Ep. 8 | Using Data to Get Results Part 2
Host Allison Hartsoe continues her interview with data analytics legend, Bob Page, who offers several key tactics for setting up a data framework and how to use it to get results. He talks about the impact you can get from data, and advises companies to focus on generating revenue before cost savings. He shares his strategy for building a data framework, including an inventory of systems, consolidating data points, and pulling in data from external events that impact your business such as weather or exchange rates.
Read Full Transcript
Allison Hartsoe – 00:06 – This is the Customer Equity Accelerator, a weekly show for marketing executives who need to accelerate customer-centric thinking and digital maturity. I’m your host, Allison Hartsoe of Ambition Data. This show features innovative guests who share quick wins on how to improve your bottom line while creating happier, more valuable customers. Ready to accelerate. Let’s go.
Allison Hartsoe – 00:33 – Welcome back to the second part of my interview with Silicon Valley big data technology legend Bob Page. In the first part of our discussion, Bob introduced us to why you should care about the pit of technology despair through is what I thought were incredible real-world stories running big data analytics for eBay and Yahoo, and then we went on and discussed whether there were any ROI impacts that could happen behind all this tack and if so, what that looked like. Now we’re going to talk about how you can learn from a leader and apply this wisdom. Here we go.
Allison Hartsoe – 01:22 – So Bob, in your experience, is there anything like a rule of thumb about how much to spend? Like I might know if I’m a product company, I spend x percent of my revenue on marketing and if I don’t spend enough, I’m not going to get enough. Is there any kind of rule of thumb for technology?
Bob Page – 01:40 – I don’t think so. I think it depends on the temperament of the company. Is it a technology company? Is a data company as a customer first company, maybe it’s a marketing company, maybe they pride themselves on support in service. What matters to you? What matters to the leadership of the company is going to determine how much they’re going to want to spend on this kind of stuff. I do think it’s critical. It’s important for a CAO or CEO to have a handle on what it is. If I went to the CFO and said, I’m currently responsible for burning a half a percent of the company revenue every quarter, and I’d like to double that. Okay, well I’m going to get his attention and going to say, well what are you gonna do with that? How am I going to get my investment back? Why should they make it a quarter percent of the half percent? So I think that yes, there was a cost of building all this stuff and there’s an ROI as slippery as it might be, but you still have to try to figure out those numbers.
Bob Page – 02:38 – At a minimum, you can determine how much does your date program cost, you know, what percentage of revenue, for example, or as a percentage of your total r and d budget or your total expense budget.
Allison Hartsoe – 02:49 – Yeah, that makes sense. And just one more thing on this. Considering the pace of change, how fast and new technologies are rolling out and how fast they’re becoming commoditized. Would it be fair to say that anyone who’s looking to pick up technology should look for the most value, which doesn’t mean the most features, the most value at the lowest cost?
Bob Page – 03:12 – I want to say yes, but there are so many [inaudible] cases. Yes, it is always about value. But is it value today vs. in 12 months because you don’t want to keep rebuilding systems. Do you need to retrain your workforce when you’re brand new technologies and people costs are in general much more extensive than systems costs?
Allison Hartsoe – 03:35 – Yeah. Well, let’s say that I’ve decided that I’m. I’m ready. I’m going to avoid the pit of technology despair. Help me understand what should I do first,
Bob Page – 03:45 – of course, admit that you’re not going to avoid it.
Allison Hartsoe – 03:48 – Yeah. Right into the pit.
Bob Page – 03:53 – Get there by being successful in all of your efforts up until now, right? Everyone’s excited. Everyone’s going to march off and they’re doing what they can to think about customers first and making sure that platforms in their system or in their reporting and everything is done the right way, and then they realized they could get more value if they were able to have data that I have or vice versa. And so I would say you’re in the pit now. This may be a little powerful, you know, it may be a deep dungeon because of only you and the prince and no other way in and out or whatever, but you can at least prevent yourself from the kind of wallowing in it. So does that make sense?
