Words by Michaila Byrne
As artificial intelligence continues to prove itself an invaluable tool for pharma, this article explores how AI can streamline marketing budgets for optimal forecasting – showing marketers how to target consumers better, where to target, and at what times.
Few things are as infuriating as having consulted a weather forecast and dressed in accordance with it, only to later return home dripping wet in a pair of Bermuda shorts and a rain-soaked floppy sunhat. Marketing budgets are similarly unpredictable, fickle things, often disproportionately concentrated in areas where there is no demand. Case in point, it makes little sense to market cough and cold products during a July heatwave. More precise forecasting would unquestionably free up time and alleviate unnecessary spending. So, as marketers bid farewell to past tactics akin to stabbing in the dark, a collective sigh of relief can be heaved as we usher in the era of the AI weatherman – here to improve predictive forecasting and help marketers target consumers in a smarter way.
After all, true knowledge isn’t general, it’s specific. In the past, sales data has been an inadequate indicator around which to be centring media campaigns. “My ethos for data is that it has to be action driving. If not, it’s actually just irrelevant. So, if you are just looking at data for data’s sake, then you are actually just looking at superfluous data analysis”, says Oliver Watherston, Head of Business Insight & Analytics, Novartis, speaking at the eyeforpharma Marketing and Customer Innovation Europe event. Alternatively, the key to spending optimisation is through obtaining a comprehensive understanding of the demand in question. Such insights are attained using AI analytics and predictive modelling techniques that generate independent level insights from collected data. So, what exactly does this mean for pharma marketers? This state-of-the-art optimisation supplies marketers with information that allows them to adjust campaign budgets accordingly as they begin the process of customising streamlined marketing strategies for specific regions.
A massive 90% of all data in the world has been collected in the last 2 years alone. With such an unprecedented level of information within reach, it can be tempting for marketers to blast audiences with content across all channels. But restraint is savvy, and in the words of Herbert Simon: ‘A wealth of information creates a poverty of attention.’ Marketers should aim resources towards the strongest demand and at the most appropriate time. AI achieves this goal as highly automated pipeline procedures extract signals and collects anything from social media to weather and sales data, updating the model and delivering forecasts to local teams. Depending on the model, marketers can track what signals are driving sales up and what signals are driving sales down.
As with weather forecasts, predictive tools in marketing are limited in the sense that they can only account for what has come before. Albert Pla Planas, Senior Data Scientist, Sanofi explains: “You can only work with the data that you have. In following the data science and traditional intelligence techniques, you get a forecast that only provides you with information that is similar to what happened before.” Regions won’t necessarily behave homogenously within themselves and predictions are largely dependent on external data with factors like climate change further exacerbating the unpredictability of seasons. To account for regional variations, a given country will generate a unique, personalised forecasting model based on different trends, data, and technologies. There is no reason to expect the North and South of Italy to behave in the same way. In Brazil, climate, pollution, dust, and humidity all contribute towards allergies whereas in Europe, allergies are driven primarily by pollen. The approach would work independently depending on the scenario and can be applied worldwide at more regular intervals.
‘‘When people are sick, their online behaviour reflects this. They’re searching for terms like cough, cold, sneeze... we realised we could use this information to really drill down and give local marketing teams immediate, actionable insights’’, says Darren Frey, Senior Data Scientist, Sanofi. Marketers can reuse this data to boost accuracy – and the proof is in the pudding. Upon implementing this, Sanofi discovered that the correlation between social media signals and demand was very high and observed an 88% accuracy in terms of long-term forecast. Anomalous seasons may not be forecast, but their potential impact can certainly be mitigated since they predict far more likely realities than traditional models. From there, it’s a simple formula: “The approach we’ve taken is that media spend should be proportional to the likely demand. Demand goes up? You should be spending more. Demand goes down? You should be spending less. This applies at the smallest level possible”, says Frey.
The ability to assemble mass data is surely in vain if not channelled and utilised for specificity. What is the point of accumulating data if the data is not actionable? Pharmaceutical marketers have an opportunity to wield AI to its full potential as they make sense of behavioural changes to create streamlined, precise, reliable marketing models rooted in state-of-the-art optimisation. Media spend should be proportional to the expected demand and with this newfound insight, marketers are fully equipped with the knowledge to weather any storm.
