But what is AI, really?
In short, AI is a computer’s ability to consume data, make predictions, and perform actions to achieve desired outcomes.
In practice, it’s a mechanism for turning ‘input A’ into ‘output B’ at massive scale. For example, turning audio clips into transcripts, turning English sentences into French, or turning car GPS information into insight on traffic hotspots.
The same is true for advertising. It can, for example, turn our definition of the perfect target audience into a bidding strategy. AI is simply a way to handle data — masses of data too large for human beings to keep ordered — and that capability is crucial in a marketing landscape in which four of every five display advertising dollars are allocated programmatically in the US and similar rates of growth are seen across the world.
Marketers achieve the best results with artificial intelligence by clearly articulating desired outcomes for an advertising campaign that map to a company’s larger strategic goals.
First, the marketer needs to understand the brand’s business goals from which marketing KPIs can be devised and achieved.
Let’s look at one easy example. Suppose a marketer wants to sell high-end appliances that retail for $1,500 and is willing on average to spend $200 on marketing to sell one.
Once the definition of the goal is clear — and it can take a lot of work to get there — we can explore ways in which that $200 is best spent.
In devising our plan, we must ask what we know about measurable actions that lead to sales. These actions include things such as signing up for newsletters, showroom visits, or assembling options for the appliance on a website.
We can then build a program with steps that map to the goal. We can even weight different components — a showroom visit is likely worth more than a website visit, for example — to make sure we favor steps that lead to desired outcomes.
We can then construct AI algorithms to feed into appropriate platforms, such as exchanges or DSPs, likely incorporating other instructions or constraints: a list of white-listed publishers; over-weighting in certain regions; limits on media spend; and so on.
With data gathering and measurement capabilities in place, we’ll launch the campaign, integrating the weighted components and additional instructions and having the algorithms constantly monitor and optimize.
The AI can run tests in minute increments of time and price, constantly working to understand what points toward or away from success.
AI, using machine learning, will be able to consistently hone in on ever more effective buying, timing, and pricing. Placements may be more or less frequent, higher or lower cost on average, and may be designated for mobile, desktop, OTT, or other mixes.
These pieces will all fall into place as the campaign keeps moving toward the increasingly cost-efficient sale of more appliances.
Unlike people, AI will invariably do as instructed. It’s crucial that it is instructed correctly. AI is the opposite of “set it and forget it.”
One key ingredient in all of marketing AI is talent, the people who know how to interpret goals, understand the programmatic advertising landscape, and can create and refine algorithms to the specific needs at hand.