The challenges facing marketers have grown significantly in recent years. Audiences are becoming ever more fragmented, budgets ever more squeezed and, in the face of competitor actions or unforeseen disruptions, such as the Coronavirus, marketing strategies have needed to evolve more quickly and consistently. As a result, brand communications must be tailored to multiple audiences, media investments must be optimized, and campaigns released to market quickly. To help solve the complexity of media plan optimization in this fragmented world, we’re turning to AI.
AI has extended its sphere of influence to many aspects of marketing, but one area where it’s only just realizing its full potential is that of media planning, allowing us to help our clients address these challenges, in two key ways:
The explosion of media options open to consumers, coming from the mobile and digital revolution, means that it is today difficult for clients to reach mass audiences, as it was in previous decades through television or press for example.
This is a phenomenon called “audience fragmentation”, meaning that in order to assemble a large enough audience, brands have to use many different media channels, tactics and messages, as well as non-paid media options like sponsorships, content sharing etc.
Traditionally, when “mass audiences” could be reached easily, this wasn’t a problem for planning systems. They could calculate the effects of a plan for one main audience (e.g. Main shoppers 25-55 years old) and the technique used was simple: an average “response curve” was created for the mass audience, that would indicate how many people could be reached in that group for a given amount of budget.
Today, creating media plans has become challenging, due to the increased number of audience segments that need to be included in a campaign. Some can overlap significantly, meaning that people belonging to multiple audience segments are at risk of being “over exposed” with ads, leading to an unnecessary waste of money for brands.
Using traditional planning tools with average response curves for one audience does not work anymore.
Planning efficiently across multiple audience segments while understanding how they overlap requires planning at a much granular level of data: not at the segment level, but at the consumer level.
We need to be able to simulate the impact of a media plan on individual consumers, not an average audience segment. This means we have to use very different data: much more granular data is needed.
We use panels of consumers at granular level as a weighted representation of a country’s population, where we understand individual people’s consumption of media, as well as they involvement with different product categories and brands.
This also means we need to use different techniques to analyse this data. This is where AI comes in: it allows us to build a simulation model of a media plan for each individual consumer, and then aggregate the effects of the plan up to each audience group.
The technique we use for this is called Agent Based simulations, which requires intensive and complex calculations, so it has started to take off with the emergence of Cloud Computing, although it is still rarely used in the Marketing industry.
This is effectively the only way to solve this audience fragmentation problem: at Wavemaker we did human vs machine parallel tests: a senior team of planners was given the task to create a plan that would optimize reach across three overlapping audiences.
They could not do it, as every time they would improve reach in one of the audience segments, they were losing reach in the others. In contrast, our AI system solved the problem in 1.5 minute.
One of the key outputs we create for clients is detailed weekly media plans, capturing how much investment is put behind each media channel, each week.
It is critical to get this right as many factors influence the effectiveness of media: memory decay (i.e. the speed at which people forget brand messages is different for TV, social media and outdoor posters for example), the actual amount of spend, the pattern of investments (front loaded, continuity, pulsing, bursting…), the frequency of exposure to the message, viewability of ads (e.g. some people block ads on their devices), etc.
It is a tricky process that requires a lot of time and experience from our planners to get right, even with the correct tools.
But again, AI can help here: not to replace media planners, because experience and know-how is a critical craft in media planning. But to help them get there faster by suggesting a good first version of a plan: it will not be the final plan, but it may be 70% or 80% of the way there, saving planners significant time.
This is possible thanks to an Artificial Neutral Network, which can “learn” to think like a real-life media planner. Within seconds it is able to run tens of thousands of media plan combinations and see their impact on output metrics.
Once the neural network is trained, a Genetic Algorithm, which mimics a natural selection process, uses the neural network, and based on it, figures out what the optimal plan is.
There is clearly a role for AI in media planning – it is already transforming the way our teams work and empowering our planners to produce the most effective and efficient media plans for our clients.
I’m certain this is just the beginning.
Stephan Bruneau is the global head of product solutions at Wavemaker Global.
The post How AI is simplifying the complexities of media planning appeared first on ClickZ.Reblogged 1 month ago from www.clickz.com