Realizing the true potential of ML means breaking out of our digital advertising framework
“Data-driven thinkingIs written by members of the media community and contains new ideas about the digital revolution in media.
Today’s column is written by Ali Manning, co-founder and COO of Calice’s custom algorithms.
In the future, the best performing and best performing brands will be the ones that can predict the future.
And there’s no reason these brands can’t better anticipate what consumers want than even the best-capitalized tech giants.
It might sound far fetched, but machine learning is about to make it a reality. Machine learning will transform the practice of marketing while resetting the relationship between marketers and tech giants along the way.
Where we have been
In the years to come, brands will compete in an entirely new playing field, and their ability to gain market share will depend less on those who spend the most on media than on those who can create the most powerful proprietary predictive technologies.
Netflix is ââalready predicting what we’re in the mood to see informed by the feedback it receives from its marketing flywheel. There’s no (good) reason why this shouldn’t work for a big brand too.
Historically, marketers have competed on what they’re meant to be good at. For the most part, that means focusing on a product’s USP (“This toothpaste makes your teeth 50% whiter!), On services (LL Bean, for example, has a very generous return policy), on the awards (“This product is a great deal”) or on the stories they can tell about themselves (Nike inspires you to be successful; Pepsi makes you feel young).
These factors are always important, of course, but brands have also gone to great lengths to accumulate as much consumer data as possible in order to get their messages to the right targets and hopefully keep their customers in the fold. .
Brands are striving to be the best at targeting and closed-loop marketing. Yet advances in machine learning promise to change everything.
Where are we going
Sarah Rose, Senior Vice President of Global Digital Operations, Data and Platform Operations at IPG Kinesso, recently wrote in an Ad Exchanger column that âmachine learning is the first step in optimized data science applicationsâ.
In other words, whether through ML, AI, or computer vision, machines can do things faster and on a larger scale than humans.
This is 100% true, and yet the concept can seem rather abstract. It’s not hard to read this language and think, âHey, isn’t buying programmatic ads already doing that? Is this slightly better targeting? “
To truly understand the potential impact, we need to think outside the digital ad boxes and Consider what ML has already done to transform industries such as finance, medicine, and sports.
Take medicine as a shining example, where we are already seeing personalized cancer therapies based on genomics.
Likewise, brands can create their own personalized predictive technology that integrates thousands of variables and fully guides decision-making.
You might ask, âWhat about creativity? And the answer is that creativity will always matter – a lot. Maybe even more. There is no reason why the combination of creations based on sophisticated predictive models and testing with new, proprietary ML technology couldn’t be as effective, if not better, than the biggest players in advertising technology.
ML in action
Here are some hypothetical examples of what this might mean in practice.
Imagine that a competing wireless brand developed improved network coverage in some areas. The brand has a chance to gain market share, but only if it is able to inform specific customer segments in specific areas. ML can help boost this brand’s performance when targeting consumers by taking into account a set of custom variables related to location, revenue, and current device type.
TTo be clear, it’s not just a matter of showing ads in certain geolocations. I’m talking about creating an advertising auction strategy for over 40,000 zip codes while overlaying a customer’s income bracket for each.
Now imagine another wireless brand, this time the national leader. This company is less focused on growing market share, as its best growth path is to sell its current customers on larger service packages and expanded family packages. In this case, the brand can use a different set of custom variables to target existing customers, such as each person’s existing plan, how long they’ve been on, and how many devices they have in their home. All of this information can be fed into machine learning software to achieve much more relevant and profitable results.
The DL on the ML
This future is not so far away.
Marchaeologists have absorbed so much data that they feel like they don’t know what to do with it. And that’s because they don’t do it yet. But as ML technology takes hold, its predictive power will increase exponentially depending on variables that can be incorporated into a model.
It also promises to unearth dozens of needle movement variables that humans might never see.
ML tools get smarter and more powerful the more you use them. This sets the stage for brands to compete on tools that can learn fastest rather than storage space or voice share.
And think about this: Once brands have their own ML, they’ll know more about their own customers than a single walled garden or a brand’s internal technology.
Plus, thanks to advancements in data storage and players like Snowflake, what used to take weeks and cost millions can now be done in hours at a reasonable cost.
This has the potential to radically change the dynamic between traders and the duopoly.
It’s not that brands won’t continue to advertise on these platforms – they might even be doing more. On the contrary, marketers won’t feel like their own customer data and campaign data is siled. They will have their own in-depth understanding of their clients and what makes them tick, which will give them more leverage.
ML doesn’t just promise to change your business – it promises to redefine the business you are in. It’s a future I think most marketing managers would commit to.