Personalization is becoming more and more important on every front as marketers develop targeted messaging and choose outreach channels for maximum effect. This is where data science comes in.
Data science is a trendy buzz term now, but at its core, it means testing and measuring, looking at the data, and making decisions based your findings.
Kaleb Ufton, director of technology at Ekoh Marketing, said that marketers can use data science to derive insights about what is and is not working with their campaigns.
“Almost all digital platforms provide a rich set of metrics that can be used to measure the success of an ad, even a specific headline or creative,” he said. “Data science takes the ‘gut feeling’ out of it and provides real data that lets us see what is working.”
It can also help with planning campaigns as marketers look at data from previously successful campaigns and show clients their progress.
Data science applies computational and statistical techniques on big data to identify groups and patterns, asking questions like, “Which customers exhibit a similar response to a message or channel?”
“By answering questions like these,” said Adam Lichtl, founder of Pacific Data Science, “The data-savvy marketer can begin to conduct A-B testing on specific groups to improve conversion. It may be that approach ‘A’ works better on groups 1, 2, and 4, and ‘B’ works better on groups 3 and 5. This sub-segmentation is based on data, but often involves subtle distinctions that are hard for a human to identify.”
Not too long ago, raw, simple data alone was often sufficient to stay ahead of the competition, as only the most progressive organizations were using it. Today, brands and agencies across the spectrum are inundated with data.
“It seems like everyone takes it seriously. As such, the key now has become how to analyze the data and properly use it in order to grow,” he said. “From a marketing perspective, this means analyzing the data so that you can put a marketing plan together to reach your exact demographic in a cost-effective manner. The competition is so tough, you don’t have the wiggle room anymore for mistakes,” explained Matthew Ross, head of business development for RIZKNOWS LLC.
Ross recommends taking a top-down approach, starting with getting to know your advertising platform extremely well. For example, if you’re using Facebook, first become familiar with the exact variables and filters you can use in your campaigns. That way, when you mine the data, you have an idea of what is useful and what is not.
“A recent study that caught my eye was from the University of Maryland. A professor-led team created a model to analyze user activities online and assign them a ‘grouping’ in order to optimize paid advertising,” he said. “They logged over twelve months of data relating to engagement, demographics, and visitation tendencies. At the end of the twelve months, the team was able to effectively target various segments of consumers.”
Tina Mulqueen, CEO at Kindred PR, said that many of her clients have been surprised to learn that thanks to payment systems and analytics software like Google Analytics, they have access to large amounts of consumer data.
“Understanding digital touchpoints like online payments or website clicks allow us to optimize those touch points to collect meaningful data,” she said. “For example, if you have your customers define whether or not a purchase is a gift, it opens up re-marketing opportunities for gift-giving.”
Asking the Right Questions
CMO of Engagio Heidi Bullock noted that the best CMOs not only report on performance, they can predict pipeline and how initiatives will perform. Data science, when applied correctly with the right data and expertise, can help create models used to answer challenging questions and even help predict outcomes.
“It really starts with being clear on what questions you are trying to answer. Once the question is clear, then teams can map out the data needed,” she said. “Typically, marketing organizations will have some of the data required – and that may be enough to be directionally correct with the analyses they have to run. If there are gaps – either with sources of data or with access to the data – that is the second aspect to identify. At that point, teams need to assess the costs involved.”
Laura Troyani, founder & principal of PlanBeyond, noted that one of the biggest opportunities for data science in marketing is in the realm of lead scoring and lead optimization.
“Lead scoring is the process of ranking leads based on their expected value to a business. Leads with high scores will often be given to sales to try and convert,” she said. “Leads with low scores will usually stay with marketing for future nurturing activities. Data science can play a key role in developing the scoring model that actually drives this inflection point.”
When organizations have large lead and customer databases, it becomes possible to run regression models on the marketing sources, website pages visited, and content downloaded.
“The models will help isolate the variables that most contribute to conversions, and therefore inform how a team develops their lead scoring model,” Troyani said. “It offers a data-centric approach to segmenting databases and optimizing when leads are passed on to the sales team.”
Alexei Yukna, director of marketing technology at 14 West, noted that marketers are faced with increasingly short attention spans in crowded spaces and must be agile.
“Those wishing to rise above must practice marketing at scale, and scale requires sophisticated technology to make decisions in real time,” he said. “Gone are the days of setting and forgetting with the occasional touch point to adjust, rinse and repeat.”
That’s why modern marketers must first have a clear view of the customer. Ensuring that all potential customer data is clean and verified will cut down on bad data and dead-end marketing efforts.
“From there, many modern platforms have rolled in AI aspects to their platform to engage send optimization, content recommends, display affinity, etc.,” Yukna said. “There are vendors that allow for API use in a real time scenario for inclusion if a marketer needs to develop a robust stack not of the enterprise variety. Everything from identity resolution online and offline, affinity modelling, and even event triggers can be found on the service level now thanks to SaaS platforms.”