optimization that “machines” and algorithms will be much better at than humans. In many cases it’s already true. Look at the way Conductrics works to serve the best possible digital experience (“variations”) to the segments where they work (convert) the best. Humans suck at this. Setting up static “if this then that” personalization rules is history.

When it comes to data analysis, finding correlations and causal relationships between user behavior and likelihoods of closing the sale, machines will beat humans easily.

Harvard Business Review called “data scientist” the sexiest job of the 21st century, but it’s already starting to be one of those jobs that machines will take over. Automated data science platforms can just crunch through more data much, much faster – without bias – and produce insights, visualize them, conduct reporting and even take instant action.

The very best data scientists will remain of course, but the “I know a little bit” kind will be all rendered useless.

Optimizers of the (near) future are still needed to come up with versions of possible digital experiences – which includes writing copy that connects with human beings, and coming up with all-encompassing UX strategy for various buyer types. And then the algorithms will choose who should be part of which experience.”

More Human Intelligence Needed, Too

This gap between the functionality of artificial intelligence and the necessity of human creativity and insight opens up the increased need for applied behavioral psychology.

As Jairo Moreno said in a Facebook group discussion, “As time progresses and the web is maturing, the layers of the conversion pyramid become commodities from the bottom up.

With functionality nearly solved, when good usability is almost standardized at a reasonable level, it will be all about persuasion.”

That leaves psychology, neuromarketing, etc., as an important point of differentiation and actionability. With your deeper data skills (or those of your data scientists), you can tell where campaigns are lacking and correlations between customer segments and purchase intent. You can cluster behavioral data at an increased level of granularity to predict actions.

But that data doesn’t become truly actionable until you can connect human psychology triggers to those data points.

Interactive Interfaces

There’s no shortage of people proclaiming 2017 as the year of the chatbot.

Conversation is the new interface, and bots are the new apps.

So where there used to be a form or a human support rep, you may soon be seeing a conversational interface in the form of AI and chatbots.

As Craig Sullivan put it in a Facebook group discussion, “I’ve been predicting for years that lead generation, contact, and various other types of forms would disappear, to be replaced by conversations. Chatbots, voice AI – all of these can do the job of converting people far better than shitty web forms.”

He also shared an article that outlines an open source chatbot. Check it out here.

A bit part of the future of commerce, user experience, and optimization could lie not just in the technical understanding of chatbots, but in understanding the nuances and qualitative things that make them effective.

Virtual and Augmented Reality

Virtual reality is another area in which we’re already seeing huge changes. Lowes and Myer have both bought into delivering a virtual experience, and augmented reality lets you see an IKEA couch in your own home.

As Benoît Quimper said in a Facebook group discussion, “Forget the website as we know it – it has 10 years on the countdown and augmented reality is a likely disruptor. Therefore, product demos, “trying before buying,” live support, etc., may all become part of a performance strategy.”

One massive constraint of ecommerce, and one reason brick and mortar has stayed so strong, is that you can’t actual experience the items you’re shopping for. You can’t pick up and hold a pair of jeans, seeing how they would look on you, etc.

We’ve tried to bridge that gap in many ways, like offering 3D rotating images or videos, and investing in ROPO attribution so customer can choose to buy and research wherever they feel comfortable.

Conclusion

Today’s CRO experts are already, on average, more technically savvy and comfortable with data than traditional marketers or more generalist digital marketers. That trend will only continue and deepen in the future.

Learn advanced data analysis skills and more traditional data science crafts – Python, SQL, R, and other big data skills. Get comfortable with AI and predictive targeting tools, and learn to play with conversational UI elements like chatbots.

Look to experimentation and optimization as a driving force in product development and other business areas, not just landing pages and UI changes. Full-stack optimization will become more and more mainstream, especially as testing tools build out features to promote this universal optimization.

Finally, as Peep commented in the article, as our robot overlords take over the more rudimentary data tasks, the ones that rise to importance are remarkably human: writing copy, persuasion, UX design, drawing insights, etc.

Pick up skills to prepare you for the future of conversion optimization at CXL Institute.

The post What Does the Future Hold for Conversion Optimization? appeared first on CXL.

Show more