The Marketing Science Institute (MSI) is pleased to announce a call for submissions to the 2018 MSI-Juniper Networks Research Initiative. MSI and Juniper Networks, an MSI member and the industry leader in network innovation (NYSE: JNPR), are working in collaboration to share B2B data with select teams of researchers to address questions of interest common to Juniper and the scientific community. Researchers are invited to submit proposals by February 16, 2018 and selected teams will be notified by March 19, 2018.
A data overview, potential research questions, and description of the submission and selection process are provided below.
Juniper Networks hosted a Q&A session for the data initiative and recorded the session. For information on how to access the session, ask other questions about the initiative, and to submit a proposal see here.
We encourage interested researchers to review the information below and attend the January 17 presentation.
Marketing Science Institute
Juniper Networks provides security, switching and routing hardware and software for the foremost wire and wireless carriers, cable and satellite operators, content and Internet service providers, and cloud and data center providers. Businesses that depend on the network to deliver mission-critical transactions, applications, and services use Juniper Networks.
Juniper Networks’ marketing organization has adopted a data-driven strategy to understand the full customer journey (i.e., the sequence of customer interactions with the firm leading to purchase). To understand this process, Juniper Networks has built a marketing data lake called Verity that integrates several sources of data from advertising logs to web logs to Salesforce.com to purchase data. By using a proprietary technology to merge different data streams, the organization is able to monitor and track the purchase decision-making steps of a customer.
In high-value B2B tech sales, purchases are made by accounts that comprise many influencers and decision-makers, all of whom are captured in the data. When decisions are made by accounts, a focus on an individual would be incomplete. Juniper’s focus has been on understanding the customer, or account so that we can determine how to assess marketing ROI, and determine how to allocate budget effectively across the marketing organization. Juniper expects that the data have the ability to be prescriptive in nature.
Data Description of Verity
The following outlines the data sources that are available from Verity along with a brief description. Data sources are noted as “connected” if they are integrated and matched by account within the Customer Journey table.
“Connected” refers to when we can tie a contact or account’s firm interactions across different stages in the journey toward purchase. For example, Juniper is able to tie together those that are exposed to our advertisements and those that came to our website. Another example is that Juniper can see if a collection of contacts who visited the website (juniper.net data) converted into a sales opportunity (salesforce.com data). This is a form of contact to account connectivity. These integrated data are summarized as follows:
- Customer Journey Table: This table combines important data elements that are extracted from the data sources below and linked (indicating how these data are merged) in this particular table (180 million records). The data are stored at the contact and account levels, enabling us to map customer-firm contacts by accounts to reveal the full journey.
- Advertising data (connected): Using proprietary techniques, we collect an impression log of most of our advertising data.
- Juniper.net data (connected): This is a complete file of all of the users that visit juniper.net. This file includes referral sources like SEO, SEM, campaign clicks, etc.
- Marketing automation (connected): This dataset enables us to track events (Webinars, offline, Executive Briefing Center), marketing email activity, click2chat, 800call, and online form fills.
- Salesforce.com (connected): All pipeline (sales interactions) data are included in this set of tables, including leads, opportunities, and closed deals. Closed deals are marked as won or lost.
- SAP – include (connected): SAP is a financial management system that provides a view into sales achievement. Sales achievement is based on products that are sold and shipped or services that have been delivered.
- Third-party data sources: Buyer intent data from third parties data sources has been incorporated into Verity. These are stored at an account level. The two primary data sources are:
- Social: These data enable us to monitor feeds from sources like Twitter, LinkedIn, and Facebook. Additionally, they provide a competitive view of social media.
- Third-party intent data (connected): These data are provided by Bombora. “Intent” data signal buying interest from specific organizations based on topics we’ve created that are relevant to Juniper Networks.
- Internal education data: The education data are integrated into Verity using an internal site called Juniper University and an integration with a training provider called Cornerstone.
Additional funds to support research efforts may be available from Juniper. These efforts may include opportunities to fund experiments, host researchers on site at Juniper Networks for an extended period of time, or purchase additional third party data. Please provide a brief description summarizing how these funds should be used. The selection of proposals will not be based on funding criterion. However, if the funding can enhance the research, then it should be outlined in the proposal.
Potential Juniper Data Lake Research Questions
Causal Attribution and Synergies for Marketing Touchpoints: Juniper Networks has prioritized each topic based on the level of interest from Juniper. Tier 1 is the highest priority and tier 3 is the lowest priority.
