NOTE: GrowthBeat is less than 2 weeks out! VentureBeat is gathering the best and brightest in modern digital marketing to help declutter the landscape, simplify the functions, clarify the goals, and point the way to success. Get the full scoop here, and buy your tickets while they last.
This post is from LinkedIn data scientist Jim Baer, who’s speaking this afternoon at VentureBeat’s DataBeat conference in Redwood City.
Today, businesses have access to data far beyond anything data-focused corporations had 20 years ago. The massive amounts of valuable data garnered from things like staff behavior and customer interactions can become a company’s biggest competitive advantage.
But how many of us are making the best use of that data? A data-driven approach to business means using all that information to optimize existing business goals and investigate new possibilities. This can mean conducting frequent experiments to make sure you’re selling as much of your current offerings as possible. Such tests can be as simple as checking whether to put the mustard on the shelf next to the ketchup or the hot dogs or as sophisticated as testing whether a feature personalized for each visitor to a webpage can raise both engagement and revenue.
You can also use data to make recommendations, inferring what will bring the greatest value to your customers based on their history and characteristics. At LinkedIn, we do this to connect our members with opportunity, whether it is identifying job openings that most suit a member’s career experience or by recommending groups of like-minded professionals to engage in expert discussions.
Becoming a data-driven business can seem daunting, especially with a crowded market of companies selling products and services that promise to unlock the power of your data. In my experience, there are four foundational supports necessary in crafting a data-driven approach to business, although your level of investment in each can vary widely depending on your company’s goals and resources.
1. Build the Right Data Infrastructure for the Company’s Goals
Data infrastructure is the underlying technological plumbing that collects, transmits, stores, and delivers data to be leveraged for monitoring the business and understanding opportunities. Without a solid data infrastructure there will not be a reliable source of data to guide decisions.
However, there is no one-size-fits-all solution to creating a data infrastructure; there will always be trade-offs between the cost of collecting and wielding data and the benefit for business goals.
For example, a gaming company may want to collect all of the data on how users play its games in order to create effective features and grow the business. This will require investing in a huge relational database that allows those building the games to ask a broad variety of questions. However, if a winery is simply trying to understand how many people are visiting their website and what types of actions they take on the site, they can rely on inexpensive (or even free) services to get those answers. They can then focus their resources on tracking the elements in their wines that best resonate with customers to craft future best-sellers.
In any case, approach the data infrastructure investment decision with a specific set of goals in mind, but retain flexibility in the system wherever possible. As your business grows and evolves, your needs will change (likely increase) and may require adjustments to infrastructure.
2. Democratize Data Throughout the Company
Data infrastructure investments won’t provide value unless the data collected is accessible. The more people who can access and use data to measure performance, evaluate improvements, and learn about the business and customers’ patterns, the better.
Democratized data allows employees outside of the technical departments to critically evaluate the company’s data and ponder implications for the business. It allows the right person, with the best context on a specific area of the business, to directly evaluate whether the data supports expectations. It also empowers a broad base of employees to find anomalies that can be important opportunities or warnings.
For example, members of one product team might want to explore the sales impact that another team saw when they changed serving sizes for a beverage product. Or, for an online dating website, the same sudden rise in site traffic that delights the monetization team may concern the security team, which suspects a bot attack. The key is to get data into the hands of those who recognize what it means and for that data to correspond to clearly-defined metrics.
3. Enable Experimentation
Experimentation tools provide the ability to test innovations and treatments and learn from the performance data before making big bets. The simplest approaches to experimentation are before-and-after types of evaluations to understand the effects of making a singular change to a test object.
However, most business questions call for more complex experiments, such as which features an auto insurance company should add to policy offerings to increase renewal rates among customers. This may involve multiple test groups of policyholders and a slew of different features to test and compare results.
The best experimentation systems will streamline the creation and tracking of test groups, treatments, and results to help simplify the process and scale it across an organization. But even a well-designed testing platform needs careful experiment design to maximize the opportunity for genuine learning.
4. Foster a Data-Driven Culture
A data-driven business culture requires the infusion of data to optimize familiar processes; it also requires a company-wide philosophy of innovation and experimentation, where employees are constantly seeking opportunities for new breakthrough products or features.
You can foster a data-driven culture by always asking for and consulting the data when making decisions. This works best when data is demanded by top-level employees, requiring hard numbers to back up claims of the benefits that a new program or feature will bring. In a data-driven culture, you’re always asking the question “What do the numbers show?”
For example, at LinkedIn, the heads of each product group give a weekly presentation to executives in which they present the primary metrics for their business lines and discuss any notable changes from plan.
The tools and infrastructure laid out above can be made or bought to fit your business goals with various levels of expense and expertise. However, the investment in these will not yield the fruit of data-driven success unless you also establish a company culture that requests evidence from data as part of the standard decision-making process.
As we enter a new era where data is more easily accessed by companies of all sizes, those who begin to leverage the massive amounts of unique data for their company will enjoy a competitive advantage. Those who don’t will eventually be left behind.
Jim Baer is senior director of Data Science at LinkedIn.
We're studying digital marketing compensation: how much companies pay CMOs, CDOs, VPs of marketing, and more
, with ChiefDigitalOfficer. Help us out by filling out the survey
, and we'll share the results with you.