Posts in Behavioral Data

personalisation

It’s the Thought That Counts

December 8th, 2017 Posted by Analytics, Behavioral Data 0 thoughts on “It’s the Thought That Counts”

Think about the feeling you get when someone you care about gives you a gift. Surprise… delight… overall, a pleasant feeling, right? That’s because a gift is thoughtful and gives the impression that some time and effort went into providing that special gifting experience.

That’s exactly what customers are craving for from their favourite brands. Customers don’t want to feel misunderstood or overlooked. They want personalised content, offers and recommendations which surprise and delight them.

Personalisation is not some abstract concept
AI has become the standard for all winners in the online retail ecosystem.

Large, multinational, growing companies are investing more and more into personalisation each year. And the stakes couldn’t be higher during the holiday season. In 2017, Cyber Monday alone brought in $6.59 billion in online sales, surpassing all prior years. Black Friday didn’t disappoint either – there were $5.03 billion in online sales.

But these hoards of customers don’t magically appear out of a hat. In this competitive environment, retailers need to earn each customer. And a growing number of online buyers expect tailored offers and purchase recommendations from the sites they visit.

So how do you get ahead of your competition to attract customers? Two words: machine learning (also commonly referred to as ML).

Machine learning has become the standard for all winners in the online retail ecosystem: Amazon, eBay, Wal-Mart, you name it, are all utilising ML. And if you’re not (that is, if you’re looking at customer data as a static thing), you will quickly lose out.

Today it’s absolutely possible to map and predict customer behaviours based on real-time interactions. Every hover, click, scroll, and type builds a sequence of events unique to an individual user. ML, together with Data Science, helps leverage data more effectively, automatically driving insight from the data and helping organisations to understand the customer better. There are ways to use insight and model output to power the various customer experience platforms in your organization. And the earlier you implement ML, the quicker you’ll deliver relevant content and convert customers before your competition does, and the sooner you can achieve peace of mind with your marketing campaigns.

Are you getting it right with your personalisation efforts? Feel free to share your experiences with me. What’s working and what’s not? You can email me or connect with me on LinkedIn.

 

clickstream data

Unlocking the Value In Your Clickstream Data

November 15th, 2017 Posted by Behavioral Data 0 thoughts on “Unlocking the Value In Your Clickstream Data”

In 2017, 65% of all US retail sales are involving a visit to the company’s website. And as you may have already realised, your website is one of the most important sources of data when it comes to understanding your customers’ wants, needs, and preferences. Assuming you have web analytics, this source of data gives you the opportunity to track and collect data on every single interaction a customer makes with your brand.

In-store you can track what a customer has bought or returned. On social media, you can get a view of which adverts they have seen, or which content they have browsed. But it’s only with web analytics that you can truly track every interaction, regardless if a purchase was completed.

Sure, you cannot track the mood the person was in when browsing nor what they had for breakfast. But no other channel gives you the chance to learn so much about your customers. From which pages and products someone has viewed to time spent on page and engagement with bespoke content widgets. From basket interactions to video plays and customer feedback views. It’s clickstream data capture which allows you to track every interaction with your brand.

The aforementioned clickstream audience gets us to a point where most organisations are today, leveraging their clickstream data in their chosen personalisation, DMP, and email tools. There is, however, much more that can be done with this data. You might say that you could look at a customer’s historical in-store transactions and work out what you think they might be interested in next. However, finding a segment of 100,000 people, in the last 14 days, who have viewed a certain product type without purchasing, is undeniably more powerful. I know which audience I would rather market to.

So, if 65% of all US retail sales are involving a visit to the company’s website, and if overall web share of sales typically ranges between 35% – 60%, there is no doubt about the impact your website on overall sales.

e-commerce sales increasing

How can we unlock the value of our clickstream data?

Web analytics is great. You can track demand, product interest, online sales, and even identify problem areas on your digital properties. These are just some of the critical functions your web analytics tool can do. However, such tools have their limitations.

The type of querying you can do is pretty limited, as are the types of visualisations you can create. The data science functionality, when available, is restricted to proprietary, pre-defined models. The schemas that these datasets are stored in are predefined and the type of data you can pull into these applications is generally limited to lookup data.

To really unlock the value in your clickstream data you only need to pull the data out of these applications and into a Big Data platform. When you do this, you can knock down the barriers of restrictive querying, focus in on individual customers (not cookies), and get clickstream data feeding in along with your many other data sources. Only then can you truly achieve an omni-channel view of your customer.

