If you time-travel to 1950 and meet someone who is curious about what 2016 is like, what would you tell them?
Perhaps that we have this little thing that fits in the palm of our hands that has access to all the information in this entire world; you will likely add that what we do with it the most is to look at pictures of cats.
This is the crux of the big data paradox. It is often not how much data you have, but what you do with it that drives success. Depending on your need, the right small data may be more valuable to you than big data.
Oops, You're Missing Out!
NorCal BMA members have access to the slide deck for this talk.
Interested in becoming a member?
Theresa Kushner, a 20+ year veteran of data driven marketing, recalled an incident at IBM many years ago where a marketing executive lamented the lack of availability of quality marketing data. It later became apparent that the real problem was that this data was stored in such a way that it was inaccessible to the needy.
This problem still persists, in different forms, with the big data phenomenon. While all data driven marketing is affected by it, B2B marketers face special challenges.
What data you really need
B2B database marketing is more difficult than B2C. In B2C, you are reaching the consumers who are largely the decision makers. B2B adds a number of impediments to this direct reach:
- Hierarchical data is complex. Large companies have a range of presence from site to headquarters to the whole enterprise. It is not uncommon for data pertaining to one site, or one segment of the company is organized differently from that pertaining to a different segment.
- Relationship among different contacts within the same company is important, but are largely unknown.
- The role of a contact needs to be known. It is important to know how far removed the contacts are from decision makers. It takes an average of five people in a corporation to make a decision. Your chances of conversion from initial contact to a final sale drops to less than 50% when you have five people to traverse in your conversion pipeline. The longer the trail, the lower the chances of conversion.
- Transaction data is distinct from decision data. Ship-to and bill-to addresses are different; business vs. home contacts add a level of complexity
- Hard to correlate website hits and other unstructured data to company and specific contacts
The data you need depends on its role in your company's strategy. Consider the following examples:
- You are customer-focused. Customer is your top priority
- You want to help your channel partners to have a better profile on customers to maximize sales
- You want to understand your operations better
- You want to explore additional monetization opportunities
- You want to do competitive analysis to one-up your competitors and be the best in class
The data you need for each of the above is different. So what does data-driven mean to your company?
How to evaluate which data is most important
Consider some example scenarios and see what data would make sense to collect:
- To fulfill your company strategy ... collect market share data and financial information
- To give you the competitive edge ... collect IP addresses of competitive installs
- To make the biggest impact on your revenue ... collect data from partners, distributors, or directly from customers
The trick is to decide what your goal is and then establish what data is appropriate to achieve it.
Data acquisition needs to keep the following in mind
- Identify the data elements you need
- Add elements available from third-party suppliers
- Fill the gaps with data discovery
- Heed the 80-20 rule: Focus on the 20% of accounts that provide 80% of revenue
Your website is your first collection point. However, you need to make sure the fields in the form match what you really need. Consider the following form, typical of what you may find online. Decide if all the fields are required.
You must also keep in mind that a portion of incoming data is fake as people don't want to enter that information.
You will want to supplement your data with IP address identification (e.g. Demandbase, Visistat, Profound).
Use next generation data sources for state-of-the-art data collection.
What kind of environment you need
There is no one-size-fits-all in terms of data solutions that cater to different companies. Database builds are specific to the needs of individual marketing departments.
Broadly speaking, they fall into three categories:
Enterprise Data Warehouse
This is a more traditional set up where financial, marketing, and sales data flow into a relational data warehouse, Hadoop Big Data cluster, or NoSQL real-time store and a variety of Business Intelligence reports and visualizations are generated on the other end.
Big Data environment
Companies with web market data slide into next generation data architecture and build Big Data environments.
In this model, a Big Data Refinery is in a macro feedback loop with a store of business transactions and interactions and a store of business intelligence & analytics.
The latest buzz is the Data Lake where structured (e.g. spreadsheets) and unstructured data (e.g. emails, PDF files) coexist in a common pool and business intelligence is gleaned from this collective data using tools to quickly 'sift' through them.
Where you start to extract insights
Collecting marketing data is an investment worth making only when there are well defined plans for extracting and using the business intelligence arising out of it.
B2B data-driven techniques for use of such collection include segmentation, penetration analysis, profiling, modeling, targeting/campaign selection, and recording results of marketing activity.
Consider the top applications of data to marketing:
- Purchase patterns, product patterns,
- segmentation. We need to pay attention to cases where segmentation of 1 comes up.
- Campaign targeting/selection
- Reactivate dormant/lost customers
- Campaign results
- ROI, optimize marketing investments
- Lifetime value, Managing customer. Without understanding the lifetime value of a customer, it is hard to maximize ROI for marketing investments. Propensity to buy models help in this aspect.
Using Propensity-to-buy model
This works best with your own data, not anyone else's. Some companies offer to use your data but their own algorithms to generate this model. This doesn't always work either. You need to understand how the algorithm was created to understand whether it fits your specific situation.
Use of propensity-to-buy model at VMware resulted in a lift of 3 to 4 times in top two percentiles/decile versus random targeting at the global level. In dollars, this represented a potential annual opportunity of over $100M.
Analyzing Store Traffic
Another Big Data project at VMware was to analyze store traffic and conversions.
The goal was to improve overall e-tail conversion by driving product level analysis and optimizing customer path to purchase.
The approach was to share product level optimization analysis with marketing teams; suggest promoting products identified as high conversions. The idea is to fit all traffic in specific buckets in this model:
The findings of this project were that upgrades get higher conversions with very low traffic, and that most marketing dollars spent on products with low conversions.
Potential annual revenue impact of this optimization was estimated to be over $2.5M.
Theresa Kushner, Vice President of Enterprise Information Management at VMware, specializes in extracting marketing value from data, big or small.
She finds her passion for harnessing data growing just as data gets bigger. Her crusade is against BAD DATA, the bane of marketers.
A 20+ year veteran in the field, Theresa Kushner joined VMware in 2012. She grew VMware's data capability starting with data governance and master data management, expanding to business intelligence and advanced analytics. Her team was recognized by The Data Warehousing Institute for Best Enterprise Business Intelligence in 2015. At VMware, Theresa helped launch the company’s women’s initiative, VMwomen.
Before joining VMware, Theresa was Senior Director of Customer Intelligence at Cisco Systems. She joined Cisco in 2006 from IBM in Armonk, NY. At IBM and Cisco she managed similar functions responsible for increasing market share, driving revenue and applying insights from company data both in the U.S. and Europe.
Kevin Bolden (VP and Global Head of Marketing Insights and Analytics at HP) says, "[As Senior Director, Customer Intelligence at Cisco], Theresa has done a fantastic job - her understanding of the domain, the information, and the business process has been fantastic."
Bryan Maach (Managing Director, Career Solutions at ACT) observes, "Theresa’s unwavering demonstration of integrity, dedication, and excellence truly inspires others around her to do their very best."
Maria Villar (Global VP, Data management & Governance, SAP) offers, "Theresa is an expert in direct marketing, customer intelligence and data management."
Theresa and Maria Villar co-authored the book “Managing Your Business Data from Chaos to Confidence,” published in 2006.
Theresa has co-authored with Ruth Stevens the book “B2B Data-Driven Marketing: Sources, Uses, Results,” just released in June 2015.