Every campaign has winners and losers. The biggest winner since 2012 has been big data. Tactics pioneered by Republican and Democratic campaigns alike over the past half decade will determine the next president and help businesses operate more efficiently, said former Republican data guru Zac Moffatt.

Moffatt was 2012 Republican presidential nominee Mitt Romney’s digital phenom and is the co-founder of analytics company Targeted Victory. Romney, a business mastermind and strong proponent of of data and analytics, hired Moffatt to discover and convert new voters. At the time, analytics were nascent on campaigns but widely used by business. “[Romney] makes decisions based on metrics,” Moffatt said during a phone conversation. “Our numbers showed [a large percentage] of voters rarely watched live TV and lived online.” His job, Moffatt explained, was to make big data accessible, understandable, and usable for campaign staffers.

READ: Big decisions with big data (Tech Pro Research)

Though Romney lost the election, Moffatt’s analytics and data team pioneered techniques widely used today by campaigns and by business. Although the 2012 Republican data product, ORCA, was widely criticized after the election, Harper Reed, Moffatt’s counterpart on Obama’s 2012 campaign, respects his work and explained, “When [Narwhal] failed, it wasn’t really a big deal.” The media exaggerated [the product’s] failings, Reed said. “Zac knows data, and his work translated well to [the private sector].”

After an early career in politics, Moffatt launched Targeted Victory, a data firm that combines business analytics, email marketing, and social media to help campaigns and other organizations raise money. According to the company’s website, since 2009 Targeted Victory has raised over 200 million dollars, sent over 1 billion targeted email messages, and added over 30 million social media users to its database.

“[Our company] is an audience-driven technology company focusing on programmatic media buying across all screens,” Moffatt said. “We’re rooted in politics, but we serve corporate, international, and issue advocacy clients and also offer a host of technology products available to any right of center candidate or cause.”

Moffatt explained to TechRepublic how the analytics industry has evolved since the 2012 campaign and what business can learn from observing how presidential campaigns leverage big data.

What are the biggest technological changes between the 2012 campaign and the 2016 campaign?

The biggest change from 2012 is the rise of programmatic and addressable advertising, which now offers candidates the flexibility to run multiple campaigns to multiple audiences with maximum reach and efficiency.

Let me put that into perspective. In 2012, Governor Romney’s presidential campaign was the most sophisticated programmatic media buying campaign in Republican history. By the fall, we were targeting three core audience segments in nine key states. This year, our client Ted Cruz’s campaign was targeting 173 different audiences in Iowa alone.

That means campaigns no longer have to spend millions of dollars on overly broad and wasteful television ads. For example, we released a study showing that in 2014, congressional campaigns wasted 75 cents out of every broadcast dollar delivering ads to voters in the wrong district(s).

2014 marked the first time campaigns had the kind of data needed for programmatic targeting, but no one yet had built a system to leverage [data]. This year, campaigns not harnessing this technology are behind the curve, fail to reach key voters and, in the process, waste precious financial resources.

So, programmatic targeting is vastly improving the ability of political campaigns to more efficiently reach their target voters.

Maximizing efficiency and reach unleashes the creative potential of your campaign budget. In a close election, that’s a huge advantage.

How was your platform coded and developed?

Our application team works primarily with Ruby on Rails, and we are hosted in Amazon Web Services (AWS). We use Amazon’s infrastructure for operations support and scalability. So, for example, we run our transactional databases in Relational Database Service (RDS), rather than managing them ourselves. We use Elastic Beanstalk to deploy our apps and automatically scale with load.

On the data side, we’re again hosted in AWS but are moving towards using Scala. We like Kafka for queuing, Elasticsearch for unstructured data and fast search, and Spark for both batch and streaming data processing. We use Databricks to provision our Spark systems, as well as for its analytics environment.

Finally, data from all of these systems, as well as data from all of our vendors, flows into our Amazon Redshift data warehouse, where our analysts and business users typically work in Tableau.

In what ways does big data help inform offline marketing like microtargeting?

What we’re seeing is that campaigns are getting much better at harnessing their data and using that data to drive their targeting determinations. Smart, efficient campaigns have moved from buying a specific site to being audience-specific.

The old model was about reserving inventory and attempting to treat digital video targeting like broadcast television. The new model is about leveraging the data and infrastructure advantages of digital.

Whether a campaign is focused on list-building, fundraising, persuasion, or Get Out The Vote efforts, a focus on audience-based advertising allows them to reach more voters for less money.

What does the post-2016 political tech landscape look like?

Looking ahead, successful political campaigns will have embraced a culture of data driven decision-making. Winning campaigns at every level will share one thing: they will have successfully … embraced outcomes driven by data and will use zero-based budgeting.

2016 is the year when political campaigns finally began to understand that developing smart ad content across every screen was more valuable, and more efficient, than blindly buying expensive TV broadcast spots. I think this change will certainly drive campaigns in 2018 and beyond.

The other important take away this cycle is we continue to live in a political environment where we’re data rich, but content poor. The explosion of smart targeting tools allows us to reach off the grid voters and others who were once unreachable.

What lessons can business learn from observing how campaigns use data?

Campaigns and businesses are different in many ways, mostly because a political campaign has a finite ending. On Election Day, you either win or you lose, so your budgeting reflects that. Procter and Gamble? Their marketing budgets and business decisions are based on a three, five, or ten year look-ahead.

That said, there are many lessons businesses can learn from campaigns. For all their differences, the biggest similarity between campaigns and businesses from an advertising standpoint is they’re both hyper-local.

SEE: OS convergence and the rise of the mobile office (Tech Pro Research report)

The most important lesson for any business or brand is that targeting specific audiences will always be more efficient than blanket advertising on broadcast television, which eats up large portions of your advertising budget with little to no ability to measure your results.

We’re also huge believers in cross-device targeting for corporate brands and campaigns alike. Cross-device targeting allows clients to deliver true sequential messaging across every screen. Over 50% of all web traffic is on a mobile device. The audiences across these screens are often more engaged on mobile platforms than on traditional channels. The data on this isn’t exclusive to politics, so it’s something any brand or business should seriously think about and act on.

Read more

Note: some quotes have been modified for clarity and brevity.