Allison Hartsoe – 04:39 – It does, it does, and I don’t know if I [inaudible] the word wallow in business anymore, which is just so appropriate. Got It, got it. So starting there is there some kind of framework or a format you’d suggest
Bob Page – 04:54 – Yeah. One of the things that I did quite a lot when I consult with companies that were doing Hadoop adoption and they were trying to figure out what’s the best way to get started, how do we don’t shoot ourselves in the foot, later on, etcetera, etcetera, and so there are some keys that I think are important. The first is to inventory your systems that impact the customer. You know, you might think on my website, my enterprise data warehouse reporting tools, but there’s also other stuff, right? Those things that impact the customer experience. Now your email campaign. Yeah, Your call center, your bug tracking tools, right? So and so keeps reporting these errors or bugs or whatever you should know that and even things like sort of indirect impact on your customer, like say your supply chain systems. I mean if you have stock-outs or overstock and tie that the customer demand and you can do, you can understand opportunities for your customers that way.
Allison Hartsoe – 05:49 – I have to detour on this for just a second because we want to do this analysis, that would have been perfect for what you’re outlining here in terms of the inventory of systems. Where there a big branding company and they were driving a ton of event and success traffic to this product retailing company and what they were doing was blowing up demand which would then reduce inventory and explode the call center, but they couldn’t see this complete loop and I think one time we figured out it was a ratio of 17 to one unhappy customer per minute. It was insane.
Bob Page – 06:23 – Yeah. If you’ve got say, if you don’t tie your business development efforts on your marketing efforts together, even though they’ve got to do with the customer, but market 10% off on all wristwatches, you haven’t talked to the business development folks, and they haven’t made sure you’ve got plenty of supply hand and talking to each other through phone calls and emails and slack channels and stuff. I’m not going to make that call, but it’s an in the data somehow
Allison Hartsoe – 07:00 – it wasn’t directly customer, but it actually is because the customer sees your organization as one entity. It doesn’t see you as you know all these splintered business units that you know your one to them.
Bob Page – 07:11 – Yup. Yup. For this inventory and the systems that have to impact the customer in some way should not be this fine detail every piece of data documented for every system that’s going to stop you from doing the work. Instead, it should be kind of high level like what kinds of customer data is available and how does it impact the customer experience or customer value. Does it have any PII personally identifiable information and kind of roughly what kind of data volumes that we’re talking about. Are we talking about petabytes? So we are talking about just a little trickle of data. It’s something that gets refreshed once a year. I mean, you know, how fast is it being is coming in and how much do you have?
Allison Hartsoe – 07:11 – So kind of [inaudible] inventory.
Bob Page – 07:58 – Yeah, yeah. And from there it’s like primarily you’re going to be focused on now these different systems and what have you, and so now it’s like, okay, I have all this now let’s go to consolidation point for that data. Call a data lake, if you want to call it customer data, Landing Zone, call it you want, but I think again, being pragmatic about it, don’t worry about what data is going to end up being relevant or highest value. Don’t worry about how long are you going to store it. Just thinking about getting the data and getting the data back to analysts and machinery and data scientists quickly. So if I could be prescriptive, I would say you want to look for the Hadoop distributed file system, hdfs. That’s where the data should be stored. It’s resilient. It’s been proven over a decade. It’s the place where you want your analytics data, your customer data,
Allison Hartsoe – 08:52 – This is like the 50-lane freeway in a sense that we were talking about earlier.
Bob Page – 08:54 – Yeah, and the data is replicated so that if any one of your lanes crashes, everything just gets rerouted, lonely problem, and you don’t lose any data. So I would say go to Hortonworks or Cloudera. They’re the two big public companies that are supporting and download free Hadoop distribution and then later, much later, once you’ve got system up and all that, you can decide that you want to pay them for things like support or for any proprietary technologies that they might have, but do that later because you can get started for free. Good.
Allison Hartsoe – 09:28 – I just want to ask a clarifying question here because one of the things you said earlier and I just want to make sure I heard this right. You’re not just talking about all the analytics data and the 50 lane highway and getting to it quickly, but I think what’s also happening here is because I’m not always in. Maybe I’m not sure I’m getting this right, but because I’m not editing the core, I’m actually able to work with data sets without messing with the master systems and moving the master systems around. Is that what we’re also saying?
