All companies have data, all companies have people: the secret to big data analytics is incorporating people into the overall process, according to speakers at GigaOM’s Structure Data.

Structure Data 2013 John Sotham BuildDirect Mohan Namboodiri Williams-Sonoma
photo: Albert Chau

Session Name: The Future Impacts Of Big Data Insights On Your Organization.


Announcer David Card Mohan Namboodiri John Sotham


Thank you for that, Rebecca. Right now we’re going to on to our next panel, our next discussion. David Card is the VP of Research at GigaOM Research. He’s going to be talking with Mohan Namboodiri. He is the VP of Customer Analytics at Williams-Sonoma, and John Sotham who is VP of Finance at BuildDirect. They’re going to be talking about the future impacts of big data insights on your organization. Please welcome our next discussion.


Good morning. Welcome back. Thanks for your attention. My name’s David Card, I’m the VP of Research for GigaOM and we’re going to have a discussion this morning about what big data does to the decision-making process and the organization within the company. We’ve got two who have done some deployments, they’re bringing in big data in real-time fashion and it has had an impact on both their organizations. What we hope to give you guys a feel for today is some takeaways – maybe some practical things, examples of how big data in real-time fashion, and analytics built on top of that has affected decision-making within an organization. In addition, what parts of the company are better-off or easier and which parts of the company take to that first? Does it kick-off in marketing and make its way into finance or something like that? Then, what does that do to the whole organization’s decision-making process and workflow, and org chart? This is kind of like the org-chart panel discussion, I hope. So we’ll start off with some introductions and then get right into the meat of the discussion. Mohan, tell me who you are and what you do at Williams-Sonoma.


I’m the Vice President of Customer Analytics, which means we look after a very broad range of analytic applications from predictive modeling to certain kinds of triggered [inaudible] marketing and email, to the retail store. We’ve had a very interesting and important big data use case around marketing attribution.


Marketing attribution. We’ll get into that in a bit. John, tell us what you do for BuildDirect.


Sure thank you. Pleasure to be here today. My name’s John Sotham. I’m with BuildDirect technologies; I’m their Vice President of Finance. I probably want to give myself a promotion here and say I’m the CFO but we will leave it at VP Finance. For BuildDirect: we’re an online retailer and wholesaler of building materials and I think big data is really important to us. It’s allowed us to really lean out our supply chain and really differentiate ourselves from the competition to the point that we’ve got suppliers willing to put consignment inventory in our model. The whole predictive analytics – that’s the secret sauce, that’s what drives the success at BuildDirect.


Why don’t we start with that? Talk a little bit about how you have taken insights from your analytics process, and what has that done to your supply chain or your – I don’t know whether it’s pricing or inventory management… What kinds of things have been affected by this?


The truth of it is, it’s broadly across the company. Everything from merchandising, what products we select to put on our web store, to how we merchandise them, and how we market them. The real benefit to the supplier and this is kind of the operational side that ultimately leads into the customer experience which is to be able to buy product from us, have it delivered to their doorstep at 40-60% less than you would see with the competition, really is the suppliers doing business in a new way. This is putting inventory and consignment in our model, and why would they do that? It’s really the predictive analytics around the data we share with them and the insight we can give them – early insight into which products are going to be winners, losers. It allows them to optimize their resource allocation and production planning to really better– take the supply and match it up with the demand so it’s that whole predictive piece that has allowed us to really lean out the supply chain. I think that’s the key piece for us.


If you don’t mind spending a little more time on this, is there a before and after picture of that? What changed when you started doing the predictive analytics? Would you get more inventory, would you get more real-time inventory from your supply chain? What kind of things shifted a bit?


Well I think just really the value, if I’m understanding the question properly. It’s really the value to the supplier. We’re no longer just a new access to market. When you look at BuildDirect from the highest level you say, Its a new channel to market.” That’s a true statement. But why would a supplier be– I mean, they’re interested in that, but what value can we bring to them that is extraordinary? Without getting into too many details, I’m cognizant of time, we’ll have monthly huddles between our category management team and our suppliers. We will go through the data but we’ll do it from a perspective of trying to show them what the insight is, as opposed to the data. So we’re really teaching the suppliers how we’re able to collect the data and analyze it and product those insights share it with them, share it with them, and they can see how they can take that deeper into their organization to plan their production. Whether or not they need their own marketing or sales staff – which frequently they don’t – so it’s really again focusing on what’s of value, and giving them insight from the data.


