Big Data Analytics Key Takeaways from DGS4

Big-Data-Analytics-Panel-Group-Photo.jpg

[su_note note_color="#e3f0ff" text_color="#000000"]

Thank you to Attunix for sponsoring this panel at the Digital Growth Summit.

[/su_note]

Watch a replay of the Big Data Analytics Panelists at Digital Growth Summit:

[su_youtube_advanced url="https://www.youtube.com/watch?v=4HwXsy-7bqg" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"] Jesse Leimgruber, Moderator, NeoReach - Tyler Malin, Exoteric - Joael Harbi, CBS Interactive - Pekka Kostamaa, Teradata


Big Data learnings and tools

What can you learn from Big Data and what tools are accessible?

Google Trends can give you a prediction of keywords' popularity for the next 12 months. - @JesseRank

I love Google Trends. You can go to Google Trends and enter any keyword that you want to research and it will tell you predictively if that trend is going to go up or down over the next 12 months. In the same vein, there is a site called Majestic.com, where you can type in any website and they crawl the whole Internet to show you every other single website that links to that page. You can find people who have talked about you or your company who you never would have guessed that they were interested in what you were doing. Obviously, you cannot crawl the whole Internet on your own computer so these tools are really important. - Jesse Leimgruber

Many big companies go through phases when they start collecting data. First, they collect the data and they look at what happened. For example, sales were down yesterday. You learn to look at the data and it's more traditional recording stuff at first. - Pekka Kostamaa

Big Data analytics not only can help companies to analyze past events, but also to make predictions. - @PekkaK

When you move up to the next level, it becomes more real time. You start collecting information and seeing what is happening right now, so you can take action directly on that. You figure out maybe the product placement is different on store A than on store B and that is why the first store is selling better so you call the store manager and you say, "Move your product. Put it in a different place." - Pekka Kostamaa

The third level is predicting and that is the big thing right now, trying to predict what will happen next. An example that happened with one of our customers, Wal-Mart, when there was the hurricane coming towards Florida a few years ago, they actually cornered the market on American flags and Pop-Tarts. They figured out that when a hurricane comes in, that people want to stock up on Pop-Tarts because those can last a long time and American flags, for whatever reason, were also wanted. They did the analytics. K-Mart also realized this, but by the time that they did, there were not any American flags left to be bought. Wal-Mart had cornered the market so they had a big market advantage based on their Big Data analytics. - Pekka Kostamaa


[su_youtube_advanced url="https://www.youtube.com/watch?v=KWPxKADXukw" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]

Using Big Data for ROI

How do Big Data analytics apply to your marketing strategy?

Make sure that your traffic is drawing in customers that are actually going to make purchases. - @TylerMalin

Being able to optimize for ROI is critical, and determining what your goals are for marketing is also important. It seems simple, but in practice it isn't. You have to look at these APIs and analytics, and make sure that they drive sales. Figure out who your target customer is, where they are living, how to talk to them and how to optimize traffic for these customers. You need to make sure that your traffic is drawing in customers that are actually going to make purchases. - Tyler Malin


[su_youtube_advanced url="https://www.youtube.com/watch?v=5-snAnlVSrc" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]

Big Data marketing tools

What tools can be used to harness Big Data and marketing power at an individual level for start-ups, new businesses and small business owners?

Make sure that you ask your customer to do what you want them to. - @TylerMalin

In almost every organization, there is already an existing set up tool that people have chosen for one reason or another. Sometimes these tools are great decisions and sometimes they are not. One simple example would be that there are three inexpensive web page builders that have massive differences between call-to-actions, color and placement. People think that these are trivial things on a website, but they actually really drive significantly different action. Again, I think that making sure that goals are tracked through and that you are really asking your customer to do what you want them to do is crucial. - Tyler Malin

Data should be accurate, it needs to be right in order to ensure that you are not inflating the data. - Joael Harbi

One of the robust tools in my opinion is Adobe. Their suite of products is great. There are so many things that you can do with it, it is a beast of a machine. Google Analytics is the free option for this and can be used quite well for smaller companies. You need your data to be accurate, it needs to be right in order to ensure that you are not inflating the data, which is why I love Adobe because we can do that with their suite of products. Once you have all of this set up and you have your team members looking at each aspect of the tracking for functionality, then you can optimize and start to target things in a really specific way. - Joael Harbi


[su_youtube_advanced url="https://www.youtube.com/watch?v=8ACwZoXhwWo" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]

Big Data tracking

How do you track all of this data? How do you install the tracking and analytics to get this data in a way that doesn't mess with the data and ruin the data?

