How Predictive Analytics is Transforming the Ad-Tech Industry
The potential of big data is thrusting AI and machine learning into an indispensable role.
- By Ira Cohen
- January 18, 2019
It's not possible to think about today's advertising industry without appreciating the pervasive use of ad tech, the tools that enable brands to target, deliver, and analyze digital advertising. For roughly the last six years, the two primary ad tech tools, programmatic ad campaigns and exchanges, have increasingly been the sails propelling a $628 billion global digital advertising industry.
In theory, by using programmatic ad campaigns, agencies can target relevant audiences and place ads on thousands of sites. In practice, programmatic advertising has not been without its challenges, from fraud to data privacy issues under the GDPR.
Today, the narrative around data is changing. If big data was once the goal, today the focus is on making data actionable. In the words of Google's chief decision intelligence engineer, Cassie Kozyrkov, "Data science is the discipline of making data useful." That challenge has been taken on by data analysts, tasked with connecting the dots between the data ecosystem and the business ecosystem.
The link between these two worlds is predictive analytics.
Not surprisingly, AI, machine learning (ML), and predictive analytics are now poised to completely reinvent ad tech for a very simple reason: volume. With data accumulating at an exponential rate, it's simply impossible for data analysts to extract relevant and timely business insights without autonomous analytics. Effective predictive analytics means it's possible to monitor bids and results and suggest the optimal bid price or whether a bid price is reasonable -- all in real time. This was impossible just a few years ago.
If this technology is so powerful, why is ad tech still not leveraging predictive analytics and data mining to their full potential?
It's Always About People
AI and machine learning don't change a basic fundamental of advertising: it's about the people. A word that comes up again and again in marketing is authenticity. Despite ad tech, brands must look at their audience not in terms of what they are but according to the things in which they believe. Consumers are looking for timely, helpful, and relevant content -- even ads. Done well, predictive analytics can ensure real-time highly personal content.
It's not just about the audience, either. Automated systems don't negate the need for human monitoring of both the processes and the results, as tempting as it is to run completely automated campaigns. AI will fundamentally revolutionize programmatic jobs, but human intelligence is still crucial to realize its potential. An automated system without monitoring is only half a system.
The fact is that programming is complicated, with tens of thousands of campaigns running every minute. Most companies wouldn't know if there's a problem until there are business repercussions because they can't see all the variables. They might be looking at effective CPM, bid requests, or simply looking at the success rate, but that metric might not tell the whole story.
Managers need to be on top of the campaigns and budget. If decisions are being made by algorithms, who manages or verifies the process? Has the algorithm lost money? How long did it take to detect a problem? Is your system providing root-cause analysis and real-time alerts? These are all issues that must be considered if you want to successfully leverage AI tools such as predictive analytics.
Predictive Analytics Is More Than You Think It Is
Predictive analytics is more than just tracking historical customer behavior and then tailoring a response based on such behavioral indicators. Predictive analytics is about getting the most out of your data. Are you?
If you're dealing with potential customers, for instance, predictive analytics can help you identify customers who most likely intend to transact by using data from existing customers (referred to as "lookalikes") so you're not wasting resources on poor prospects.
What happens when you realize you're working with incomplete or dirty data? There are workaround processes. Data stewardship, for example, which can include CRM (data) diagnostics, cleaning (removing incomplete, redundant, error-filled, or obsolete records), and detail enrichment (adding missing fields to existing records from external sources) will help you maximize your data for more accurate business insights. You can also match third-party data with your own data to create more accurate personas.
The Dividends of Data Privacy
It's been four months since the GDPR was rolled out, and marketers are still trying to understand all of its implications. One thing is certain: privacy laws such as GDPR are affecting programmatic advertising, which is based on tracking and targeting individuals and their Internet behavior.
Nevertheless, the fundamental belief that personal data belongs to the customer and not the enterprise is reasonable, and it should motivate organizations to rethink their relationships with customers. That's not a bad thing.
Predictive algorithms already offer the ability to leverage aggregate, anonymous data that will benefit the enterprise while respecting personal rights. Companies should also develop incentive plans to motivate customers to offer data with mutually beneficial results. Although privacy laws may seem restrictive, respecting customer data offers the potential to turn cynical buyers into responsive fans which can only benefit both customers and organizations.
Today, the potential of all that data is thrusting AI/ML into an indispensable role, not just as an ancillary feature. With predictive analytics, everything can be monitored, so enterprises can take bigger risks and reap bigger rewards. Predictive analytics also enables organizations to focus more on customer satisfaction. That's significant because despite the new technology, the fundamental goal of effective marketing hasn't changed: better human-to-human connections that enable you to more persuasively tell your brand's story.
Ira Cohen is co-founder and chief data scientist of Anodot, in charge of inventing and developing its real-time multivariate anomaly detection algorithms. He holds a Ph.D. in machine learning from the University of Illinois at Urbana-Champaign and has over 15 years of industry experience. You can reach the author via email, Twitter, or LinkedIn.