The 5 steps to unlocking the value in your clickstream data are:

1. Customers buy from brands, cookies do not buy from channels. Do Big Data.

Most organisations have realised that customers expect an omni-channel experience and have started (or are about to embark on) a Big Data platform project to handle this. It is of paramount importance that the clickstream data be a part of this project. Clickstream data can be tricky to work with, as Web Analytics vendors do not always provide this data in a format that is easy to use. However, with so much data on existing customers, as well as prospects, this data source is just too valuable to ignore. When the clickstream data is accurate and made available in your Big Data Platform, in a schema that makes sense, you can achieve a true, omni-channel view of how your customer is interacting view your brand and begin to understand and model how best to serve them.

clickstream data

2. Data Science. Stop ignoring 90% of your data.

When querying and segmenting, web analysts tend to focus on high-value data such as purchases, cart adds, visits to key pages in the conversion funnel, etc. You might then segment this data by some key dimensions, product types, geographical properties, first-time visitors, etc. Realistically you might be able to take these key metrics and apply 5 or 6 layers of segmentation to get to something interesting, be it an audience or a metric. Even in this scenario, you have probably ignored 90% of the data you have for each visitor included in the audience. Think of all of that data your Web Analytics tool collects with each tag fire. Then multiply this by the number of tag fires each visitor generates. You have so much information on your web visitors; why ignore it? Using data science techniques, you can begin to consider the other 90% of your clickstream data and begin to find behavioural patterns. For example, just because I have viewed a certain product or service more than 3 times, or added something to my basket before abandoning, does not mean I am ripe for retargeting. Conversely, just because I have not completed these interactions does not mean I am not interested in buying the product. The truth may lie in everything else I have done on your website. Start considering the full picture and use open source libraries when doing so.

3. Reporting, it is a business-wide problem.

Every business needs to produce reporting that tracks sales, revenue, and demand. Clickstream data is a vital component of this, particularly for tracking digital sales but perhaps most importantly as a measure of demand for your various products and services. If you do not have your clickstream data in your Big Data Platform you are left with the unenviable task of manually trying to combine separate reports into a single sales dashboard. Worse still, you may simply ignore the clickstream data and not report on your demand levels. But with your data consolidated in one place, not only can you easily produce high-value omni-channel business reports but you can also be more agile in answering any follow-up questions, with the required data in one place and ready to use.

4. Monetise your clickstream data, it is already costing you a small fortune.

Web Analytics tools are not cheap, and the more data you collect the more you have to pay. My previous points have examined optimal ways of storing, processing and querying your data, allowing you to learn more from the data you already have. Once you have accomplished this and created various scores and segments for your existing customers and prospects, you need to get this information into the applications you use to communicate with your customers. Connections between your Big Data Platform and your marketing tools (DMP, personalisation, customer contacts, and anything else you use), be it for online or offline customer communication, need to be established. At that point, you can automate the data flows and ensure customers are receiving a relevant, personalised and consistent experience with your brand, across the different channels that are available to them.

5. Automate everything you can.

Within an Analytics function, there are a limited number of people. Typically, these people are highly skilled in one or more areas of Analytics, be it querying, data science, visualisations, or data engineering. By automating as many processes as possible, you can ensure that the resources available to you are delivering new insights and are seen as a value driver rather than a potential constraint or blocker. When it comes to clickstream data, the aim is to automate the ingestion, processing, and auditing of this data in your Big Data Platform. This frees up the data engineer to work on new data sources and ensures your analysts and data scientists have up-to-date, ready-to-use data at all times. The same then applies to getting data out of your Big Data Platform and into the applications you use to communicate with your customers. Establish these data pipelines and automate these flows. Without this, you risk becoming a blocker to the business, as the time to delivery of new outbound data sends will be very long.

So, if you’re looking to boost sales and revenue, clickstream data will be your go-to source of data to find the customers to reach out to. It should also be an essential component in your businesses data strategy. If you get this data into your Big Data Platform, in a structured and usable format, and establish the required outbound data pipes, you will be able to give your analysts the tools they need to increase data value, and your marketing team will be able to maintain true, omni-channel engagement with prospects and customers.