Without proper tracking and collecting, you won't have any data to do anything with. There are kind of two sides to this that are a bit tricky. You can literally track anything, from tab management solutions to placing tracking pixels on every aspect of the page, but what does that really do apart from inflating your server costs and driving everyone crazy? There probably isn't enough manpower to track everything, so you have to ask yourself what matters. - Joael Harbi

Without proper tracking and collecting, you won’t have any data to work with. - Joael Harbi

Making the correct decision comes with experience: you develop an eye for data and start to see what works and what doesn't. Still, trends shift and the way that you track your data will most likely change over time as different things become important. The reason tracking is so important is because once you understand what you need to track, then you can track it. For example, I work with video at CBS Interactive and our biggest API right now is video consumption because we want people to consume all kinds of videos and ads. - Joael Harbi

Another example: I was working at an ecommerce company that wanted to sell apparel, so we were looking to collect data on traffic to those specific inventory items. Once you determine what it is that you are really after, then you can begin the process of saying, "Hey, I want to put this tracking pixel here and this is what it is going to be mean." - Joael Harbi

It’s very easy to be tracking the wrong things and then to not have your analytics set up properly. - @JesseRank

It's very easy to be tracking the wrong things and then to not have your analytics set up properly. I can't tell you how many times I have seen analytics suites that are firing too many times or are not on every page on the website and people are missing certain pieces of data, which messes up the data you are collecting. At the core, it is so critical to make sure that whatever data you are tracking, it is accurate and clean and is set up properly. - Jesse Leimgruber


[su_youtube_advanced url="https://www.youtube.com/watch?v=3Mh_gqn7YV4" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]

How large corporations use Big Data

The end goal of Big Data is to help with business intelligence and lead to smarter decisions. What are some large corporations doing right now with Big Data? How are these large corporations using Big Data?

One example of a large company using Big Data would be for path analytics on your website. One of our clients wanted to know, "Why are people cancelling their accounts?" So, the first thing that we did was to look at the cancelling account and the overall customer value. Maybe the account isn't worth keeping, because, for instance the account runs at a loss because they utilize the call center too many times. - Pekka Kostamaa

All data about a customer’s path needs to be tracked together, including their online and offline experiences. - @PekkaK

So, our clients can look at the customer’s path and see what led to their cancellation. This might include multiple sources and not just their web interactions, because maybe they called the call center or visited the branch in person. All of this data needs to be tracked together so that there is a path from each customer and you can identify the common path. - Pekka Kostamaa

One of the things that they actually found out was a very trivial thing, that a lot of the customers cancelled the account after they physically went to the branch and asked for a refund of a fee. If the bank denied the refund of that fee, they would cancel right away. The fee was only $10, but the customer value was more than $10 so it made sense to give the customer that refund because they were worth more. If the fee is like $100 and the customer value is less than $100 then it wouldn't be profitable and the company wouldn't want to make the change. These are very advanced analytics and once you figure out where the path leads, you can decide what to do about it. - Pekka Kostamaa


[su_youtube_advanced url="https://www.youtube.com/watch?v=BWTsyCTQueQ" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]

Collecting and organizing Big Data

How do clients actually get the data because a lot of it is so silo-ed, it's not like the Google Analytics suite that captures everything, so how do clients make these associations with their data?