Bob Page – 10:00 – Yeah, that’s a good point in it. You’ve got a bunch of operational systems now that work, don’t mess with them. All you want to do is pull data from them. Got It. Into some other system where then you can do all your fancy stuff.
Allison Hartsoe – 10:15 – Got It. So it’s like my 50-lane highway is sitting on top of my office building of traditional data sets.
Bob Page – 10:22 – Yeah, data highway in the sky, but yeah, I would not prescribe people think about this as a rip and replace kind of thing. You’re not saying we’re going to throw away. Our existing systems are going to put this in place frankly, because you want some quick wins and very easy to think, oh wow, wow Hadoop system in for a 10th the cost of my existing gold plated analytics and enterprise data warehouse, maybe I should just do that, and I can get a whole lot of value because I’ve reduced my costs, but what you don’t think about is the fact that first of all, the CFO and the CEO, they care about that stuff. I think they care about additional revenue more and so think about your effort here in consolidating data about users and providing new user products or experiences as an opportunity to increase revenue sources and that’s something that’s a whole lot sexier.
Bob Page – 11:20 – Also, just to be really clear about an sort of further cement, this idea that you should go ask her so quick wins on the revenue side instead of kind of do cost reduction, probably years of business analytics encoded into your data warehouse and reporting systems today, years, do you want to just try to rebuild all that stuff in this new system that’s going to take what, six months a year for you to just put that stuff and then giving you the same numbers because you got to run these things.
Allison Hartsoe – 11:20 – Strategy,
Bob Page – 11:56 – so I would say don’t go there. We would not be able to produce this product or produce this value on our existing systems or we can with this new systems. Eventually, eventually people are going to be so excited about all the things that you’re doing that they’re going to say, could this be used for cost reduction? And you can say, well, in fact, we’ve been experimented with that, and we have some good news. Right, but you shouldn’t be leading with cost reduction in my mind,
Allison Hartsoe – 12:25 – that’s interesting because I think a lot of people do lead with cost reduction because it gives an immediate payback on this particular number on our expense.
Bob Page – 12:34 – I know that they do and I know that the Hadoop vendors also talk about things like warehouse optimization, code words, right? Like we used to say, know sizing or whatever, but this is similar, and I say that might be the way to go. Maybe you don’t have a lot of business logic. Maybe you don’t have years and years of hundreds of thousands of reports and tables and everything else that is sort of institutionalized in the way you do business. If that’s the case, and it’s a quicker win to show some cost reduction, then great, but I think for a, for anyone who’s been at this for any amount of time, my advice would be put that off instead. Show Positive ROI of your system by generating new income because [inaudible].
Allison Hartsoe – 13:17 – Yeah. Well, new income is a. That’s very attractive, especially when organizations work so hard to get it and keep it in the first place. So on your framework, I think we’ve hit three points so far, the inventory of the system building that consolidation point for data and it sounds like we just had a third one which was using that new system to increase revenue, right? Yield data and products that increased revenue that. That makes sense. Are there more elements to the framework?
Bob Page – 13:45 – Yeah, a couple more I’d recommend, but let me just finish on this one. About the new products. How do you get there? I would say Hadoop now supports SQL pole in a technology called comes with equal access, so all your analysts can get you reporting systems can attach to it and what have you, but also there’s something called spark and spark is of the way in which people are doing a lot of analysis today. It’s very, very fast. It also comes with Cloudera and Hortonworks Hadoop distributions. So if you have those things, your data scientists quickly all. So that would be the third one. So the fourth would make the data available to all the departments.
Allison Hartsoe – 14:24 – Okay. I’ve heard a lot of heated discussion around this. I really want to hear your point of view.
