How Next-Best-Action CX helps – before, during, & after a crisis


Share on LinkedIn

Using a next-best-action approach, driven by AI, has proven to help companies elevate CX by delivering relevant and timely in-the-moment individualized recommendations during everyday buying and service journeys.   But in just weeks, Covid19 transmogrified our world, rattling lives, markets, and businesses.  Customers’ journeys became anything but every day.

BC (before Covid-19), AI-powered digital interactions on websites, mobile apps, chatbots, virtual agents, and voice assistants were already on the rise.  Two years ago, Accenture surveyed 25,000 consumers in 33 countries and found over half no longer cared if they were interacting with humans or AI-enabled technologies[i].  Once Covid-19 took hold, those that cared for human interactions had a lot less choice, and digital channel use went off the charts.  Seismic shifts fused home and work life, removing any doubt about our reliance on digital channels and online experiences.  And in countless cases, we quickly learned which brands were best prepared, agile, online, scalable – and sadly, which weren’t.

As with any black swan event, underlying dysfunctions quickly surfaced, with many experiencing poor digital performance, services sometimes completely unavailable, tone-deaf messages, and excessive wait times.  But digital superheroes arose as well, showcasing their agility and empathy – instantly pivoting to leverage their core competencies in helping battle the crisis, and by providing reassuring messages, practical services, and resources to efficiently address immediate needs.  For example, some telecommunication companies amped up network services to essential businesses and to consumers forced to work at home; certain automakers turned factories into medical equipment plants; perfume factories pivoted to making sanitizer; some pharmacies offered free delivery to at-risk individuals – to name a few.

But although many rose to the challenge, most couldn’t do so with surgical precision AND lightning-fast speed.  Even with the right intentions, they still found it nearly impossible to supply situation-sensitive tailored services while simultaneously balancing business realities and the need to respond swiftly.  So, in the interest of speed, they were forced to use a few sizes fit most approach.

No doubt, we will learn countless lessons from these events.  It’s been a stark reminder of the value of:

  • Organization-wide planning and coordination
  • Preparedness simulations and scenario planning
  • Being data-driven
  • Having a speedy and coordinated response system at the ready
  • Listening in the heat of the moment, and not being impervious to context
  • Making targeted recommendations with an empathetic and personalized touch

And that is precisely why having a next-best-action engine in place is crucial before, during, and after a crisis.

Why a Next-Best-Action approach? 

When a crisis occurs, what customers want is immediate and practical help for their situation and a reasonable path forward that makes sense.   In other words, what you do NEXT in every interaction matters.  They need fast solutions to the current situation they’re adjusting to.  But a few sizes never fit all.  For example, some are struggling financially; some aren’t; some are caring for others; some need care; some face upheaval as work fuses with home life.   

For banks that means educating some customers on loan and deferred payment options, but others on investment advice.  For insurance companies, it means providing rebates and suggesting adjustments when current policy premiums don’t match coverage needs, and for others new coverage recommendations.  For communication service providers, it means messaging those who need better services, and reassurance to others that their high bandwidth service won’t degrade.  The list goes on. 

BEST means selecting from a variety of potential actions and picking the one(s) most appropriate for the moment at hand.  We’ll explore a little later how this optimization and ranking works.

Finally, ACTION has two meanings – noun and verb.   As a noun, it means selecting from a variety of action types – from marketing messages to sales offers to service recommendations.  As a verb, it means acting in real-time, conversationally.

To be empathetic just remember the customer is always right, right?

Nope.  Recall I said what customers want is “a reasonable path forward.”   Well, that’s what reasonable customers want.  And let’s face it, all customers are not always reasonable. 

But if a brand listens, is fair, is consistent, and presents equitable solutions to current needs, then the majority of its customers will be satisfied.  That’s the balance you want to strike and the reputation you should strive to uphold.

There is a bevy of useful messages, recommendations, and actions you should be taking to help customers.   Yet unless you are among the select few that rehearsed for this crisis, you’re facing a huge challenge crafting these, getting them approved, and then activating them in various channels.  And through this crisis, you also have to constantly update and adjust them.

What you need is a system that assists you in developing, approving, and activating these new messages.  And if you want to do that efficiently, you need them in a single location.  That way, anytime there are inbound interactions, the channel requesting them on behalf of the customer can simply make next-best-action requests – and not have to load anything or change its logic.  For outbound communications, the system senses customers’ unique needs, and proactively takes action when it makes sense.  An engine that supports a next-best-action methodology does all this. 

Wait, what’s a Next-Best-Action engine?

Simply put, a next-best-action engine is:

A multi-channel, centralized decision authority (a hub) that dispenses prioritized decisions connected to but abstracted from the channels.  The hub leverages logic that combines business rules with machine learning to determine the optimal recommendations during any interaction.

