Our desire to plan for the future is inherent – whether from a business or a personal perspective.

The need to make sense of our direction of travel and what obstacles or support we may encounter along the way, is deep-rooted. It is a rare and unusual person or organisation that proceeds without any idea or focus of what might be coming next.

So, it’s not surprising that we’re looking for guidance from the huge quantities of data we can now access. As big data has grown, so too has our ability to analyse it – to process, crunch and dissect the secrets held within the numbers at speeds and scales impossible before the arrival of algorithms, artificial intelligence (AI) and machine learning (ML).

But as we search for future lessons and foresight in numbers from the past, we must all be alert to the realities – and limitations – of what is possible. Avoiding blind trust, recognising that there could be holes in the data is the first requirement.

Look at what we’re currently living through – there are obvious problems with using past data to predict future behaviour when the world is faced with a situation it has simply never experienced before. 

We must also be mindful that algorithms are built by people and people are full of biases – conscious and unconscious – so the tools to analyse this data at speed are not free from human preconceptions and partiality. Artificial intelligence is not bias-free intelligence – the recent A Level algorithm fiasco demonstrated that for all to see. 

The full benefit comes by combining data analysis with human creativity and insight. It is in the space where the data and the humanity can work together, that the real power lies.

In many cases, data and AI can get you 90% of the way there – it is the final 10% that is the most challenging. 

Data can’t easily adapt, but people can. Modelling may have told you that a promotion advert would be most effective if it ran in May, but it will be a person who knows that would have been tone deaf and entirely inappropriate if it had aired during the peak of Covid-19.

Building empathy and human understanding between the business and the data can ensure scenario planning accommodates more flexible, long-term plans – recognising that uncertainty is highly likely, if not inevitable.

Within this context, we can still work toward greater effectiveness and better return on investment (RoI) while accepting that the future is not fixed, and the business trajectory not preordained.  

Marketing in this environment is as complex as it is uncertain, and commercial mix modelling (CMM) can help marketers make the most of their existing budgets and maximise RoI without getting bogged down in the data. Our philosophy is that the best approach involves using human intelligence and experience and enriching that with artificial intelligence.

With CMM we can combine structural equation modelling (SEM), Bayesian statistics and ML to assess the effectiveness and efficiency of sales promotion measures as well as the influence of environmental factors and competitive activities and then simulate the expected effect of planned measures. 

This involves hundreds of variables – all containing data points down to transaction levels. This really is big data in action as the data sets of the individual models are in the GB to TB range. 

Live modelling and agile learning work together to support marketers’ decision-making. With this, businesses can get through the current – and future – difficulties.

It’s not a case of humans vs machines, but rather finding the balance where human intelligence adds the insight to the machine intelligence to make the data work for this real, complicated and unsettled world. 

Analytic Partners were a sponsor of MAD//Fest x DMEXCO@home on 23-24 Sept. MAD//Insight will be releasing the content highlights over the next few weeks.