Attribution (tier 1)
Given the numerous, potentially synergistic marketing instruments, what is the relative effect of each in driving revenue? Current approaches, such as path analysis, may not recover the causal effect accurately, raising concerns about interpretation of effect sizes. Further, purely experimental approaches do not scale to the desired level of granularity. There exists potential in combining experimental methods with models (e.g., structural approaches) to obtain cleaner measures of attribution.
Related, about 80% of Juniper’s business involves interactions with its partner firms, including resellers, consultants, and distributors. So a related topic of interest is the sales attribution problem across Juniper and its partners.
Synergies (tier 2)
How do the various marketing instruments work together or in substitution to drive sales (e.g., acquisition, cross-selling, up-selling, and retention)? For example, lead-generation from digital advertising can enhance salesforce productivity. Two immediate aspects worth investigating are (i) the determination of potential complementarities between marketing instruments and (ii) the determination of a potential sequence or pathway of optimal interventions.
New Marketing Variables (tier 2)
To date, information on data events, sales calls, training sessions, trade shows, and so forth have been limited, suggesting new potential insights into how each drives lead generation and sales.
Overtouching & Frequency Capping (tier 2)
What is the optimal timing of the various marketing touch points? There is also limited research and knowledge about the frequency and timing of individual marketing interventions as well as groups of such activities. For example, are three display ads and two emails within a day the same as two display ads and three emails?
Randomized controlled field experiments provide the most reliable data for determining the causal effect of marketing instruments. How should such experiments be designed (which variables, which channels, which customers, etc.), and how much replication is required to ascertain the attribution of causal effects?
How can theory inform the design of experiments? To what extent do theories regarding the underlying mechanisms that drive the causal effect of a marketing instrument or synergies between instruments matter for designing experiments?
Experimental data also can be streamed into the data lake to provide more complete insights into marketing causals, and also to assess whether experimental effects are stable over time.
Heterogeneity (tier 1)
How do marketing effects vary across industries and customers? What are the implications for targeting, customization, and revenue enhancement?
Granularity (tier 2)
A given account may comprise multiple decision-makers, such as the purchasing department, engineers and finance. How do the various individuals within the account drive the decisions of the account? How do the effects of marketing vary across account members?
How should one aggregate information for decision making? Suppose one determines that marketing spending in aggregate drives total sales. Total spending can be disaggregated by vehicle (e.g. promotion) which, in turn, can be further disaggregated by marketing tool (e.g. discount or e-mail). Similarly, sales can be disaggregated by channel. In other words, what is the right lens for decision making – the effect of an email campaign on an account, or the effect of marketing spend on revenue, and how can the multiple levels of aggregation be integrated?
Sales Force Planning (tier 3)
How should one set sales force incentives? Despite the long extant literature, there is considerable economic theory on the role of salesforce incentives and their potential unintended consequences. Moreover, the extant literature has not considered the co-existence of sales force and other marketing activities.
To what extent do sales outcomes reflect the effect of salesperson ability (adverse selection), salesperson effort (moral hazard), and/or other contemporaneous firm marketing efforts?
Does sales training affect sales outcomes, and what are the implications for sales compensation?
How do other marketing instruments interact with sales incentives?
How should territory alignments and call plans be optimized?
Juniper Webinar with Additional Information on January 17
On January 17th at 10:00 AM PST, Juniper Networks will host an optional presentation and Q&A session to afford interested researchers additional information and enable them to ask questions about the data. The content of the presentation will cover the underlying data infrastructure, a review of the connected data, and a high-level review of the content used by the organization. Phone numbers and Skype links for joining this session can be found here. Please dial or Skype in 5 minutes before to ensure the Skype application is working properly.
Researchers are asked to submit a three to five page proposal outlining their research idea along with their CVs by February 16, 2018.
Three to five proposals will be selected based on the scientific contribution of the research and the fit with Juniper’s priorities. The scientific contribution will be evaluated by a committee whose members include JP Dubé (Chicago), Carl Mela (Duke), Sanjog Misra (Chicago), and representatives from MSI and Juniper Networks. Decisions will be communicated to selected teams by March 19, 2018. After signing a confidentiality agreement that will allow for publication of research, teams will be flown to Juniper’s Santa Clara offices for a one to two day meeting to initiate the research projects. Any completed research papers stemming from this research initiative are to be submitted for consideration in MSI’s working paper series.