That is the big question. We used to say that people should put everything into a single data warehouse, but the data volumes are getting so big these days that typically it lands into a data lake. If you do not clean it and wrangle it, it can quickly become a data swamp. - Pekka Kostamaa

You have to be very careful that you keep the data clean. - @PekkaK

You have to be very careful that you keep the data clean. You can have multiple sources load it into the lake, but you have to be able to analyze it in a single tool so there is this discovery process. This is where professional data systems and professional data specialists can help. You have to know what is there and what makes statistically good sense, how to use tools like sentiment analysis against text, market segmentation, collaborative filtering, etc. - Pekka Kostamaa

You have to come up with the application and then you operationalization it. The operation phase requires high tech applications. We had one application that we really liked that sent ads based on customer background. We used this to send ads to customers that were withdrawing funds from an ATM so it needed to be quick and we needed to figure out what to send them quickly. - Pekka Kostamaa


[su_youtube_advanced url="https://www.youtube.com/watch?v=pn_25EXYMOo" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]

Big Data trends

What insights do you think we will eventually be able to gather? Where do you see the Big Data space going?

Proper analysis and implementation of data can give you a marketing advantage over your competitors. - @TylerMalin

I think, in general, a culture of curiosity and incorporating Big Data and analytics is going to be the differentiator between successful companies and non-successful companies. The amount of marketing advantage you are able to achieve through properly analyzing and implementing what you learn from Big Data is going to very quickly make it difficult to compete with other companies that are doing it better. I think that is something that we will start to see very quickly, in any sector. If you want to be successful, you have to start incorporating these things now. These are areas where we can start to train and analyze and make big differences hopefully. - Tyler Malin

Solving big problems is really key, such as solving resource allocation on a mass scale. It's an issue for any one human to wrap their head around the massive resource allocation of the planet and Big Data is something that I think is going to go there and help solve some of these problems. - Jesse Leimgruber

Based on the type of content that people consume, their behavior on the site or the app changes. - Joael Harbi

I think, touching on the issue of no black boxes, because then you have this exposure as to what is good and what isn't and then you can make business decisions, life decisions, global decisions based on things that have previously worked and not worked. Definitely the no black box era, that is quickly going to lead everyone out that is still kind of living in that black box era. - Joael Harbi

The other thing that I am seeing a lot of is content. I see that people are really interested in content, whether it is mindless celebrity gossip, or a really important piece of news, people like looking at content, engaging with content, and sharing content. What we have seen is that based on the type of content that people consume, their behavior on the site, the app, or the connected device changes. For instance, Big Data allows us to know data like that people who are reading articles about the TV show “Big Brother” are more likely to consume videos all the way through. Or, people who are taking quizzes tend to have a different behavior, and then you can analyze what people like in order to predict and target content for that person. - Joael Harbi

I also think that people are changing their traditional media consumption devices, and are going on to mobile devices, tablets and OTT devices such as Apple TVs, Fire TVs and Rokus, etc. We are now in a very robust way tracking user behavior on these OTT devices and trying to understand what people do. For a lot of people, that is their main means of consuming entertainment. Many people are pulling the plug on cable providers. Where Nielson has kind of been doing it in a black box way in the past, and completely monopolizing that industry, now it's kind of going to be tough to do that because people can track media consumption in many very accessible ways. I can tell you what has been consumed on Apple TV this morning so it's a bit scary, but cool. - Joael Harbi

Everything is being tracked now. - @JesseRank

Everything is being tracked now. I love the analogy you always overestimate the change that is going to happen in one year and underestimate the change that is going to happen in ten years. If you look back to a year ago, we had Apple watches and such, but ten years ago we didn't even have iPhones. I think that analogy applies really well to Big Data because ten years ago, Big Data didn't even exist. So next year, it probably won't be that much different, but in ten years I think it will be dramatically different. - Jesse Leimgruber

I see the next big wave as being Internet of things and everything will be on the web. - @PekkaK

