BrainTrust Query: How Can Retailers Avoid Drowning in Big Data?
Through a special arrangement, presented here for discussion is a summary of a current article from the M Squared Group blog.
Last week, I met with a company that had the most complete data that I have ever seen. From web traffic to email/direct mail to transactions and so on, this company had invested in their data infrastructure as a key asset. But they were still dissatisfied. They weren’t making money from the data.
The complaints of that digital marketing team were the same as companies with less complete data:
"Management keeps asking questions that are interesting but not actionable."
"We haven’t had time to create the measurement approach."
"Control groups are not part of the company culture here," and so on.
The main complaint boiled down to one thing: "We don’t know the right questions to ask."
This issue is not uncommon. When you look at companies who are the most successful at leveraging Big Data, they have the following characteristics:
- Fewer questions to start, rather than more. Rather than investigate everything under the sun, these companies have taken a "step-by-step" approach focusing on identifying opportunities for quick wins from relatively quick analysis. The goal is to increase company confidence in the accuracy of the data and analysis as well as to demonstrate how many profitable actions can be identified and taken based on data-driven insights.
- Begin with the end in mind. Rather than research "why people buy more red t-shirts in Tulsa on Tuesday," each analysis should begin with a hypothesis of what the answer might be, and an explanation of the action that could be taken based on that finding. That way, the ROI of data-mining Big Data can be clear and meaningful.
- Measure and publicize the results. The value of Big Data is based strictly on the actions that can be taken by analyzing it. The actions must be driven cleanly by the insights (so others can see how the analysis led to conclusions which led to marketing programs) and must result in a change in customer behavior that drives incremental revenue and profit. Whether pre/post, vs. prior year or vs. control group, results must be measured to ensure they are valid.
Organizing data to answer questions is one of the challenges of Big Data but the greater one is focus. We will, more than ever, be able to ask almost any question about our businesses and the data will be there to answer.
But the key is — which questions to ask.
Discussion Questions
What are the key questions that any company should be asking when attempting to leverage Big Data? What suggestions would you add to those offered in the article?
Fewer questions are better than the usual “laundry list.” If I could ask only 4 questions, they would be:
– Which group of customers are raving advocates and why?
– Which group of customers are neutral and why?
– Which group of customers are disappointed and why?
– Why do our prospects buy from our competitors?
The insights gained from these four groups are the foundation to a solid business plan for growth.
The win is to use the least amount of data that can deliver the most impact. Less is always more. I sat in a meeting at GSK a few years ago and the executive I was meeting with said “I don’t want to play Trivial Pursuit, I want to play Monopoly.” Boiled down, facts are not worth much if you can’t create value from them.
In particular, I think 1 and 2 above are very important tips when thinking about Big Data. The other tip I would include is to be sure and source from more than one data source. Just because you see a bump in red shirts doesn’t mean it is a trend you can replicate in other markets. Validate before you expand the approach.
Finally, I agree that focus is key.
With big data, a way to gain more value faster is to look at the steps that key associates go through to get their jobs done. Then leverage front-end tools that are easy to use and run against the data sets and focus on making 2 or 3 of the steps they go through faster and easier, and with more informative than ever.
Take a retailer planner…they have x-steps that they cover each day. Impact 2 or 3 of these steps to start.
Big data is not just about lots of new things to report on and measure, it is about velocity and volume and of course, value. Use it to make your teams’ jobs better and in the end, the shopper experience will be more exciting!
Tom…Big Data Guy….
More important than the questions are the answers. Figure out what you need to know, what you would do with that information if provided, and how it would impact the business. This is the difference between data/knowledge/insights.
The questions (and their priority, importance, and order) will follow from targeting what you seek to understand. I would suggest becoming very focused on what you want to know. Do not attempt to wrap your arms around the Empire State Building. You will fail. Rather, grab a hammer and nails and bang two boards together. Then, connect those two boards to two others. Repeat.
I am a big fan of progressive performance over trying to get to to perfection (and as the example shows, always having to postpone it because there is new information and it is next to impossible to synthesize what has already been collected).
The key questions companies should be asking about leveraging their “Big Data” (assuming they have said data) should start with one simple one: what are you trying to or hoping to achieve?
If it’s incremental revenue then there are questions to ask leading to which customers present the best and biggest opportunity; whether the opportunity is about incremental transactions or growing the average transaction, etc.
The point Mark makes above about starting with a hypothesis is also hugely valuable. Most companies, and nearly all CEOs, have hypotheses that related to specific questions that can be answered with data. My personal favorite is the one posed by a client CEO, prior to a meeting, when I asked how the business was doing. His reply was that “comps are down 2-3%; I think it’s _____ (customer segment.” My reaction: we can tell you exactly which customers are driving the sales declines, but you need to look at the actual customer data rather than speculate.
As a fundamental starting point, big data should illuminate specific insights about how the business is performing and when looked at from the standpoint of customers, it’s pretty easy to begin identifying opportunities to leverage the data to drive incremental, and profitable, sales.