Can they take more risks, or [inaudible] lowering the risks, maybe?


We’re lowering the risks, yeah. Absolutely.


Okay. Mohan, one of your chief initiatives was marketing attribution. Tell us a little bit about how that worked, maybe before and after applying new big data analytics. Maybe you didn’t do it before, actually [chuckles].


I think the way back story is that basically all the channel champions – so everyone knows that we are a catalogue or direct mailer – but we’re also a very substantial e-commerce business. So we have many different marketing channels in play. Each channel would have their own math about I sent you a catalogue, it’s my credit, I organized email, I sent you targeted display, I paid for the search budget,” and so we had 150% revenue organization. It’s been quite a journey. I would say that the political and internal organizational conclusions of all that are not completely worked out. But of interest to the audience here is more on the technology side and the analytics side.


We are, with our partners, basically flighting a time-to-event modeling approach to attribution. So if you make an order, we’re looking at everything we’ve done to you that we can attach to you. So: every catalogue, every email, every time that we know that you’ve come to the site, why did you come to the sigh, every remarketing impression that we’ve been able to serve, and trying to disentangle that so that all the dollars add up to one. That’s been a super interesting technical challenge. I think we’re very pleased to–


Has it resulted in a shifting in spending? Do you do send out fewer catalogues, or do more search, or do less display?


We’re sort of in the early days. Williams-Sonoma Inc. is Pottery Barn, Pottery Barn Kids, Teen, West Elm, Williams-Sonoma and so we’ve actually just completed the analytics for all the brands, and now we’re starting to–


Make those decisions now.


Make informed judgments.


Okay. Did you do this with internal staff only or did you have a partner, or a series of agencies helping you out with that?


We worked with a partner who is both a technology and an analytics partner. Their name is Upstream. We worked closely with them. When it comes to negotiating, we’re customer one. But I think we were pretty important for them in terms of this approach and then I know other retailers are looking at them.


Sure, sure. We have two examples of the big picture– doing a marketing mix-modeling and evaluating spending and things like that is a cross-functional. It’s a marketing function and an advertizing function – but it’s cross-channel. You were talking about a lot of supply chain assistance. Is this an indicator of where big data analytics kicks in first? They’re somewhat different – one’s obviously supply chain and one’s marketing. But neither one of you guys were applying it to manufacturing or anything like that. You talked a bit about real estate on the phone. Is that a pattern? Is there a place where big data analytics can prove themselves first inside of a company? Or does it just depend on where you start it? John, do you have a theory on that?


It’s interesting, I don’t know that there’s a place where it’s going to evolve first and demonstrate success and move on. The obvious one, and certainly in our company, was from a marketing perspective. Trying to better understand the customer, trying to understand how to market to that customer and the lifetime value. But, I think before that, an organization needs a culture where it’s prepared to make the investment and time in big data and analytics. To understand that the way to get real value and turn that data into insight is to squash the notion that– for example, at the beginning you can think big data is a silver bullet. We’re just going to trust what it says. But we’re not going to go deeper to understand the insight. So I think for us, we preach that data and analytics is important department-wide. But really then empowering people to be curious and ask questions. Get integrated and involved in data and analytics. In that sense, it probably started in marketing first but it’s when you teach users how to really cut into it and start harnessing it, and it’s ok to ask questions, be curious. That also helps validate the data. Other than that, people, I think, have a difficult time trusting it.


So it started in marketing and then it spread and you saw the true value of it when you were looking at it across things like supply chain management.


I think so.


You started in marketing too or is there a path deeper into– where some of these same systems and processes are going to be used in different parts of the Williams-Sonoma organization?


Historically, before it was big data, and even before it was data mining, it was database marketing. I don’t want to come off as the luddite in the room but we’ve been doing this for a long time within marketing in Williams-Sonoma Inc. A lot of other direct retailers have some component of– they don’t have to explain the propensity to purchase based on a marketing stimulus.


Catalogue guys get that from day one.


So that’s there. I would say that it’s about putting the business question first. In marketing or where my experience is, it’s very clear if you’re driving program success by leveraging vast volumes of data, having a correct problem and a way to intervene in the real world with the answer and you can say ” This is what I accomplished as a result of that,” life is good and easy. I think there are good applications in supply chain for absolutely sure forecasting. But we were talking a little bit about the data warehouse thing. What are you going to do with your data warehouse? It’s kind of that you’re already sunk. Right? What are you going to do with your big data installation? I think you’re similarly sunk. If you’re not starting on that road with business problems in mind.