Bob Page – 14:29 – Well, so on the positive side, you’re going to be surprised at the insights that the company is going to come up with when they have all the data available to them. You know, marketing never thought to look at call center data before, but now it’s all right there and so let’s do some of that analysis and see interesting things that come out or don’t come out. Now there’s a bunch of couches, right? You’ve got a locked down, any PII, right? As appropriate. Don’t let every random person, every random thing about every random person and also you’re going to have chaos on your hands if you don’t document what the data is. You know when you did your analysis, did you filter out robots and different systems are not necessarily aligned in your system. You have a field called gender, and the values are either or blank. And I have a system in my system, it’s called sex, and it’s a zero or one or two. So how do those map all? We have documentation about that somewhere.
Bob Page – 15:24 – So I’m not suggesting that you do data normalization effort across all your systems. I’m saying you should at least document them and if you can provide some code snippets or whatever to help people have a blessed way in which they should be looking at these things.
Allison Hartsoe – 15:42 – That way when I hear people struggling with this, what it often comes down to is this part of the organization is using this field to make the numbers look good in this way, and this part of the organization is using this very similar but slightly defined a different way so that unification matters. Is that something that the CFO or you know an office related to the CFO should actually get in and try to sort out?
Bob Page – 16:09 – I think so, yeah. If you think you’re going to solve governance with technology, then you’re much more technologists than I am. Governance is not a technology problem, but it needs to be backed by technology. If you can’t enforce it or assisted with technology that it’s. It’s hard to do. It’s hard to govern the data, but there needs to be some set of agreed upon. So its gold standard for the things that matter the most to say the CFO, right? Who’s driving the business and setting up budgets and looking at contribution, how things are being measured and rewarded. Now inside your own organization, you might have your own version of that same number because it better reflects the way you were thinking about it or the way to an industry group thinks about it.
Bob Page – 16:59 – Give you an example. At Yahoo, we had numbers that we would report to the business on things like visits and visitors and session length durations and all that fun stuff, and we could defend it, you know as this is what we’ve measured directly and what have you. But the sales folks in the business, for the most part, didn’t care about our numbers because all they want to do would be to see how the numbers changed from a prior period or how they relate to other businesses within young. But that’s not who they competed against, where they could compete against themselves or with other businesses. They competed with other web properties, and so they were relying on like a ComScore or somebody and those numbers and those numbers. Those are samples. They’re terrible numbers, blah, blah, blah, but you know the reality is your advertising, you don’t want everybody’s internal numbers.
Bob Page – 17:51 – You want some sort of standard that goes across all of the possible advertising places and folks like ComScore are the people who you go to so so that’s why I say you might have different sets of numbers but coming on what your needs are and who your audience is for the same thing, but you’re going to only have one set of kind of north star, but you’re going to govern the whole company, and the CEO and the CFO have to drive that.
Allison Hartsoe – 18:16 – Yeah. That’s a whole nother conversation of what that North Star is, and I know it has spent as long as a year trying to figure that out. It’s a difficult thing.
Bob Page – 18:24 – It is. It is. One more and that is okay. Now that you’ve got your system, you’ve got it in place. You’ve got the integration, your data scientists and your analysts or generating value or reducing cost or whatever they do and your different departments are getting very interesting insight. Now it’s probably time to ask what data is missing.
Allison Hartsoe – 18:44 – Now that I’ve got all this, what am I missing?
Bob Page – 18:47 – For example, I’ll give you two examples. One is we just bought another company. You’re doing a merger.
Allison Hartsoe – 18:47 – Oh goodness.
Bob Page – 18:55 – So you want to put together some kind of framework for bringing in new data and making that sort of resulting seeing available back to the new company as well as you know, the rest of the organization. But the other thing is that you know, you might take an honest look and say what external events have an impact on my business, like whether for example, or a foreign exchange rates or whatever, and if those matter to your business, you might as well go get the data and bring it in so that you have it for analysis. So you could see different causes and effects when my marketing program is what drove all this new value from France, and then you realized, nope, the foreign exchange rate flipped and suddenly France thinks that everything I’m doing is really cheap. Okay. So it’s important, and it’s easy to convince yourself of something that isn’t actually true.
