Amanda: Welcome to the webcast.
I wanted to introduce our presenters. We have Amos Schwartzfarb presenting from Black Locus. We have Dhruv Bansal presenting from Infochimps and we have Tim Gasper presenting from Infochimps.
Today, we're going cover high-speed retail analytics, courtesy of a new approach of big data. Just a few things of housekeeping before we get started. We'd love to have your participation. Please feel free to raise your hand on the GoToWebinar doc. We may not answer your questions during the event but we will definitely have a Q&A session after the event, we will try to address as many as we can there.
We will be recording the event today so if you need to get off, we will send you the recording and a few days via email so look for it there. One more thing, we have a Twitter hashtag if you're joining us and live on Twitter, it's #bigdataretail. We will look for your comments in the twitter sphere today as well.
We'd like to open up with a poll, a quick poll just to get started. Our first poll today is now live on the screen. If you could take a moment to answer the question for us, we will show the results in just a moment after some votes come in. The votes are coming in.
The full question, we ran out of characters as you can see, the full question is, how are you leveraging all the data your organization collects from sources such as CRM, websites, social, POS, pricing intelligence, etc.?
We are going to close the poll in just a moment so if you haven't yet voted, go ahead and do so now. So the results of our poll are live on the screen as you can see. 40% of the audience today is not collecting data. 30% is collecting but not fully analyzing. 30% is analyzing but only historically.
What we'd love to see is more and more of you answering analyzing in real-time while predicting and forecasting in real-time and we hope that this webcast today addresses that. Without further ado, I'd like to turn it over to Tim Gasper, product manager for Infochimps to tell us a little bit about retail and big data.
Tim Gasper: Hello everyone, this is Tim Gasper as Amanda stated, I'm our product manager here at Infochimps. I think that poll is really interesting, it just goes to show that while a lot of us are doing some sort of small scale analytics in our organization, there are still a lot more work to be done in terms of making all that data accessible and some of these flags that I go to now are going to make some of that point.
First, to just establish, I mean, all of you are going to be pretty familiar with this entire concept of big data and the technologies that are behind it and it just goes to show that, as Gartner said, it's all about volume, velocity, and variety, and as a retail organization, we see this in particular because there is so many transactions that are happening.
Whether we're talking about people in-store or going online for our products. But big data isn't just about large volume, right? It’s particularly the variety piece that I think we need to make sure that we capture as well that you don't necessarily have to have had a byte of data in order to take advantage of big data technologies.
If you have data coming from online sources, mobile sources, social sources, your partners, your supply chain, your inventory, these are all different kinds of data models.
Yet, the best insights that you can clean, are going to be from combining all those sources together, and using technology that can unify that all.
I really like this quote that I saw on EMO Magazine which talked about how we're shifting from this concept of multi channel, which I think all of you are very familiar with, to more of an omni channel approach, and that it doesn't matter how customers are interacting with their business.
They're going to be touching every possible information channel they can. They're going to be bringing up their mobile phone and looking at reviews of a product right when they walk in the door.
We're going to experience the showrooming phenomenon where people come into our store, check it out, get that experience of talking to a person, and then go on to Amazon and buy that product.
The best way that we as retailers can really take advantage of all these different information sources and make customers stay within our boundaries is to take advantage of data as much as possible.
This is going to sort of beating a dead horse here but some of these staff I think are really interesting and I'd love to really bring up the numbers where over 50% of customers are making multi channel purchases.
Over 40% of people are using their phone and are purchasing directly off their phone, and social has really added another dimension where now, if we're not penetrating and getting into the peers of the customers that are coming into our doors, then we're really losing an opportunity to close the sale because over 80% of consumers are going to trust their peer recommendations over everything else.
Finally, this just ties it all together, this graph shows how today, a combination of buying online and being influenced online is over half of the purchasing decision and it's really taking the "e" out of e-commerce and making online and offline the all-one experience.
