title
 

Digital Analytics. What on earth’s next?

, Posted by Darren Hutchins in Blog, Darren's posts, Web Analytics

Darren Hutchins, Technical Analytics Manager at Mezzo Labs, looks at what the key trends & innovations may be in the year ahead.

 

Up until a few years ago, most people would regard digital analytics as dry, unsexy, and just simply not exciting in the slightest.

Images of Moss and Roy from the popular Channel 4 show ‘The IT Crowd’ would serve as a key stereotype when thinking about such topics.

However, things in the world of web analytics have changed remarkably quickly  and much for the better.

The digital data space is now a haven for “cool” innovations, enabling many brands to transform how they advertise their products, shape their digital and physical experiences for consumers and even how they communicate directly with them.

So what do I think are the key up and coming innovations & trends in the space in the year ahead?

 

Machines have emotions too….

 

Ok, maybe I’m jumping the gun, but empathy? Yes absolutely.

Imagine the thought of a business being able to understand their consumers on an emotional level, seems crazy right? And may seem personally intrusive to some.

Well think again!

You’ve got your standard quantitative survey tools (ala Usabilla, Qualtrics), sure.

You’ve even got your more strategy driving qualitative customer tools (like Foresee).

However, you can argue there’s too much uncontrolled bias swimming around in this data oasis.

How could you really tap into genuine consumer thoughts, beliefs, and sentiments without spying on them or with some sort of brain probe? Charlie Brooker thinks this is already happening. Well, in reality emotion recognition tools seem to be just the way of gaining such insight. There are a fair few knocking about right now.

Emotional recognition tools

One of the cooler solutions I’ve seen comes from Affectiva.

It’s a cloud-based solution, called ‘Affdex for Market Research’.

They term it facial coding, and it’s literally surveying consumers and their reactions to served content on digital sites.

Every single facial expression is recorded and translated to the relevant perceived emotion and all this through the use of a bog-standard webcam….. cool huh?

Also, a little creepy….

The results are aggregated and displayed in an easy to use dashboard.

This data allows companies to forecast KPIs like brand recall, purchase intent and likelihood to share, to name but a few.

Obviously, a lot of people will have justified reservations about such technology, in the wake of privacy scandals in the media of late, but this is certainly a new horizon that cannot be completely ignored.

Alternative emotion-based observation techniques

On a slight tangent, I’ve been reading a fascinating book recently by Martin Lindstrom called ‘Small Data’.

He’s a self-professed data detective going to fairly extreme lengths to help failing brands turn themselves around, and invariably become hugely successful.

Why am I mentioning this you may ask? Well his methods are somewhat unorthodox and crucially bear relative significance to these kinds of tools.

He surveys consumers in their natural environment, usually in their homes to understand how they use, interact with, and regard certain products in their daily lives, akin to Louis Theroux in many ways (minus the comical British awkwardness).

Lego’s turnaround in fortunes in the mid to late 90s was helped in large part from an observation he made of a child, and how proud he was of a pair of trainers he wore. They were battered, and on the face of it seemed of little spectacle, but crucially stood as a symbol for his skateboarding prowess

Something inconspicuous to many, but to Martin, provided the key insight that children attain a social currency amongst each other from mastering a skill of some kind. This led to him giving a recommendation that Lego should return to its original core competence, but with a slight twist, in selling products that required a certain mastery to build.

Needless to say, their fortunes began to return pretty quickly.

Body language the key

Anyway…… The granularity of observation is the key shared concept here, and whilst businesses conducting this level of research is not practically feasible, tools like this will make landmark discoveries of this ilk a lot more attainable. Personally, I think is very exciting for the future of digital & e-commerce.

Some big players have seen the potential opportunity in this, with brands like Kelloggs, CBS and Mars, using such tools to optimize their content and spend on advertising.

After all, Albert Mehrabian’s 7-38-55 rule dictates that tone of voice only accounts for 7% of all personal communication, 38% attributed to tone of voice, with the largest percentage (55%) being for body language.

If we as businesses go by this theory in regard to customer communications, emotion recognition tools of this nature are unquestionably an untapped resource right now, which could change the business landscape for the better.

 

I will be eagerly awaiting the next uses of this in the coming year….

 

 

I want that pizza, you know which one……

 

Staying on a similar theme, eye tracking is something brands are thinking about, and in some cases, quite creatively.

Bringing innovation to the culinary market, who else but, Pizza Hut!

For the hugely unsociable amongst us, or just for the ones of curious mind, Pizza Hut now have an alternative way to order their famous pizzas.

A tablet device displaying numerous toppings (as above) constitutes just that.

Integrated within the tablets are eye trackers, which track the movement of your eyes across the screen of toppings. A sophisticated algorithm then calibrates your optimum pizza choice based on the time spent ogling the delicious individual toppings.

Overall, a good little commercial use of eye tracking technology.

Target (the retailer) have started using sensors within their stores to track eye movements too, which really excites me.

Retail can be a great market for this idea to really flourish.

 

 

I had no clue at all that was happening…..

 

As analytics professionals we all strive to, and take pride in getting that golden nugget, that piece of insight, which shows us why our products aren’t selling as well as we thought they should be, why our marketing ads aren’t quite hitting the mark, and so on.

The only catch is, getting to these mini euphoria moments can take a while, and all the while businesses are losing out.

At the last Adobe Summit, I went to a session where they spoke about how cool it would be to automate this work, effectively giving companies more analysts in their workforce (just not your generic human kind) without the detriment to costs.

