Session III - Data Strategy

Participants

John Carnahan, Ian Small, Rahul Pathak, Ali Ghodsi, Ben White, Mark Madsen, Chris Wensel, Reynold Xin, Greg Rokita

Members Excused

Cindi Thompson

Administration

• Approved raising member count to 12.
• Approved two new members:
  • Reynold Xin
  • Gwen Shapira - https://www.linkedin.com/in/gwenshapira/ - pending Gwen accepting invitation
• Approved Council Charter.
• Approved Council Statue.
• Going forward all amendments to Charter and Statue will be explicitly transcribed in Council Statue.

Topic Discussion Notes

Introduction

In a larger enterprise there is usually a disconnect with what a Data Strategy is. If you talk to a COO of a company, their notion of a Data Strategy is how do I capitalize on the data that we have make the capitalization an asset of a business. If you talk to a CTO of a company, the strategy is how to take the data from all the independent systems, how do we put them together, how to leverage it effectively, how do join it, and how to manifest data through toolset that makes it useful. For Chief Marketing Officer of a company, the meaning is very different; it is, how do we leverage the information from the data platform to store marketing data, to be able to advertise and use it effectively for marketing. People typically come to engineering group, and they ask—what is our data strategy. If you come out and if you talk about data strategy as a platform or datasets involved, the non-technical people get confused; they have a very different notion of it.

Questions

How do people deal with this? Is there a way to join those different notions of a data strategy into something that as a group we can fall behind? We are not aware of any books on this subject, so it makes sense for us to discuss this, fill the gap and solve the dichotomy. This would significantly reduce the confusion, especially in medium and larger companies.

Open Discussion

In a very large organizations, COO, CTO and CMO are viewing the strategy though their individual lenses with their own filters set. They are not all interested in all the pieces that are not related to them. So what does that mean for me and my job, even though the rest of those lenses are relevant to someone else?

Most companies are in a very early stages of articulating a Data Strategy with capital ’s’. Many of them have a technical data strategies, a marketing data strategy. A large corporation may divide data strategy into a Business Strategy, a Technical Strategy and a Functional Strategy. A Business Strategy is what sales should do with data, what should product do with data, what marketing should do with data, etc. A three-tier approach would be: what is the overall Business Strategy that we are trying to implement using data, which really is only expressible as how you are going to drive top line or the bottom line, and all the things that come out that. The first chunk of that Data Strategy always has to do with either revenue or profitability and some other metrics, almost everything else is taken out of that bucket. That’s the conversation we want o start with, and take the conversation to what does that mean for you to develop Technical Data Strategy for CDO, or Functional Strategy for the head of marketing, Functional Strategy for head of operations, or back to the revenue or profit for the COO, etc.

Business Data Strategy in a very large corporation may have three pillars. The first is to use all the data that was available to us to drive efficiency in the overall business, to either increase profitability or more likely free up additional margin for investment in other areas. How does pillar I of the Business Data Strategy manifests itself? Different aspects: we can use data to better manage customer churn, we can use data to better, more effectively optimize capex, we can use it to drive incremental revenue across products. What, however, is the tangible asset that you get from business strategy? The only metric that was relevant to measure Business Strategy is hard $ return. The Business Data Strategy is not manifested in dashboards or tools. But how do we link the fact that we changed something and that resulted in $ change? To drive that change we can drive better decisions and better decisions were powered by a set of tools that were provided. But the whole point of implementing a strategy is that tolls are useless without the results. The outcomes is what will make business take notice. Business strategy is focused on business outcomes. But taking it one level deeper, what is causing the change in revenue? For example, if you found that decreasing network latency by 10% may increase review by 20%, how do we tie it back to Business Data Strategy. Do we need to invest in engineering or hardware to accomplish this? With numerous projects going on in parallel. At the executive level, you just talk about business results. Each project can drive ROI calculations, but the infrastructure team would supply the fuel for the individual teams. The investments in infrastructure are strategic for the business. The overall infrastructure spend is strategic, because companies see that more likely there will be return; top down approach: we need big data infrastructure. Once the individual teams brought RIO, the executive noticed that the investment in infrastructure was a slam dunk and they were asking how to get more. One of that investment allocations strategies is as follows. One of the teams creates pipelines and datasets, other teams tag the usage of the application, then the ROI can be tired to infrastructure work.

In a medium size companies sometimes there is no luxury of huge investments that can be swallowed by sporadic, successful teams. How do you get to the level of detail of what things are needed to be done to have largest impact? This comes down to the vision of a CEO. The first pillar of the strategy, to make everything more efficient is easiest to understand. Test driving individual squad teams approaches is important in making surety correct approach is followed. In big companies funding may not be a problem, but getting access to data, especially if there are privacy laws involved, is. Iterative approach helps in defining what data could be accessed.

Pillar II of the strategy is, essentially, making the customer experience better. The return on that investment is a happier customer. This pillar is much harder to sell. Pillar III is generating revenue off of data, directly. Creating a data service from which a revenue could be created. In very large company, if you can attach revenue to something, it is very easy to understand. If you say the customer will be happier, people look at you say, I don’t know if this is a good idea.

The overall Business Data Strategy for the company can have a three pillar business strategy that all hide off the infrastructure, but all pillars leverage the platform enabled data to different corporate ends. The justifications is each case are very different. Architectural improvements make the business strategy possible and allow efficiencies at the business level. The difficulty of associating business, technical and functional strategies can be tackled by both deductive and inductive reasoning. Most businesses tend to be very reductionist — we need to fix this problem, that problem, and they are unaware of the gap that exists in some areas. What generally tends to happen is, data strategies often starts top down—we need to do this more efficiently—can you do this with this thing I heard about called big-data? But the more you understand about the data, the more bottom-up approach happens. Now we realize we can do so much more with this data. (Feedback: this is spot on.)

The Data Strategy is not one thing. It is a framework in which there are entire portfolio of data-driven business improvements. Those improvements materialize themselves in a variety of ways, but the Data Strategy has to be a framework that encompass lots of different types of improvements to the business, whether it is terms of metrics, better marketing campaigns, better Capex decisions. People need to progress from: I want to solve problem X to: we can solve more problems. The Data Strategy is a framework that encompasses many things one of which is developing a language that allows to communicate about data opportunities. The language allows people to understand where opportunities lie and which part of the framework they belong to. This classification allows everyone to create shared understanding and allows them to accept that event if something does not show on a dashboard, it is useful in a different context. This important not get the first no, and but to gain understanding and acceptance. Creating the framework is important up-front to start people in neutral state instead of ‘no’.

Adoption metrics, productivity metrics and satisfaction are often taken into account when measuring success of platforms. When messaging to executives, the $ is often expressed transitively to increased productivity of platform users.

October Session IV Topic

Technical Data Strategy