 

References:
     – Digital Commerce 360: 60% of U.S. Retail Sales Will Involve the Web By 2017
     – Deloitte: Understanding Consumer Shopping Behavior
     – RetailNext: Brick & Mortar Vs. Online

 

 


Bal Basra

Bal Basra is a Solutions Consultant at Syntasa, assisting customers and prospects in extracting the maximum value from their big data projects. He brings tremendous knowledge of analytics products and methodologies from previous positions at Adobe and TUI. Connect with him on LinkedIn.

 

retail online in store brick and mortar

Retailers Can Have Their Bricks and Click Them Too

August 30th, 2017 Posted by Analytics, Behavioral Data 0 thoughts on “Retailers Can Have Their Bricks and Click Them Too”

Despite what you might have read, retail is not dying. Sure, brick-and-mortar retailers today face significant competition from their digital counterparts. But with the right omni-channel strategy, they can outperform online-only stores by providing the best of both worlds – the convenience of online e-commerce along with the human experience of physical stores.

Last week, Macy’s announced its new President will be Hal Lawton, a former eBay and Home Depot executive who is credited with building Home Depot’s stellar interconnected retail experience. Macy’s knows that the path to sustainability involves a unified online and in-store strategy and it has plans to expand its data analytics and consumer insights.

That’s because today retail stores are sitting on an enormous mound of customer and enterprise data, which includes point-of-sale receipts, online visits and purchases, warehouse inventory, and so on. And all of these data points are extremely valuable with the right data analytics strategy and technology in place.

In particular, predictive behavioral analytics has allowed retailers to know when to do what and where. As a result, a store can maintain optimal inventory levels and anticipate what a customer will want to look at on their next visit. It can also pair up a customer’s in-store and online activities to ensure a seamless customer experience and optimal conversion rate with each visit. Imagine, a sales representative having the most up-to-date customer information at their fingertips to help the customer determine the next best action.

It is these kinds of capabilities that will allow companies to stay relevant and win big.

If any of this resonates with you, tweet at me or email me to share your thoughts and experience with analytics.

 

data trove

One Man’s Trash is Another Man’s Data Trove

June 28th, 2017 Posted by Behavioral Data 0 thoughts on “One Man’s Trash is Another Man’s Data Trove”

Companies have been hoarding too much data on consumer demographics. There is an over reliance on demographic data for consumer insights and business decisions. Luckily, a select group of companies are uncovering a data trove. With the advent of predictive analytics, they are realizing that anonymous and behavioral data can shed more insight on consumer behaviors, and that it’s this kind of data which can accurately predict customer intent, and not age, gender, or location.

A single, anonymous visit to a website can provide sufficient insight about the entity in order to identify, segment and predict their propensity for a particular outcome. Entities exhibit tendencies and behaviors of intent, regardless of their age or where they come from. In much the same way that you can catch a thief based on their past behavior, you can also “catch” an intent by analyzing past behavior.

Algorithmic modeling can capture behavioral data and treat each data point like a unique thumbprint, representing the unique sum of all outcomes carried out by that one entity across time. This kind of model is not only precise, but it also gets better with time. The more behaviors that get mapped against intent, the more accurately the model can identify the propensity for intent and outcomes, and deliver the next best offer at the appropriate time.

So embrace your anonymous data. It’s worth so much more now than ever before. If that’s not spinning anonymous trash into golden data, I don’t know what is.

Do you have a question or comment?  Tweet at me or email me.

SHARE THIS ON: Twitter     LinkedIn     Facebook     Email

Hey, Media: Do You Know Who Is Consuming Your Content?

January 30th, 2017 Posted by Behavioral Data 0 thoughts on “Hey, Media: Do You Know Who Is Consuming Your Content?”

Fake news always seems to be in the news these days. Media companies are working themselves into a frenzy wondering why so many disaffected readers have turned towards click-bait headlines and conspiracy- monging websites, instead of opting for their own tried and true content.

So, how can a website hold onto an increasingly divided audience when there are so many other – more ideologically tailored — options to choose from?

I think the answer is simple. Mainstream news outlets have done enough solid reporting throughout the presidential campaign to earn the trust of the broader American electorate. But the content needs to be placed in front of the right pair of eyes. If The Washington Post wants its liberal readers to remain loyal subscribers, it should send them frequent updates on investigations and new allegations. If The Washington Post wants more Trump supporters to visit its website, it should place its vigorous reporting on the shortcomings of candidate Hillary Clinton front and center, but only when these particular readers visit the website.