Analyst AI tools

Soon after, Adobe released the Virtual Analyst.

Luckily, I had the chance to use this tool last year in and boy does it get to the nub of an issue quickly.

In summary, the Virtual Analyst constitutes the combination of 3 key features newly integrated within the Adobe Analytics product.

These are Anomaly Detection and Contribution Analysis, and Intelligent Alerts.

Anomaly Detection enables the ability to pinpoint the potential factors that have contributed to given anomalies in the data. It utilises statistical significance to highlight only factors that really could have caused the anomaly to occur.

Contribution Analysis then allows a deeper dive into these found contributory factors, and calculates the percentage likelihood that each factor has caused the peculiarity in question.

The key part of the overall capability in making it feel as if you have a virtual analyst working for you, is the Intelligent Alerts feature.

Intelligent Alerts can proactively spot anomalies in your data and inform you of these whenever you choose for it to, every second of the day almost if you wanted!

You can create alerts to trigger on anomaly contributing factors, based on different statistical thresholds, i.e. 99.5, 90%, etc. and relativity measures if contributing factors are close to these thresholds.

Uses

A good use of it I experienced previously was being able to identify individual marketing campaigns causing slower growth or even decreases in a given marketing channel’s traffic acquisition strength.

It is a feature that I expect to see a lot of companies using in the year ahead, as it has huge potential power to uncover blind spots for many marketers, and challenge conventional ways of thinking and decision-making.

 

 

Why can’t we all be data analysts?

 

In a world where companies have ‘Big Data’, vast quantities of data from many different sources, there is a need to make more efficient use of such data.

Traditionally in the past, a person would need to have a fairly niche skillset in programming and statistics, in order to make sense of the reams of data held in databases.

However, modern-day analytics tools have enabled a considerable shift away from these kinds of black-box systems. The systems used now have simple graphical user interfaces, making it much easier to conduct analysis of data.

This is a fantastic step forward for all companies digital, allowing their analysts to explore many different topics and attempting to prove/disprove hypotheses galore!

Modern day data expectations

Let’s take this a bit further. Expectations now change, a level of insight and reporting once deemed too much, is now normal service.

Businesses see the value in digital analytics more than ever. Some restructure their business around it, where maybe in some cases each digital team has their own set of micro digital KPIs and as a result each team wants their own analyst to help them achieve these.

So, you may think there is only one option: hire more analysts, and sure, you could.

However, in the current marketplace the supply of such professionals isn’t unlimited as more and more businesses undertake digital transformations.

They will cost more than they ever did too, from the demand-pull inflation for such resource.

Empowering the organisation

Don’t fret though, there’s another way.

What if you trained your non-data analysts to use these systems to report and analyse data.

This is what a lot of people in the industry are referring to as data democratisation.

This is what I helped pioneer at a previous company I worked for, and it proved a big success.

Of course, this doesn’t mean that you suddenly need less analysts, or your collective analyst workflow decreases that considerably.

It also requires a well put together training programme for upskilling the personnel in these tools.

However, it gave two instant advantages:

  • Less menial requests were coming through to the dedicated analysts in the team, allowing them to invest more time in deep-dive insight pieces
  • More business areas started to make key decisions based on data rather than gut instinct and other intangibles

Get ready for hesitation

Sounds great right?

However the whole business world is not digital, and therefore you have to expect and be ready for hesitance and a certain level of friction from some business areas.

All in all though, this is a topic that could potentially revolutionise how businesses make use of data, but don’t be fooled into thinking this is a overnight silver bullet delivery, thorough planning and alignment of business objectives is key to success here.

 

 

A.I. – Let’s get customers and robots talking

 

We all know about chatbots, they’ve been around for a while now.

Up to now though they’ve mainly been used by companies as an extension of their customer service proposition. We must expect that in the coming year and beyond, that chatbots could become the main (and potentially only) way companies communicate with their customers.

With that said, tracking these interactions at that point will become a necessity.

Chatbots in use

One of the big brands using them currently is H&M, who use the popular Kik platform.

Their bot serves as a personal shopping assistant effectively, to help you on their website.

To start with, it asks you various questions to understand your fashion style, before calibrating that information to recommend outfit choices for you to peruse.

If you decide that you agree with any of the proposed choices, the bot will redirect you to H&M’s mobile site, deep linking you straight to specific product page, giving great ease of purchase.

How it works

The really interesting part here is how the calibration of the customer’s fashion style is executed.

They use an algorithm which makes use of analytics captured from the chatbot, in real-time, before seamlessly providing the suggestions of things you’ll like.

In terms of truly understanding the customer, this kind of data goes above and beyond what you can traditionally capture from user site journeys, and will prove invaluable to businesses in so many sectors.

How to implement

Doing some research on how to implement such tracking, and integrate within modern analytics tools, the following two articles are some of the better ones I’ve come across to help get to a point where you hold such data.

 

Summary

Referring back to the previous point around increased ubiquity of bots, making such the primary digital communication style, you could quite easily predict where customer expectation could get to very quickly in this context. There is a potential that the chatbots could fully administer the point of sale without the mobile site journey redirection, breaking the otherwise one touchpoint experience.

In summary, chatbots aren’t going away anytime soon, and could quite feasibly change the way e-commerce businesses communicate with their customer, so it’s certainly a case of watch this space for now….

All in all, 2018 is likely to be a very exciting year in the space, with traditional analytics capability shifting away from focus, with the lens firmly on artificial intelligence and data democratisation.