I see the next big wave as being IoT for the Internet of things and everything will be on the web. We can only imagine what kind of applications that is going to bring. So one scenario that we have been throwing around in our development lab is the idea that the trashcan, it will know what you are throwing away and your refrigerator knows what is in there. So, okay, he just threw his egg carton away. Now we are out of eggs now so your refrigerator can send a message to the store saying "we are out of eggs" and now an Amazon store drone will drop eggs to your front door. So the whole interaction, the whole paradigm of how we are going to live and work is going to change. The cars that drive by themselves, maybe nobody will have a need for a car anymore in their garage because you pick up an Uber app and the car shows up in two minutes at your door and drives you automatically where you need to go so we don't need to have a car per person. We need to have a car per ten or twenty or thirty people so we can help the environment that way. I think there are a lot of possibilities and opportunities with this new Internet. I don't think we really know where it is going to take us in ten years, but there are a lot of exciting changes that we can see possibly happening. - Pekka Kostamaa


[su_youtube_advanced url="https://www.youtube.com/watch?v=xrfzdDXXUHQ" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]

Big Data visualization

Audience Question: How does data visualization play a role in all of your individual jobs and whether unstructured data needs to be visualized to make better decisions?

I love data visualization. Data is huge and people can have a difficult time processing it quickly. - @TylerMalin

I love data visualization. Because data is huge, the scale is massive, people can have a difficult time processing it quickly. Dashboards can be helpful in that way if they are properly created. In general, when you take a business's data and all those silos that are tracking what certain people need to know, you will usually never be able to find two people that will interpret the visualization the same way. There are a bunch of tools that can be used for data visualization and on the open source side, there is a lot of innovation going on right now in the space. From that perspective, it is about actually spending time, the data team, making sure that you are dealing with the decision makers and that they are actually able to make decisions based on the Big Data. - Tyler Malin

We actually have a data visualization group because, like you said, everyone interprets things differently. I've done this -- it's very interesting to go up to two intelligent people and ask them about the data to see their differing takes on it. - Joael Harbi

The goal of data visualization is to make things as simple as you can. - Joael Harbi

In my mind, it is very clear as well, but everyone is going to approach things a little differently, so the way that I like to approach things is from a standpoint of making things as simple as you can. - Joael Harbi

I know that a lot of people on my team are using Tableau, where they are building a lot of in-house stuff and we have some really awesome Python coders that are building these pipelines and are creating awesome visuals. It's important to remember though that you can have so many pretty visualizations, but you need to know what they are saying. - Joael Harbi

You have to ask yourself the question "Who cares?" so you can figure out why it matters. If you can't answer that question, then you can't move forward and that is where I think the data visualization comes into play because you can create the coolest looking things but if no one is going to use them, then it doesn't matter. - Joael Harbi

In my organization specifically, it is very data heavy and people are using these things. We decide what to create based on what we have seen before internally, obviously right now a huge push on social and content type pieces, a huge push on anything that is failing, meaning is an app crashing or is a video not loading. We want to better understand those fail rates and see those things in one view. - Joael Harbi

There are many ways to look at the data and we have to make sure that there is something of value there. - @PekkaK

We don't build our own visualization tools; instead, but we partner with different systems because visualization is so important but difficult to do. There are so many ways to look at the data and we have to make sure that there is something of value there. - Pekka Kostamaa

Take Google Maps for example. If you look at the traffic information, you see right away where it is red and this indicates which freeways are busy. Think about that same data in tabular form. It would never be valuable in that way, it wouldn't be easy to access. - Pekka Kostamaa

There are so many data visualization tools that give you a picture of the social network and that would be impossible for you to see. Sure, you could plot it into an Excel spreadsheet and see the connections but it is just 0's and 1's. When you put it into a nice visualization, you are going to quickly see what your groups are and how that works. - Pekka Kostamaa

I think visualization is going to become more and more important as we rely more and more heavily on Big Data. It's like we say, a picture is worth a thousand words. The brain is very good at looking at pretty colors and shapes and more visual things. - Pekka Kostamaa


Audience Question: What are you doing on the predictive side to keep up with always evolving platforms?

