Predicting the future with predictive modeling.

Especially in companies this kind of 'biases' sometimes occur. Example: a company decides not to deploy a major media campaign this year in order to achieve its cost target. Shareholders happy and back 'on track' with the annual target. However, this has a negative effect on brand awareness and preference among consumers: the brand is less top of mind. Sales decline significantly, resulting in a drop in revenue. The cost savings by not deploying a media campaign this year is nothing compared to the negative impact the drop in sales has.

In six steps, we will look at the right trade-off between short and long term: how do you know if your present bias is preventing you from investing in the long(er) term?

So the question is how to make better decisions that are also smart for the longer term. How can we eliminate "present biases," the reflex to go for short-term advantage, and get more focus on longer-term opportunities? Even in your business, wrong choices may be made because of present biases.

The answer is data and predictive models. These can help us look into the future objectively under certain assumptions, and that's a powerful tool for making choices for the longer term. And to eliminate biases in our behavior. But how do you look into the future with data? It sounds like a utopian idea, but much is really possible through clever application. Read the roadmap below that can help you do this. With a successful application, you may be able to outsmart the competition.... Because in 2022, the majority of companies still cannot look into the future with data.

If all goes well, you already know through research and years of experience how consumers react to your product. Step 1 is to dive into the data and research you already have available. That sounds like an "open door," but many companies sometimes forget that they are already sitting on a mountain of information and experience. Perhaps you launched a product in the past; what happened to the customer database then, did it increase in number? Look at the data and information you have and try to answer the question: did product launches done in the past lead to more customers? And what did it do to costs?

Based on the insights from step 1, you will compile a concrete list of findings. Then you start trying to look ahead and make assumptions about what you expect to observe if you were to introduce the product now, under current conditions. Draw up hypotheses: expectations you have that you want to test. In addition to your internal data and research, include at least the following:

Next, explore whether you can actually answer the hypotheses. Do you have all the data available? If you don't have the data, you can't include it in a model for testing. See if you can still obtain or perhaps need to collect the data. Delete the hypotheses you cannot test with data. It is important that you include all data that could possibly have an impact. After all, you want to take all factors into account.

Put the data you have collected into a predictive model where you want to use data to predict the effect of a new product on the customer base. Look closely at what the explanatory value of the model is from a statistical standpoint so you know whether the model and your assumptions actually cut wood. In other words, can the model make good predictions or are the quality and assumptions flawed? If so, go back to the previous steps.

You are now going to really look into the crystal ball by setting up so-called 'what if' scenarios. 'What if' the current customer base shrinks, can this new product attract new groups or bring back runaway consumers? Or: 'what if' competitor A discontinues its product does my new product then have a greater impact? In this type of 'what if' always applies ceteris paribus (all else remaining equal), you assume that the other factors are constant and, unlike inflation in this case, remain the same. After all, you can't have all scenarios chew up the model in detail.

Set up as many "what if scenarios" as you can that may be realistic. Keep going until you get a feel for the mechanisms: that you understand what happens when you turn different knobs: economic conditions, competitor behavior or pricing etc At some point you will see which situation will be the most plausible in 5 years. This is an iterative process; the point is that through "what if scenarios" you actually try to understand the possible future with the marketing mix in mind.

Translate the outcomes from the above exercise into daily practice and translate it into policy. What should we do tomorrow, what strategic choices should we make with long-term effects included?