Discover PerformanceHP Software's community for IT leaders // November 2013
Snapfish: Connecting data for fun and profit
HP’s photo site turns a small Big Data experiment into a full-blown customer experience makeover.
Great intelligence into your customers’ behaviors can drive real revenue gains, but not if you can’t get a Big Data analytics project off the ground. You’ll need enterprise-wide support for technologies that transcend yesterday’s data silos and slow feedback loops.
As HP photo sharing site Snapfish recently learned, the best way to garner that support isn’t with a full-court press of high-level evangelism and gloom-and-doom PowerPoint decks. Instead, a quiet, low-cost demo let the technology speak for itself.
“We thought of it as a science experiment,” jokes Venkat Mynampati, senior product manager at Snapfish. With the support of engineering director Manas Chaliha, Mynampati’s team used the free community edition of HP Vertica to create a quick trial of potential site customization improvements.
“I gave free access to multiple people, and everyone who saw it realized the ease and speed with which they could use it to solve problems,” Mynampati says. “Suddenly we got widespread adoption.”
Support from across the business made for a fast and painless analytics implementation that’s helping Snapfish sell more product and improve the user experience.
Leveraging the clickstream
As a web-based retailer, Snapfish was held back—not by its customer data—but by an intelligence feedback loop that couldn’t really know customers, or anticipate their needs. Mynampati’s team saw golden opportunities that were untapped:
- Static product promotions failed to capitalize on customer behavior or preference data.
- Customer photos sat unsorted in albums instead of being pre-populated into ready-to-buy photo books.
- A time delay prevented popular card designs from being featured promptly.
Snapfish’s Vertica experiment began with a simple desire: to use its customer data, notably real-time clickstream, to personalize the customer experience, to offer more delights, and make it quicker to buy them.
Over the next 18 months, Snapfish expanded the Vertica environment one virtual machine at a time. Today, the system processes more than 21 billion records, which Snapfish is using to reduce cart abandonment, raise per-customer revenue, and keep users on the site longer.
Building revenue through customer experience
The keystone of the recently launched initiative is a product recommendations engine that uses multiple sources of customer intelligence to show users the products and card designs most likely to interest them. Additional features include:
- Highly targeted email offers with reduced lag time to contact.
- Pre-populated photo books using customers’ most recently uploaded photos—solving the big problem of customers not having time to get started on a photo book.
Mynampati boils down the expected business value of the recently launched initiative into three simple pillars:
- Save money—With ad-hoc reporting provided by Vertica, Snapfish has eliminated the costs of external vendors’ static reports and analysis. They’re also reducing marketing costs by relying on targeted email messages rather than expensive, low-conversion mass promotions.
- Make money—Snapfish conservatively expects the new recommendations engine to initially improve conversion rates by 6 percent—a figure that should net several million dollars in additional annual revenue. But the new analytics power will let Snapfish optimize iteratively, so even greater improvements are expected over time.
- Increase site stickiness—Using a combination of behavior and psychographic profiles, Snapfish provides a customized experience that takes individual family situation, personal style, and regional trends into account. Not only will visitors stay longer, but Snapfish expects customers to share photo albums and custom products with friends and family more often, which results in greater per-album revenue.
Targeting faster performance
With the Vertica system in place, many other avenues to an improved customer experience have become evident to the analytics team. Among them: If Vertica can predict what types of products a certain visitor might like best, couldn’t it also predict where they might go next, based on usual behaviors and that customer’s own past site usage? With that, Mynampati explains, you can realize a dramatic improvement in page load time—perhaps not on every page view, but on most of them.
Thus, Snapfish is harnessing Vertica for performance optimization, developing a two-phased initiative to dramatically cut page load time.
- Phase 1: Measure load times for every page, using the data to identify and remediate pages and services with the worst performance.
- Phase 2: Build a predictive model of what pages a given user might go to next, pre-build the most likely pages on the fly, and send them to edge cache, where they can be served instantaneously to the user when requested.
“With every millisecond of delay, we lose customers,” says Mynampati. “We expect this to cut the time users wait for page downloads on Snapfish by 50 to 60 percent.”
An analytics evolution
Snapfish expects the site performance enhancement to be the first of many extensions that leverage its investment in Big Data analytics—including more features that touch the customer experience and others that don’t. But by starting with a simple demo, building support, and then focusing on its critical business objectives, Snapfish proved that strict priorities and restraint can be the keys to success in the transition to Big Data BI.
To learn more about how Snapfish is harnessing the power of Vertica, listen to the Snapfish webinar in the Value of Vertica series. And visit Vertica.com for more on the power of customer analytics in retail and beyond.
HP Software’s Paul Muller hosts a weekly video digging into the hottest IT issues. Check out the latest episodes.
Welcome to a new reality of split-second decisions and marketing by the numbers.
Looking toward the era when everyone — and everything — is connected.
Introduction to Enterprise 20/20
What will a successful enterprise look like in the future?
Challenges and opportunities for the CIO of the future.
Dev Center 20/20
How will we organize development centers for the apps that will power our enterprises?
IT Operations 20/20
How can you achieve the data center of the future?
What the workforce of 2020 can expect from IT, and what IT can expect from the workforce.
Preparing today for tomorrow’s threats.
Data Center 20/20
The innovation and revenue engine of the enterprise.