Steve Lowell, Global Direct of Data and Insight
Steve Lowell, Global Direct of Data and Insight 31 July 2020

Is predictive analytics modelling still predictive?

Will consumer lockdown habits become long-term habits? Exploring the future of predictive modelling, post-COVID…

Steve Lowell, Global Direct of Data and Insight
Steve Lowell, Global Direct of Data and Insight 31 July 2020
Share article

Will consumer lockdown habits become long-term habits? Exploring the future of predictive modelling, post-COVID…

It never ceases to amaze me just how predictable we are as consumers. Time and again, campaigns using predictive modelling demonstrate a better performance than those where the targeting is less robust.

But what happens when something seismic happens in the world (such as a global pandemic) that alters consumer behaviour overnight? Can you still count on the power of predictive modelling? 

Predictive modelling – how it works

The basic premise of predictive modelling is that past behaviour is a solid predictor of future behaviour. Samples are built from the way consumers have engaged with a brand in the past and are used to build statistical models which gauge the likelihood of consumers behaving a certain way in the future.

For example – responders to past marketing campaigns are used to build models which predict the response to future campaigns.

Post-COVID predictive modelling

But does this premise still hold true as we start to come out of the COVID lockdown period? It’s fair to say that we’ve all made certain adjustments to the way we engage with different brands and industry sectors during this time – from adopting online grocery shopping to signing-up to streaming services to while away the hours.

With these changed behaviours in mind, what does this mean for the future of predictive modelling and can the way consumers have behaved during lockdown be used to predict the way they’ll behave in the new normal?

From my perspective, there’s no doubt that predictive modelling is here to stay as a marketing discipline. But we may have to think a bit differently about how we build these models in light of the lockdown period and the changes in behaviour we’ve seen.

Here are a few points to think about…

1. Lockdown consumer or loyal consumer?

The first consideration is to understand the degree to which your consumers have been behaving differently during lockdown – and establishing whether this was a blip or if these behaviours are set to stay.

Does the profile of a ‘lockdown consumer’ look similar to your typical consumer – not just in terms of the way they interact with you, but also in terms of their profile?

We’re working with an online food retailer who have seen an increased volume during lockdown but a very different profile of consumer. This starts to pose the question as to whether the content of their welcome journey is still fit for purpose when communicating to this new audience.

In addition to the traditional analytics you might use to understand your audience profile, you should consider overlaying human insights too. For example, even if your audience might be the same age or income profile, they may well have a very different attitude towards you and your brand.

2. Set up trigger campaigns

Now might be the time to consider different means of building campaigns. Outbound, direct campaigns are typically built either from some form of predictive modelling or trigger-based – where you react to the way in which a consumer engages with your brand (for example by following up directly after a web visit).

If you have the infrastructure in place, perhaps consider how you can deliver more trigger activity to maintain the performance of your direct comms. This doesn’t require any form of prediction, it’s simply responding to how a consumer is engaging with you.

3. Test and learn

Consider the way in which you build your testing programme to gain maximum learnings of how consumers engage with your comms in the future.

Just because your audience might be different, doesn’t necessarily mean that they’ll engage differently with your communications. Having a robust test and learn platform in place to gather insights from your activity will be crucial during the next few months.

4. Sit tight

Finally, hold your nerve and avoid making any significant overhauls to your campaigns. No-one knows just how different the new normal will be. We may well revert back to pre-lockdown behaviour as the months pass, in which case your existing targeting may well still be valid.

Avoid making any significant changes at this stage. Diligence and a clear understanding of customer behaviour is what’s needed at present.

Share article

How to use social media to boost in-store sales

Indicia 21 October 2020

Let's get analytical

We're sure of it: we can solve your analytics challenge to create new value in your business. Get in touch and we'll get started.

Want more?

Why a route to D2C sales must be the top priority for brands

Author: Indicia Worldwide team 09 October 2020

Our October 2020 webinar laid bare the challenge brands face in rolling out a direct-to-consumer sales channel – and the importance of getting it right for 2021 and beyond…

5 great ways to bring mobile experiences into retail

Author: Indicia Worldwide team 26 August 2020

Can technology revive retail post-COVID? Here are 5 ways retailers can embrace mobile to transform the shopping experience…

Visit blog

Work with us?

Get in touch to see how we can create new value for you