The insights that you're getting from your website, and the insights that you're getting from people walking through your doors, into your brick and mortar location are really going to tell you about the unified view of what that customer is. You cannot keep them separate anymore.
To wrap this all off, and kind of complete this introduction before we pass it off to our chief scientist, as well as Amos over from Black Locus.
I want to bring up the different types of retail analytics that your organization may be trying to achieve and as you can see, some of this has to do more with being able to segment, and target, and market to our customers better.
Some analytics that you may be trying to use your data to do is going to be more around price, optimizing your supply chain as well as your inventory, or you may just be trying to learn more about your customer and trying to monetize the activity that your customers are interacting with you with.
That might be, market [inaudible 08:06] staff analysis. It might be customer ROI and what the lifetime value of a customer that comes in through my doors, and how can I figure out customer satisfaction?
They're interacting with me in forums, online. I need to tap into all that information and understand what are they saying about my brand? Not only can I sell to them, but how can I improve my products and sell it better tomorrow?
Just to finalize that point, with the market basket and shopping cart analysis, it used to be completely separate, the offline world from the online world but now, the things, you're learning from people hitting your websites are exactly the things that are going to impact the foot traffic coming through your doors in your brick and mortar store.
Unifying this all is really the key here and that's why it's all one big deal, it's all unified.
Why aren't more organizations collecting, unifying, and analyzing all this data? I think, as that poll suggests, there are many challenges, three of which I think are the most important ones are going to be staff, expertise, and time.
It's hard to find individuals who are really good at what we would maybe call data science, who can really drill into the data, mine it, explore it, and come out with insights that you never thought possible that have that expertise.
Some of these new tools like the Hadoop, which is probably a word you've heard of, if not very familiar with. These people are difficult to find, and ranting yourself up with the training to do that is not simple.
Finally there's no time. Competitors are adopting these technologies and hiring on these people. How can I achieve this market basket analysis? This predictive real-time pricing intelligence? Tomorrow, not a year from now, not two years from now?
I'm going to pass it off now to Amos, the VP of customer development over at Black Locus and he's going to talk about his application, his to get application in their service and how they can really help you overcome some of these issues.
Amos Schwartzfarb: Thanks Tim. What we do, is collect a massive amount of data from the internet and the way that we do that is that we have some proprietary machine learning technology that is very good at understanding what is and isn't a match.
The way that we work with your customers is they'll give us their entire product catalog, which could be hundreds of thousands or more SKUs and we basically break down every single product into hundreds of attributes and then crawl the web looking for different combinations of those attributes. Then we pull that back into our platform and do some analysis on top of that.
As you can imagine, that's an enormous amount of data and it's very hard to figure out what to do next. Part of what we do in that is to give you the ability to prioritize and act by giving you a simple set of rules on top of this enormous data so that you know the right product, the right price, and the right time.
As you can see from that quote that was up on the screen, we're able to provide an enormous amount of ROI to our customers and from an ROI 75 company, they saw their best order ever almost immediately following starting to work with us and they've only seen improvement since then.
Amanda: Now we would like to watch another poll. This poll is, "How are you currently acquiring prices assortment intelligence? There are some answers on the screen, we'll give it just a few minutes. Everyone place your votes and we will show the results in just a moment.
We have about 35 of you that have voted, 35%. If you want to go ahead and place your votes now we're going to show the results in just a moment. A few more votes coming in. Last call for voting. Closing the poll, we'll see how you guys answered.
So 29% of you go manually out to the web, 14% are using a third-party software vendor, 14% are using a consultancy firm and 43% just doesn't know or is using their gut sense for pricing.
Amos: Yes, that's not an unfamiliar response to sort of what we're hearing in the space too and that's where it gets really exciting to talk to some of you and how we can help you get a really deep, rich insight into what's actually happening, who your competitors are, how they're pricing their products relative to your products, what their assortment is relative to your assortment, and I think that's a really important piece to understand.