Who is the champion? So you’re a central analytics organization. You guys are centralized as well, right – although where does it live in BuildDirect?


At BuildDirect the analytics team reports directly to the head of IT. I wouldn’t always recommend that to be the case. I think it depends on the leadership in the department. The gentleman running IT’s background is more in terms of business process and optimization. He’s really looking at the needs of the organization in a broader context. As part of the finance group I have my own analytics guy. I kind of rely on that I can ask some of the ” What if?” questions but at the end of the day there will be weekly analytics huddles and my analytics guy will be part of that, just so there’s a cohesive strategy going forward. Right now it does report into IT.


So you vendors out there, there’s some sales channel issues, right? You work for the CMO right? You’re purely in the marketing department. And you guys do make the decisions on whether you buy teradata or whatever else right? Not so much?


Not so much, no.


What kind of decisions do you help– do you build the specs?


We’re in the advisory role for those things.


For the IT guys.


It’s still provisionary.


Okay. Alrighty. Since you’ve been using these analytics processes, what has that done to the organization? I’m teasing the organization question a little bit deeper. We’ve got analytics but it’s in IT, analytics that’s in marketing. What does that do to the rest of the organization? What does accounting– what’s finance’s role in the context of big data analytics? Who do you talk to?


Again, I have my own financial analyst, but I think part of– and I also will participate in the weekly analytics huddles. I find it’s really important to make sure that there’s alignment throughout the organization and you don’t silo analytics. When you’ve got the data scientists you really want them interfacing with as many people in the organization as possible, so that they’re able to predict user needs. The other side of that is that the users will then have access to asking the questions and to better understand the data. So I think it’s really important to try and– it’s got to report into somebody from just a control perspective and make sure that there’s a right cadence in what people are working on. But it really needs to be distributed widely and that could be, as we’ve done it, just weekly analytics meetings where you bring in different departments’ heads and constituents to determine what the next integration of strategy or questioning is.


Weekly meetings sounds like a lot. How do you do your communications at Williams-Sonoma?


I think our primary customer is marketing but I also own our ad hoc and analytic research queue and that’s what touches–


Forecasting and modeling?


Touches the broader organization. Like potential store closure. Many-many departments come to us with the ad hoc question.


They come. You’re responsive rather than pushing things out to them in lots of cases?


There’s a little bit of both but if I had the crystal ball and watching the clock, I’d say that analytics center of excellence, wherever it’s housed, is sort of our present and I think that is the future. Where you have business analysts across the business that is wise to analytics but you can’t disassemble the methods people. I think you need to keep them together. I think that you need to try to up the level of the organization to be able to consume the products of the central function. Then those really smart business analysts represent the domain knowledge and inventory and finance and real estate and so on.


So it sounds like actually from both – and I may be reading too much between the lines, or trying to connect too many dots, to mix a few metaphors – efficiency is one of the things that was achieved, if you pull off your marketing mix analysis, it will be about efficiency. You’ve gotten maybe new customers or maybe better relationships with customers on the wholesale side, who are working closer with you, so we’re seeing benefits from using this kind of analytics in very distinct roles. What do you think will be next? Will it be a time to market thing, will it be an increase sales vs. increase profits? Any speculation for where it will go next at the company?


Well our next thing is given that we’ve assembled this enormous asset is, we’re finding it very useful for targeting.


Okay. What’s BuildDirect looking at next?


I think it’s really driving customer value. Making sure we understand how this benefits the customer.


So would that be a retention thing, or increasing average sales?


It would be all of the above. Finding new consumers, and just providing more value to existing consumers.


Very good. Well thank you very much. I enjoyed the conversation.

firstpage of 2
  1. In my opinion, Analytics is the biggest area of leverage for the space.

    We’ve invested a lot in the pipe and infrastructure for big data so far. Customers are now asking how they can take full advantage of it.

    Analytics might not be a silver bullet, but it’s Big Data’s “killer app”.

  2. What is old is new again. I was involved in analyzing big data sets (for micro targeting) over 15 years ago, and I am sure I was not pioneering it. The message was the same back then: collecting lots of data and having powerful analytical tools do not provide answers or insights. It takes smart, curious people who are dedicated to finding the insights that really make it work. Skimping on investing in people means the company will have fewer opportunities for success.

  3. Reblogged this on analyticalsolution and commented:
    Experience really does matter! May not be a Silver Bullet but better than flying by the seat of your shorts!!!


Comments have been disabled for this post