A lot of thought goes into the website design process, months before it is even built, using predictive analysis. — Joael Harbi

We have an emerging platforms division. They work very closely with our business intelligence team, to go into the massive amounts of data that we have stored as well as our real-time data to build predictive analysis models. Take, for example, the Colbert site. We just redid the site and it was all based on predictive analysis from other pieces of Colbert content, and from other sites that resembled it, from user IDs that we were able to gather and looking specifically at the behavior of these people and what they wanted to see on the site. A lot of thought goes into this process, months before it is even built or created and the corporate design ended up being a great success because of that predictive analysis that we were able to do based on data that we had and real-time data. - Joael Harbi

Customization is really important in visualization. - @TylerMalin

Customization is really important in visualization. It's really very subjective. Building the ability for you to make decisions on how your want your data displayed is important and there is definitely an emerging language around visualizing data itself. - Tyler Malin

We know that a lot of people are using Waze and Google Maps so they all see that same information that the 405 is busy, so where are they going to reroute people? If I am going to be there ten minutes from now, where am I going to be then? That is predictive analytics. Trying to predict the best alternative routes is interesting and visualization can come in to help with these problems too. - Pekka Kostamaa

Someone was starting to look at how Waze rerouting through neighborhoods was affecting real estate value because it's moving people towards Brentwood now, and the housing values go down because there is too much traffic. - Tyler Malin


[su_youtube_advanced url="https://www.youtube.com/watch?v=pZoTYEPQ9C4" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]

Simplifying Big Data

For non-savvy clients: how can we visualize and simplify the data for clients who want the value but do not know how to get it right away?

Provide your data in a format that the client is already used to looking at. - @TylerMalin

I think looking at reporting analogous to what already is in use in the industry helps. Provide your data in a format that the client is already used to looking at. I think that interactives are our future, to be able to play with the data and plot things out is helpful and meaningful. I also think visualizing with video is important and allowing the data to show trends over time. There are some really interesting projects in spaces that you wouldn't necessarily think that people are using data and it's almost an art. There are visual artists who are most influential with coming up with the best ways to view this data. - Tyler Malin

Do little training sessions for people to help them understand what the metrics mean. - Joael Harbi

If you have a niche that doesn't have a very specific program to handle that data, I would look at a generic big data company like the Cloudera, or Splunk, because their expertise in taking unstructured data and finding the perfect visualization is spot on. The cool thing is, although these are not self-serve programs, most of them have self-serve tools, and you are able to put your data in and someone will get online with you and tell you what visualization they sense for your needs and they have hundreds of options. - Jesse Leimgruber

What I have found is that it's consistency that helps them recognize what is going on and doing little training sessions for people can also help them understand what the metrics mean so that the language isn't foreign to them. They know what a unique visitor means, for example, and they are not intimidated by it. - Joael Harbi


[su_youtube_advanced url="https://www.youtube.com/watch?v=5D_3La3rNxM" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]

Social impact on Big Data

Audience Question: Does the social impact change the path of the user, as opposed to just looking at the previous data that you are collecting on that user?

When you look at data on a mass scale, you can see how similar people actually are. - @JesseRank

What's crazy is that when you look at data on a mass scale, you can see how similar people actually are. We analyze social media celebrities so we see that there are influencers that have 100,000 followers, which is such a small fraction of the population but then those 100,000 followers also follow several other influencers with 100,000 followers and within this niche, within this community, the followers are very similar. They buy the same things, they use the same things, etc. It's difficult to say whether or not there is difference looking at the high data vs. looking at the individual data because they are used for different things, but it's amazing how when you have access to all the data trends start to emerge from it. - Jesse Leimgruber

You need to know what your customers did yesterday. If something happened, then what are they doing today. - @PekkaK

It's an interesting question. You need to have the data so you can start looking at it and analyzing it. You need to know what your customers did yesterday. If something happened, then what are they doing today. It is a learning process. You want to know what was the state of your customer base yesterday and now today what is it "more" of so when you update the data, you don't really want to throw it away, you want to compare. - Pekka Kostamaa


[su_youtube_advanced url="https://www.youtube.com/watch?v=b58bybFuiqE" width="840" height="460" controls="alt" showinfo="no" rel="no" fs="no" https="yes"]


Big Data Analytics Panel Digital Growth Summit