Sort of what we're able to bring to the table for you which is not just the competitive price on your products but who your competitors are an aggregate as well as the category level, the brand level, the product level, what their assortment looks like, what you're carrying that they're not carrying, vice versa. That's one piece of it.
As we look broader across just the opportunity of figuring out how you price the when, meaning, how you price the right product at the right time for the right person. There are a lot of different elements that we need to bring into the picture.
It's not just about competitor pricing, it's not just about assortment, it's also about seasonality, it's about social sentiment, inventory levels, brand equity, all the things that you see on the screen.
It becomes really, really important at this point to have a platform that allows you to capture and very quickly understand and act upon the data. What this slide is sort of talking about here is a little bit about, who we are and how our business is trying to transform your businesses and provide even greater ROI to you so what we do today is very simple.
We provide you pricing intelligence on your entire competitive plan landscape. Over the course of the rest of this year, we're going to start to do more prescriptive-type pricing and our bigger vision of a company which we are starting to deliver on with some [inaudible 15:33] customers is to also help you actually price your items.
Just to be clear about that, that doesn't mean to be the lowest priced item, that means to figure out again, the right price at the right time with the right products so that you are maximizing your revenues for your company.
I think this slide is sort of a really good picture of really everything that comes into the picture here. It's not just about the inside analysis. It's not just about some software development that can kind of get at the problem. It's not just about this really great machine learning technology that we have that's good at matching things.
It's about putting all these things together and finding that piece in the middle, as this chart shows, the data science in it all. In order to do that, you have to have a really robust platform that allows you to get to that.
Just to get a little more tactical on what that looks like for us, it's a long process that happens very, very quickly. It's about going out to the web, and collecting ton and tons of information.
It's about extracting from that, what the important pieces of information and adjusting that into our system, and our system has to learn from that so that it can start to analyze it and then deliver something back to you. This has to happen very quickly so that you're able to act and react to what's happening in the marketplace.
I think that this is a good transition and a good opportunity for me to introduce Dhruv to talk about how Infochimps can really help to support the overall platform and to be able to deliver this.
Dhruv Bansal: Thank you Amos. Hi everyone. My name is Dhruv, I'm one of the founders, I'm also the Chief Science Officer at Infochimps. At Infochimps we make big data infrastructure simple.
We're proud to knit together solutions for our customers with best to brief big data technologies that we've researched and become experts at using and deploying so that our customers don't have to.
We're very happy to be working with Black Locus. We have a lot of customers spread across a number of industries, everything from mobile advertising to social media listening, couponing space and of course retailing as well.
We're going to take a moment to do a quick poll to get a sense of how everybody feels about big data.
Amanda: This last poll is, "What are your top priorities around analytics?"
Today's webcast covered high-speed analytics in the retail space. I'm just trying to gauge from you all, what are the top priorities around analytics for you can the organization?
Our voting is live, please place your votes. We'll give it just a minute and then we'll come back and share. OK, a lot of votes have come in, this will be last call for votes. If you have not yet voted, go ahead and place your votes. Few more coming in. We will close the poll and look at the results.
So 29% of you are looking to improve your effectiveness of your marketing analytics. Nobody is looking to enhance their pricing and assortment intelligence. I have to say, just my opinion, you have a lot of ground to gain there. 14% of you are looking to get more predictive with your inventory and supply chain management. 43% is looking to be better understand the customer base. 14% have other priorities with analytics.
I'd be interested in knowing what those are, if you feel like chatting them over to us, go ahead and share at the end of the webcast.
Tim: Everyone feel free to toss any questions on to the side. We're looking forward to answering them at the end to getting more information or dive deeper.
Amanda: And back to Dhruv.
Dhruv: We ask about use-cases because there are so many questions to ask about in the world of retail. Especially when you combine the various kinds of data sources that are available to our retail enterprises.
The Infochimps platform enables to get application like providing like providing some of the underlying technology infrastructure that drives these solutions to a lot of these types of questions. Everything from analytics and applications of workflows, all the way around to various use cases that are possible.