By clustering their audience in a clever way, media companies can hold onto their readership – even grow it. And media consumers are rarely one dimensional. Once they’re in, they will move beyond the content that drew them in and they will check out other verticals (perhaps a cat video?).

It’s more important than ever for media companies to place an emphasis on targeting readers with the right content. This should begin with the use of sophisticated tools such as behavioral analytics, which allows a company to cluster its web visitors based on, for example, how they navigate content, how much time they spend on each story, if they read the entire store or scroll to the end, if they click to watch an embedded video, or if they skip or decide to sit and watch an entire ad before a video starts. Understanding these kinds of behaviors helps media companies to serve each web visitor the most relevant and personalized content.

Netflix did this brilliantly, and guess what happened? They are no longer remembered for their “Be Kind, Rewind” days. They rose overnight to become one of the preeminent media companies of our day. And Youtube gives video recommendations without you even knowing that they are recommendations.

It’s time for newspapers and other traditional media companies to shed their old ways. When we get people on both sides of the political divide to trust real news, and warm up to the stories that may question their world view, we can start having constructive debates over the future of our country.

IoT and Behavioral Analytics: A Perfect Marriage of Big Data

January 18th, 2017 Posted by Behavioral Data 0 thoughts on “IoT and Behavioral Analytics: A Perfect Marriage of Big Data”

Big data is about to get even bigger. As the Internet of Things (IoT) grows as we connect everything from our cars to our FitBits, inventory pallets to coordinated networks, so does the need for sophisticated data analytics processes like behavioral analytics.

First, it’s worth noting that IoT places the onus on dynamic analysis. Think about what it is that we like so much about self-driving cars. They process information instantaneously to produce the most efficient, and safest, outcome possible for the rider.

The days of gathering data to inform next quarter’s business decisions are over. Today companies need a direct input analysis and output vector in order to achieve business outcomes with the least amount of digital touches required from the customer. After all, customers are expecting instant satisfaction, or else they’ll move on to anther site as quickly as they came.

Additionally, machines today are better at predicting not only large-scale outcomes – like the state of the traffic on your local interstate – but also individual human behavior. That’s what a team of MIT professors proved last year in an experiment that compared how a computer system fared in creating predictive algorithms for an unfamiliar dataset. The computer finished ahead of 615 human teams out of 906, and worked exponentially faster. It even produced better results in predicting human behavioral outcomes (such as dropout rates) by selecting more relevant data than its human competitors.

This is all good news, because not only will IoT make behavioral analytics processes stronger by increasing the data pool by several orders of magnitude. It will also make it more valuable.

Take FitBit, for example. After the company has gathered ten years of data on millions of users, imagine what a sophisticated algorithm will be able to infer from your heart rate, monitored hour by hour, day after day. FitBit will have enough information to create advanced counterpart identification models to not only diagnose a user’s health problems, but also estimate what they are at risk for. Another example, is at Tesla Motors, where acquiring 1 billion miles of customer driving behavioral data will allow engineers to design and feed data into smarter, autonomous vehicles.

That’s the beauty of behavioral analytics, synced to IoT. It should make your heart race with excitement.

Trump Beat the Pollsters, So Here’s What You Can Learn From It

December 21st, 2016 Posted by Behavioral Data 0 thoughts on “Trump Beat the Pollsters, So Here’s What You Can Learn From It”

After Donald Trump was elected president last month, many pollsters had to eat their words. In fact, one prominent expert, Princeton professor Sam Wang, went on live TV to follow through on a pledge he had made before the election: that he would eat a bug if Trump won. “After all, I was wrong. A lot of people were wrong,” said Wang, before munching on a cricket.

All bets aside, the election results were a jarring reminder of the limitations of polling as a science. Polling is a small numbers game. In order to predict the outcome of an election, pollsters divide up voters into demographic categories (age, geographic location, ethnicity, socioeconomic status, and so on). Based on responses from a small subset of people in each group, they infer what the population at large is going to decide.

In the era of big data, that sounds a little backwards.

If your company is basing its online sales strategies on similar calculations, it’s got it all wrong. The best way to determine the outcome of a consumer is to watch what he or she does across different devices, and across time, and map that onto past observed behaviors. The more data you have on prior customer behaviors and their outcomes, the more accurate your next prediction will be.

This type of demographics-agnostic calculus is what’s called behavioral analytics. With it, you’ll never have to eat a bug on live TV.

 

Copyright 2018 SYNTASA®. All Rights Reserved.