Predicting the future requires understanding the recent past.

The world changes in an instant. Just a few months ago, political experts confidently explained why Donald Trump could never win the Republican nomination; now he is president.

Many marketers are equally challenged at making predictions. They approach data analysis and predictive metrics under the same general assumption that the world will remain fairly static.

Brands put past campaigns under the microscope, assess what did and didn’t work during an activation. Essentially, they look in the rearview mirror and apply that knowledge to the road ahead. It’s a disaster waiting to happen for business planning just as it is on the open road.

We live in a world where anything from the economy to a tweet can shift our priorities and behaviors. Data that was valuable months ago are far less significant today and essentially useless tomorrow. Hidden within the firehose of data are the metrics that matter: leading indicators that are most predictive of future business outcomes.

These insights are truly crystal balls providing a clearer picture of how various marketing initiatives will perform against a brand’s business objectives. More importantly, these metrics can be re-aligned as the world changes, providing early warnings if a planned media activation will no longer achieve its intended goal. Ultimately, they can be used to optimize marketing expenditure and mix to maximize the likelihood of success.

But how do marketers pan for gold in a never-ending stream of data? How can they know what matters now let alone next quarter? Typically, marketers have made the mistake of turning to market mix models, using advanced statistical techniques to relate movements in key explanatory variables like price or advertising to a business outcome. This technique is great for establishing an ROI for each media channel and can be of use for media planning and phasing. However, it can be slow and cumbersome to build and implement and, most importantly, generates results by looking at historical data, often the past three years—a pivotal mistake if you want to foresee future trends.

Key Lead Indicators

While marketing mix modeling will continue to be an important tool, marketers should supplement this work by identifying key lead indicators through advanced analytics. Key lead indicators are the set of recent factors most strongly correlated with future business performance.

The technique has been successfully applied to highly seasonal businesses, discrete sales events and new product launches. Take, for example, retailers who have periods of peak demand: Mother’s Day, back-to-school season, Halloween, Christmas, etc. Marketers might find that 20 weeks out from the event, the lead indicators that are best correlated with success are fairly general, such as the general state of the economy, consumer purchasing power within the target segment, confidence and usually unaided awareness.

As we get closer to the launch date, from four weeks out, things start to get a lot more specific. Unaided awareness will still feature, but we’re also looking at data from Pinterest, Twitter, Facebook, time spent on website, depth of website visit and so on. Things are now more specific as consumers have already decided they will spend; now they are deciding where.

Using retail as an example, Pinterest provides an example of a strong leading indicator. The number of pins begins to increase steadily long in advance of the event as users are posting pictures, discussing within their social network and generally becoming engaged with the event.

Course Correction

There are three important things to note with the leading indicators approach.

1. It’s not a forecast: Rather, it is a course correction framework that lets us know the probability of hitting key business targets and, critically, what to do if the probability is low.

2. It shows correlation, not causation: Rather than being direct cause and effect, the lead indicators should be thought of as indicators of customer engagement, which itself is indicative of future success.

3. There’s no one set of indicators: Lead Indicators vary by vertical, brand and across time.

Predicting the future requires understanding the recent past. What happened a year or three ago will not only fail to accurately predict the future but may lead you down the wrong path. Marketers need to look at the leading indicators around them every day to find their crystal balls, rather than trying to drive their business full force ahead with their eyes firmly planted on the rearview mirror.

Read the original article here.