Of course for Black Locus, who concentrates specifically on the underlying pricing and assortment analytics, Infochimp is part of the solution that applies their technology as well.
I want to dig in just a little bit into the underlying structure of what Infochimps solution looks like. In almost every case we're starting out with some real sources of data. This is everything from internal ELP systems, point of sale, foot stream from your webs, rating review data, mobile data, social media data, anything.
One of the first things you need to do with data is adjust it, so the Infochimps platform comes with an ingestion layer that helps you collect and analyze data in real-time from all these various sources, modify and normalize that data in real-time, or else run predictive analytics.
If you have heard of tools like Hadoop, the platform is well-equipped with all those next generation big data processing technologies. Then of course, we have stored that data in databases so that we can ask questions and build applications.
Specifically, for Black Locus, an example of how they use our platform is we start with prices that are available across the web, as well as internal customer data that can be sourced with various cluster modules.
The Infochimps platform goes and collects that data, it's then passed into an analytics layer that's then ultimately the output of which is thrown into the database, which provides Black Locus' own customers with some of the pricing information they're using to make decisions.
Ultimately, we provide a series of benefits to folks who are interested in getting into big data in the retail space. First and foremost, we're an end-end solution. Big data is a new technology.
Talent that understands how to use it is rare, and we provide you with a complete and flexible big data foundation that you can then build upon.
With yourselves as your outsource big data partner in this regard, we're trying to get you to become as effective as possible, generating the insights that you need, quickly, without any sort of capital investment because we're a cloud based solution, without investing in new talents since we manage the entire operation for you, and tailored for your business.
Our professional services are happy to write customized solutions that connect to the data sources that you need to care about.
Going back to... I’m sorry.
Tim: Just to tie it back to sort of those different analytics use cases that you saw, using different big data technologies is going to provide you that technology brainwork that's going to make a lot of those different use cases possible from the perspective of having the infrastructure and the tools available to tackle those problems.
Solutions and companies like Black Locus, are really going to provide that deep insight, that application layer that's going to provide the specific insights you need to accomplish a use case, right?
Particularly when it comes to pricing and assortment, Black Locus is an expert there, and if you're looking to tackle some of these other use cases, Infochimps is happy to figure out what sorts of tools and technologies are going to help you get there.
At this point, we're going to look at some questions, right Amanda?
Amanda: Yes, we have some questions from the audience. If you have other questions at this time, please feel free to chat them in.
Here's a question that came in from Bob in San Francisco. "My company does not have a central pricing function. Do we need one in order to maximize the value of the price and assortment intelligence?" Amos, you want to answer this one?
Amos: Yes, sure Bob, that's a great question and it's something that we hear all the time.
I think what's really great about solutions, both of Infochimps and Black Locus, is that we're able to, and this is sort of Dhruv's point as well, we're able to become that function for you and in many cases we're helping to either help you figure out or execute on what your strategy is so we can become your centralized pricing function for you, with your direction of course.
Amanda: We have another question here. Tim, I think this one might be best for you. The question is, "What resources do I need to have on staff in terms of analytic programmers for big data?"
Tim: Yes, I think that was a good question and maybe I'll start this one and even let Dhruv finish it if he has some thoughts.
In order to accomplish a lot of these different analytics use cases, you may have some analysts on staff already, which may be using Excel, or some sequel databases and things like that to ask their questions, and these types of individuals are going to be able to interact well with something that you get from either Black Locus or Infochimps because we're trying to make that the application workflows as well as the technology a lot more easy and managed, so that way you guys don't have to worry about that.
Obviously if you have programmers that want to get up to speed on these technologies, they can do that as well with these things. Especially with Infochimps, you have complete access to the underlying servers and databases.
