Constant Contact

Matt Laudato, Director of Big Data Analytics, Constant Contact, Matt Laudato is Director of Big Data Analytics at Constant Contact, where his team uses the latest Big Data tools and techniques to provides insights ...  More >>

Small business can mean different things to different people – it might mean generating revenue of less than 500,000 dollars per year, or employing fewer than 10 people. Regardless of the definition, most small businesses have several things in common: they were founded by one or two passionate people, they are laser focused on using that passion to fuel their success and, most relevant to our discussion here, they are time and resource starved. The typical small business owner is CEO, CFO, VP of Sales and, Chief Cook and Bottle Washer. Most importantly, they are experts at doing more with less.

  The benefits are potentially great, and as with most transformational technologies, a little foresight and planning go a long way

The other important truth about small businesses is that they have been historically underserved by the software and services industries, and so typically make do with whatever they can run on their office PC. The emergence of Software as a Service (SaaS) companies has changed the landscape, however. SaaS offerings provide small businesses with cost-effective software functionality and open the door to using Big Data Analytics to help both the SaaS provider and their customers.

The key question we address here is how can SaaS companies use data and analytics gathered from hundreds of thousands, or even millions, of customers to help each of those customers – and to grow their own business – at scale, economically, and strategically?

The business I’m most familiar with (SaaS digital marketing tools) presents a unique opportunity for doing just this, with lessons that can be applied to other SaaS businesses. Constant Contact’s more than half a million customers use our tools to send millions of email campaigns every year to over 300,000,000 email subscribers. These customers and their subscribers cross every industry, demographic, and geographic boundary imaginable, and the data that we collect has historically languished – unknown in many cases, and certainly under-leveraged. When we made the decision to invest in Big Data, it became clear that by sticking to a few guiding principles, we could do what few other enterprises had done: transform our business by leveraging data and analytics at scale to both help our customers and help ourselves. This transformation is well in progress, and our principles are worth sharing.

Quick wins add up

It’s easy to get caught in the trap of the latest trend, and Big Data is no exception. Too many enterprises sink a great deal of money in building up Hadoop and related infrastructures, but are rewarded with little tangible business value. We took the opposite approach and focused on implementing a minimal infrastructure to solve very specific, high-value problems. For example, we knew that individual email recipients had inbox viewing behavior that was time and day sensitive, but we hadn’t built a model that could be used to determine the best time for our customers to send emails. This initial use case leveraged data in a way that we hadn’t before – across all customers and all recipients – and not only improved our customer open rates by nearly 7 percent, but proved to executive stakeholders that a small investment could have material, positive impact on our business. That set the stage for further, more strategic, investments and projects.

Keep it simple for customers

Small business owners are experts in their domain, and pretty darn good at all of the other day-to-day work required to keep the lights on at their shop. But they are not, in general, data and analytics experts, nor should we expect them to be. We consciously avoided trying to sell customers on ‘advanced analytics’ or fancy reporting. Instead, we asked a much harder question: how can we transparently bring the power of our back-end analytics to bear on their problem, at the moment when the problem was most relevant? This led us to a deep analysis of our product, and of the friction points that customers encountered when using it, as well as how those points then negatively influenced customer success – namely, the creation and sending of a successful email, where success is measured in open and click rates. By doing this, we were able to focus on building a few key analytical models that helped customers craft more effective emails.

Architecture and governance matter

There is always a balance when building out new infrastructure like Big Data between quickly realizing the potential of the technology, versus taking time to address important strategic areas such as information architecture and data governance. We realized early on that Hadoop, because of its computational power and near-linear horizontal scalability, could replace much of our current ETL processes that absorb and transform data from production systems into data warehouses and other back office systems. But we were strategic about the opportunity and chose to implement architecture and governance improvements in a ‘just in time’ fashion. Specifically, when new data sets became available from production systems – ones that we previously could not have analyzed due to their size or change velocity – we asked two critical questions:

• Does this data set represent a new architectural pattern, or is it a variation of an existing pattern?
• How can we use the project for obtaining this data to advance our data governance strategy?

By asking these questions, we were able to balance the temptation to ‘throw everything into the data lake’ and producing a data swamp in the process against the strategic opportunity to advance both our information architecture and governance processes. Doing this slows down the pace of obtaining the new data set, but is more than offset by the incremental improvements in architecture and governance.

Big Data, when you strip away the buzzwords and the hype, represents a highly scalable, cost effective way of managing large data sets for operational reporting and deep analytics. By focusing on the guidelines presented here, Constant Contact has been able to improve customer satisfaction, advance our information architecture and governance, and grow a new set of data science and predictive modeling muscles in the organization. Other SaaS companies have the opportunity to do the same. The benefits are potentially great, and as with most transformational technologies, a little foresight and planning go a long way.