One of the keys with data and analytics initiatives is to put the information in the hands of people who can act on it: store managers, district managers, and merchants. Even better, present well visualized data that clearly depicts the “so what.” Data that is easily accessible and can be visualized and manipulated by non-technical end users will have the most impact.
Does not big data require a lot of forethought and a decent data warehouse as foundation? Throwing data into an inefficiently designed warehouse is just going to result in a morass of data that’s going to be difficult to make sense out of.
Large volumes of data have been around for some time; few companies have mined it efficiently enough to get genuinely actionable insights out of it.
Well if we are going to have a productive discussion about this, we had best begin by clarifying the crucial difference between very large data marts and “Big Data.”
Of course relational databases are expanding in size. Transactional and shopper data pour in every day. Data mining and modeling techniques are gaining sophistication and scale to match.
But the story of Big Data is about something far more vast and dynamic. It’s about the data flows inherent in search, social and mobile media participation. It’s the data of everything online and it is fundamentally different from the very large marketing databases we are used to.
Big Data flows too fast to download and analyze. Using it is about pattern recognition—a whole new class of methodologies and analytics.
So the first question any company must ask itself when it decides to tackle “Big Data” is: “Are we sure we know what we are talking about?”
Any company using data should ask: How can the data be used to inform strategy across the business if we change the goals or metrics through which we assess our performance? Lots of organizations just extend the platforms they are using and then expand the data asset accordingly. But what would happen if they used customers as the basis of analysis? How would that cause their perspective to change?
In order to use the customer as a basis of analysis, they must define their ideal loyal customer and determined a strategy for best serving that customer, with appropriate feedback mechanisms in place. Lastly, they should incorporate measurements to ensure they are hitting a predetermined goal.
I am always interested to see that the answers to the poll normally cluster around one winner. I hope that is testament to the collective wisdom of the group and also an indicator that the answers are proxy for best practice in one form or another.
The answer to the company’s issues are wrapped up in the three factors raised at the top of the article:
1. Asking questions that are interesting, not actionable
2. Lack of commitment to forming a measurement plan
3. Lack of use of control groups
These are essential disciplines for database marketers and I’m not sure why many organizations resist their embrace.
That’s the really interesting question.
Retailers have a wealth of information about their customers – from online transaction records to social media data – but the key is drawing insights from all the data, regardless of its channel or source. Organizations that are able to capture and analyze the data gain a significant advantage over competitors. That being written, companies need to ask themselves all the typical “new process” questions like what are the goals and what resources are needed, and many more. Here are a few:
The key is to integrate new types of data with traditional business data that organizations already have. By opening up access to the entire corporate ecosystem and incorporating data from all sources, retailers can use big data analytics to achieve a super-charged view of the customer to improve service and sales. We are seeing a number of top retailers beginning to do this and it is their best-kept secret—they are getting very serious ROI on data integration in the mega millions.
The only question is, “Is it actionable?” The fact is that both big data and small data are HISTORY. We should all learn from history to avoid mistakes in the future. This, however, has proved very difficult for most. Take the lessons of history to heart: nothing is free, data analysis does not get one customer into the store, neither does it convert a shopper into a buyer.
Put your money and efforts into things that make you, your products, and your employees better. If you can use data to find out what you are doing wrong so you can correct it, great, but wasting a lot of time and money on big data is just your hunt for another silver bullet. How long have you been looking and how many have you already bought? How much good did they do?
Some of the early benefits that retailers have seen by adopting big data technology is lower cost of storage. This opens up various avenues to leverage the data which until now was not stored or couldn’t be retrieved as easily before.
One of the clear wins which most retailers could get from big data is forecasting. Increasing the accuracy by even one basis point can lead to huge benefits.
Some of the key questions to be asked are not about big data at all, but about the focus on things the company wants to fix. Data is just a means to find the solution, not the solution itself.
First, how much do you actually trust the data you’r already capturing? I’ve had leaders of major retailers tell me that they have some serious data cleansing to perform before they even start to think about analyzing the data.
The next step is to determine scope and roles of the leaders. E.g., how are the CFO, CIO, CMO involved? Then, what can be done with the tools you already have implemented? Does this automatically mean you need to invest in analytics tools?
Big data is more than a matter of size; it is a way to uncover insights and opportunities from new and emerging internal and external sources of data and content. Big data analysis capabilities should include an enterprise-class big data platform, predictive and content analytics, and decision management to give your organization a competitive edge. Capabilities should be designed to complement your existing information, analytics and content management infrastructure, so you can get started quickly and achieve game-changing results.
Jamie Tenser said: “Big Data flows too fast to download and analyze. Using it is about pattern recognition a whole new class of methodologies and analytics”.
Which brings out the practical problem of whether retailers can respond quickly enough to fast-changing information.
I work for a major multinational supplying software to the retail industry. The existing software is not suited to rapid changes resulting from analysis of big data, nor probably are most retailer’s systems.
So even if you can mine the data quickly enough to make practical use of it, that then raises the further question of what changes retailers (and software companies) need to make to be able to respond in real time.