The cool thing though is that you don't have to necessarily have domain specific knowledge to do things like Hadoop or machine learning in order to gain some of the insights that those things can provide. Infochimps tries to simplify those things into programming languages that more developers are familiar with such as Ruby.
Amanda: A very similar question came in from Minneapolis here. Amos, I believe this one would be for you. "Do we need to have a robust infrastructure and technical team in order to support price and assortment intelligence?"
Amos: Another really good question and the short answer is no and the longer answer is, that's exactly where Black Locus and Infochimps can help you out.
The infrastructure is basically supported by us and the team, you already have a team in place and our role is to help your team maximize the value of the data that we're pulling off the web and analyzing for you based on your business rules.
Amanda: One more question for you, Amos, this looks like it came in from Omaha. "How long does it take to get set up and how often will I receive updated intelligence?"
Amos: Another good question. Set-up for something like this is fairly quick, although, it's relative by the number of products that we're monitoring for you, and so it could be anywhere from a couple of days to a week or so and you're up and running and starting to collect information and in terms of the frequency of updating, we can return frequency very, very fast.
In other words, you can get updates daily, you can get updates more frequently than daily but most of our customers, because they're just learning their way into how to look at, use, and analyze the data, we tend to recommend taking your highest velocity skews and looking at those more frequently, like daily, and your lower velocity skews and monitoring them weekly as you learn how to maximize the value of the data, and then over time, you can transition to a faster velocity for everything.
Amanda: Thank you Amos. Another question just came in. "What enabled high-speed analytics versus normal analytics?"
Tim: This is Tim from Infochimps.
I think that's a really great question because I think that boils down to what we're trying to accomplish with the entire topic of this is that, there are many stages upon which your analytics can evolve and improve and get you closer and closer to being predictive and forecasting the future and I think the first step, obviously, is to just collect your data.
The second step is, let's start doing historical analysis. Look at the data you already have and try to find some trends in that. When things really get to be high-speed is when you can take advantage of more real-time technologies that can look at things in a more streaming fashion.
When a tweet gets tweeted, can I analyze that immediately after it happens? If a stock price becomes immediately available, can I have access to that? If a product gets sold, can I immediately see that and include that in my analysis?
That's where high-speed analytics really comes in, is being able to take action on data immediately, and then use that to start making predictions about the future and enhance our models so that we're actually training our systems to get smarter and smarter over time.
Amanda: We have one last question, "What does an Infochimps engagement look like?"
Tim: Great question. An Infochimps engagement is pretty much split into three parts, I'd say. There's more of an implementation aspect where we go in and help set you up with the tools that you need in order to get started with big data and doing some of this analysis.
The next step after that is a subscription model where we maintain and help you keep those systems up and running and make sure that those systems accessible to you in a way that you need to use them. The third aspect is, we have some professional services, so whether that's helping you write your algorithms or if it's going in and helping you get some best practices and methodology, we're always happy to that as well so, maybe Amos would be interested in talking about what a Black Locus engagement would be like.
Amos: Sure, a Black Locus engagement looks something like this, where we will work with your team very quickly to figure out how to ingest your product catalog.
Once we have that, most of the work, or actually all of the work is on our side and then, as I mentioned before, within a couple of days we will deliver to you an actionable account, at which time our success managers will work with all the necessary members of your team to get everyone trained to create the different use cases within your organization.
Then from there, there's a process around monthly and quarterly follow-ups to ensure that you're gaining the most value from this system and that we're also learning and adapting your strategy within this system as it evolves.
Amanda: There are no other questions coming in. Last call for questions. If you have one, send it now. At this time, I'd like to thank Amos for coming over and joining us for the webcast. Thank you so much Amos.
Amos: My pleasure.
Amanda: Dhruv, I appreciate you being here. Tim, thank you so much.
Tim: No problem.
Amanda: The next time we hold this event, I'd love to see that poll with more and more of you answering towards using high-speed analytics in real-time and in looking into the future of making predictive decisions, that would be great.
Thanks you so much. We'll see you soon.