Bob Page – 19:48 – But I don’t put that first. I don’t put, I mean second, I don’t put that in the data collection phase because frankly, it’s of marginal ROI. You don’t know yet what you don’t know, and so you can always go back and get it later.
Allison Hartsoe – 20:03 – So it kind of go for the outcomes razor approach.
Bob Page – 20:06 – Yeah, exactly.
Allison Hartsoe – 20:07 – Well this has been a fantastic framework to share with everyone. It’s a lot more detail than I’ve personally heard before and it’s clear, you know, from your expertise that they’ve been through the ropes on it a couple of times or through the wringer. So I really appreciate that expertise. You’ve got the arrows were standing in the front. I think in general. I just want to say that I really appreciate the work that you do and the advice that you give to other people in the industry as well as my own company to help guide them through that pit of technology despair that now unavoidable pit of technology despair. I think all these suggestions will be helpful and help people know exactly what to do when and why. So thank you for that.
Bob Page – 20:53 – No, by the way, thanks for having me on. I listened to all the episodes. I really enjoyed, and they are so dense with information, but yeah, I’m really enjoying it, so it’s a pleasure for me to be on as well.
Allison Hartsoe – 21:03 – Thank you. I don’t know if that’s a blessing or curse that you can’t listen to it at like two x speed, but I know what you mean. Like I do that with most books and other technologies. I try to plow through them as fast as possible, so I’ll take that as a good thing. Thank you, Bob.
Bob Page – 21:17 – Well you didn’t put so much information in them. They’re like, listen to them at 2x.
Allison Hartsoe – 21:22 – too much meat. So let’s summarize a little bit. First, we talked about why should I care about the pit of technology despair? We talked about technology, at least from the big guys when they’ve gone through the whole process, they’ve kind of, as Bob was saying, taking the arrows and that output has become technology in the industry that’s easier to manage, easier to gain the benefits from smoother systems, easier to operate and I think we agreed that, you know, you’re not going to avoid the pit all together, but you can find a way to work through it more smoothly. And I think looking at those, perhaps more recent technologies that come out like Hadoop and Hortonworks and Cloudera are actually great places to start because they’re taking advantage of the front half of that curve where smoother systems have actually come through. So really good stuff there.
Allison Hartsoe – 22:17 – And then second, we talked about what kind of impact can you get and the bottom line there is we’re going to worry less about the platform ROI, and we’re going to think more about the output because ultimately when you have performance from the platform, there are fewer questions about the value of it. So if you’re unlocking insight, new features, customer happiness, customer perceived value, then I think there are a lot fewer questions about the ROI of the platform, and there are easier ways to prove the value that your group is bringing the organization overall. So that was a really good section two. Now the third part was what you should do next?
Allison Hartsoe – 23:03 – And we talked through a pretty extensive framework, but it basically comes down to a handful of pieces. I think there are five pieces here. Inventory the systems, consolidate, use them, share the information and findings, and then go back and get what you were missing. Repeat and refine all the way through the Bob. Did I capture that about right? I hear you. Good. Good.
Bob Page – 23:31 – You could have done a lot faster.
Allison Hartsoe – 23:34 – I doubt that. Don’t have nearly that level of knowledge. Well, Bob. Thank you so much for joining us today. As always, everything we talk about and transcriptions of this is @am available at ambitiondata.com podcast. Remember everyone, when you’re using your data effectively, you candle customer equity. This isn’t magic. It’s just a very specific journey that you can use to get results. Thanks, everyone. See you next time.
Allison Hartsoe – 24:10 – Thank you for joining today’s show. This is Allison. Just a few things before you head out. Every Friday I put together a short bulleted list of three to five things I’ve seen that represent customer equity signal, not noise, and believe me, there’s a lot of noise out there. I actually call this email the signal things I include could be smart tools. I’ve run across articles, I’ve shared cool statistics or people and companies I think are doing amazing work, building customer equity. If you’d like to receive this nugget of goodness each week, you can sign up at ambitiondata.com, and you’ll get the very next one. I hope you enjoy The Signal. See you next week on the Customer Equity Accelerator.