Essentially, it’s the brains in finding the BEST action(s) at that moment (of interaction) and delivering it at breakneck speed (in under 200 milliseconds).

That’s what it is.  But how does it work?  What’s under the bonnet?  Figure 1 illustrates the major components:

Figure 1 – Components necessary in a Next-Best-Action Engine

Moving from bottom to top, the engine must include:

  • Guiding operating rules & principles – A module that holds the codified set of organization rules and governance guidelines, such as contact policies, annoyance thresholds, and product applicability and compatibility rules.  
  • Historical customer data – Often referred to as interaction history, this is the storehouse of all customer communications and responses.
  • Contextual data – This is the latest behavior, emotional, situational, and environmental data available associated with the customer and the interaction.   It can be vital in determining the customer’s immediate needs and intent.
  • An actions repository – This is the library of potential (candidate) actions that are evaluated and arbitrated.
  • Real-time customer analytics – This is the machine learning layer.  Here, each action is assigned a propensity and value, each interaction a journey and stage, and each customer a set of scores, such as likelihood to churn.
  • Next-best-action arbitration – This tier performs the final next-best-action ranking.   P = Propensity and V = Value.  A simple formula (P x V) allows for all candidate offers that survive other rules like eligibility, to get subjected to a final stack ranking in terms of overall priority.
  • Action presentation layer – This is the channel interface layer.   For actions to be useful they must be presented to the right channel, and this is where that integration happens.

Notice the variety of technologies: customer data processing and analytics, a rules engine, and an arbitration layer.   All of these components must work closely together, and as such a next-best-action engine built to do this differs from a stack of pure-play technologies.  If you attempted to build this from scratch you’d have to wire together a rules engine, ETL tools, streaming data tools, a CDP (Customer Data Platform), an offer management system, modeling tools, and other business intelligence services.  That’s a lot of moving parts to integrate.

So what you get when you implement a next-best-action engine is all that in one cohesive system, purpose-built to check rules, crunch critical data, filter down to candidate offers, apply AI, arbitrate remaining actions, and then deliver the winners to the channel of choice – again all in under 200 milliseconds millions of times a day.

But why invest in all this?  Won’t things just go back to normal?

No, they won’t.  There will be a new normal.

Wind the clock way back to 2019 (feels like years ago doesn’t it).  As we learned in Accenture’s study, businesses (as well as their consumers) were already reliant on marketing, customer service, and automation technologies to streamline buying journeys, reduce costs, and improve service.  And some businesses were already using decision hubs to provide 24 / 7 next-best-actions, resulting in faster resolution times and better customer experience.

In the months ahead, digital experience bars will go even higher.  More customers, shopping and looking for service online, will be even less tolerant of long wait times, poor service, and irrelevant recommendations.  In short, this crisis just put customers’ digital expectations on steroids.

Yet with any crisis comes conundrums and hard trade-offs.  Consumers expect brands to provide flawless digital services and CX.  Paradoxically, as companies watch labor, capital, and other resources dwindle, they’re realizing they must invest in digital transformation and CX initiatives.   And they also know any digital investments made now must improve (or at least protect) revenue, nearly instantaneously, while simultaneously increasing agility and cost-efficiency.  There will be no tolerance for stalled or poorly performing projects.  They must show results and show them fast.  They must orchestrate flawlessly through a multitude of digital channels, including chatbots, human-assisted portals, and mobile apps. 

Tall orders indeed.  But orders that must be filled for survival now, and thriving later.

Many have waxed on about being agile, nimble, flexible, and fast-trackers.  Admirable aspirations indeed, as perfection is the enemy of good enough in a crisis when quick actions save lives.  As Barkley put it, in their recent POV – Open brief for being a useful brand during Covid-19, “Your brand needs speed to usefulness.”[ii]  Agreed, but I’d add you always need this.  With today’s technology and systems, you shouldn’t have to face a trade-off between fast-acting and being highly personalized and germane. 


Drastic times demand drastic measures.  Events such as what we are experiencing now, force us to take risks, innovate in ways never imagined and operate with real urgency.   The landscape has shifted in ways (and at speeds) we never imagined and some ideas, previously ignored or unnoticed, are surfacing as viable and superior long-term solutions.

The Next-Best-Action approach to customer engagement falls into this category.   When coupled with the right engine, its ability to deliver in-the-moment relevance to individuals, instead of the one size fits a segment method, worked before Covid-19, is working now, and will become the de facto standard for delivering great CX going forward.   A silver lining indeed for those interested in true customer-centricity.

[i] Accenture,, 2017

[ii] Barkley,, 2020


Please enter your comment!
Please enter your name here