Compound disruptions are making it tougher to respond to your customers. The shifting regionality of COVID-19 creates unexpected spikes and shortages, and the U.S. economy is officially in a recession. Rising social consciousness sets new parameters for connecting with consumers as new data privacy rules take root.
To drive growth for 2021 and beyond, rely more on adaptive modeling using machine learning, artificial intelligence or other mathematical modeling techniques to solve the many challenges businesses face.
Adaptive modeling approaches adapt by discerning, learning online, and integrating model errors, then adjusting in real-time. Adaptation uses the error’s sequences as they become available. This means, in today’s continuously disruptive world, adaptive modeling can identify and adjust to emerging data patterns across media consumption, purchasing behavior and supply chain demands.
We are not suggesting that you toss aside historical data. Past performance has its place in setting the baseline or starting the model.
You can further extend adaptive modeling by using a modular and flexible framework of a mixture-of-experts regulated by a gating or weighting network. Each expert is a model with a different realization of the system parameters.
The gating network performs online adaptation of the weights given to individual model estimates based on performance. This approach improves estimation accuracy, creates a quicker response to changing environments, and delivers numerical stability and computational demands. A search algorithm in a feedback loop can periodically enhance this mixture-of-experts by creating a new and enhanced set of models to use.
This approach will keep you from reacting to fleeting consumer behaviors with the use of adaptive techniques and smart data processing. One key is to build the appropriate lag time to discern whether the observations are “noise” or truly a pattern.
When the country was in lockdown, for example, Samsung’s Behind the Screens Study showed media consumption of streaming services went through the roof — nearly 60% of all TV time. Out-of-home media (OOH) became moot as everyone stayed put.
But as consumers have begun to navigate the “new normal,” adaptive approaches can help identify and adjust your engagement with those who have begun to pursue activities outside the home, whether shopping, traveling, or going to bars and restaurants. OOH, for example, is making a strong comeback as more customers head to the grocery store.
Catalina buyer intelligence shows shoppers are making fewer shopping trips but spending more. Adaptive modeling is an invaluable tool to help you find and keep new buyers who were introduced to your brand when their favorite was out-of-stock. It will also identify loyal buyers who may have strayed. Both require recalibrating offers through media the buyer is consuming now, not six months ago.
Adaptive modeling can also inform your pricing strategy. A loyal user may have always bought at X price, but if they have lost their job, they now may only be willing to purchase it at a fraction of X. Conversely, purchase conversion rates on some products have improved. For instance, drug stores need not put promotional dollars against Zinc, a supplement that many doctors have prescribed to boost the immune system.
Individual campaigns also can become more precise by using adaptive modeling. Continually confirm the authenticity of your brand messaging with your core customer by communicating that you are respecting their privacy boundaries. Also use it to devote more attention to regional shifts, whether it is a COVID-19 spike near a college campus or a seasonal change.
Do not underestimate the role adaptive modeling can play in your supply chain and inventory projections. It becomes a safeguard against getting stuck with too much inventory as it adjusts to the post-stock up period. Adaptive modeling can also pinpoint potential surges in demand, so you don’t miss out on consumers ready to buy your product in the future. It can even pinpoint when and where that purchase is likely to occur.
In these times of compound disruptions, align adaptation with your company’s strategic objectives. Use it to solve concrete issues with your supply chain or to determine how best to reach a specific consumer (or consumer segments) through which channel, with which creative and message, and which pricing prompt purchase and build loyalty.
Dr. Wes Chaar is chief data and analytics officer at Catalina.