We’ve come a long way since the Mad Men era of marketing and advertising. The modern marketer is as much mad scientist as inspired creative, thanks to the proliferation of data and tools to access and analyze it. Just as inventor Thomas Edison was able to run 10,000 different experiments, changing a little something here, a little something there, until he found the formula that worked, the best marketers today use their creative side to come up with hypotheses, and can then perform experiments to prove their usefulness. The science of marketing has become increasingly, well, scientific.

Given the revelation that hyper targeted marketing is more effective marketing, it’s no surprise that almost every marketing department or digital agency out there will tell you they’re “data-driven.” But at this point, “data-driven” marketing is more of a minimum requirement than a differentiator. Without data supporting them, marketing recommendations would be no better than opinions, and analyzing ROI would be imprecise, at best. But if every marketer is incorporating data into his or her campaigns, why is it that most marketing messages, no matter how they’re delivered to consumers, seem to be forgotten within a matter of seconds? And why is it that marketing funds get spent with executives and teams struggling to answer the simple question “What did we get out of that?”

Jon Brody, CEO and co-founder of Ladder, a growth marketing agency, says “The problem is that people have too much data and are making fewer good decisions because they’re sodata-driven. You need to be ROI-driven,” he says. Effective marketing is about optimization. That’s why Brody and his team, who have worked with everyone from startups backed by Y Combinator to Fortune 500 companies, approach marketing like scientists in a lab.

In short, he says, “We run marketing experiments to help businesses grow.” This type of approach has worked for Ladder, and it can work for any team that takes the following five tips to heart:

1. Don’t be afraid to take risks

When you’re experimenting, you’re going to find out what doesn’t work on your way to finding out what does. “Data driven marketers build hundreds of audience segments and cross them against thousands of targets,” says Dennis Yu, Chief Technology Officer at BlitzMetrics, a provider of courses on Facebook marketing, “But with so many combinations possible, each experiment can have only a few dollars of budget and a few units of effort against them. Most marketers cannot scale to this efficiency, since they are of the ‘I already tried that’ mentality.” Don’t fall into this trap. Keep experimenting, keep taking risks. Failure is built into the process. Learn from it, and proceed.

2. Take advantage of the tools available

There are new technologies appearing almost daily that allow marketers to measure performance. “Native platforms like Facebook are increasingly as good (or better) at optimization than any human or even the current crop of ad tech tools that were built to optimize what you’re already doing,” Brody says. The opportunity that these platforms present is enormous.

Larry Kim, CEO of mobile marketing software company Mobile Monkey and a popular speaker at digital marketing conferences, is one of the top marketing “mad scientists” of our age. He uses tools like WordStream to give him the data he needsto analyze how large the rewards are for having an above average CTR (clickthrough rate) in Google Adwords, leading Kim to conclude that “Google so greatly rewards high CTR/Quality Score ads (and conversely penalizes keywords with lousy CTRs) that an awesome hybrid solution becomes apparent: use your content marketing efforts to cover informational keywords with SEO content and commercial keywords via PPC.” Sound mad? Don’t worry, read Larry’s case studyand you’ll see how his experiments can help you save a lot of money on your next paid search campaign.

3. Don’t work in silos

You’ll be using much of the same data to inform all aspects of campaign development — from strategy, to creative, to execution, to placement. When your team works closely together throughout every step of the development process, each team member gains a better understanding of how his or her work affects the efficacy of the entire campaign and drives ROI. Every member of your team should be ROI-driven.

4. Remember what you learn

Your team shouldn’t have to reinvent the wheel every time you begin a new project. Ideally, you should be keeping track of what you learn — what works and what doesn’t for certain types of businesses or industries, how audiences respond to certain messaging, what tools are most effective — so that you’re able to start building some institutional knowledge that new employees can later tap into.

5. Don’t stop optimizing

Just because you’re seeing a lot of success with a particular campaign doesn’t mean you should rest on your laurels. There’s always room for improvement. Plus, the more you experiment, the more you learn. If you’re keeping track of that additional insight (see above), you can apply it to your next campaign.

The more you experiment, the better understanding you’ll have of which data actually matters. Moreover, collecting additional data that ties marketing spend to campaign performance — across the marketing funnel — will give you a better idea of how to best spend the money in your marketing budget. That’s what being ROI-driven is all about, and that’s what leads to real growth for your company.

Josh Steimle is the author of Chief Marketing Officers at Work and the CEO of MWI, a digital marketing agency with offices in the US and Asia, and despite being over 40 can still do a kickflip on a skateboard.