Next-Best-Action Is Dead. Long Live Next-Best-Experience.
Next-best-action engines optimize what a brand wants their customer to do. Next-best-experience repositions decisioning for what the customer needs from brands.
Stat of the Week. 20%–30% lower costs through AI-powered next-best-experience, 15%–20% higher customer satisfaction, and a 5%–8% revenue lift. (McKinsey, October 2025.)
In the last ten years, top marketing engines have focused on questions customers never actually asked. Next-best-action (NBA) decisioning looks at every possible move a brand could make, such as offers, upsells, or win-back emails, and sends the one most likely to convert, always with the brand’s goals in mind. But what customers really want to know is, what do I need right now? McKinsey’s answer to this is next-best-experience (NBX), which shifts decision-making to focus on the customer’s current needs. Their client work shows this change can lead to a 15%–20% increase in satisfaction, a 5%–8% boost in revenue, and a 20%–30% drop in cost to serve (McKinsey, October 2025).
There’s another reason to make this change now. Accenture’s research shows that AI agents are starting to interact with brands on behalf of customers, acting as go-betweens that don’t care about your brand. When software is involved, flashy campaigns matter less than the actual experience, which is what NBX aims to improve. The best approach is to shift away from next-best-action as your main method and rebuild around the customer’s key moments, starting with the one you currently handle the least well.
The engine answers the wrong question
NBA became popular because it measures the chance of conversion across thousands of possible interactions, and it beats batch-and-blast methods by every marketing metric. But it misses what happens outside those numbers. NBA engines focus on actions that help the brand, so moments where no offer converts, like a third support call about the same billing error, a delivery that never arrived, or a renewal that quietly failed, are ignored. Yet, these are often the moments that matter most to customers.
A telecom company ran a top NBA program. Its outbound next-best-offer engine reached millions of subscribers each month with personalized upgrade offers, and its conversion rates led the industry. But during that time, customer exit interviews kept highlighting one problem no one was fixing: long hold times after repeat billing complaints. The engine didn’t respond to this because there was no offer to make, so valuable customers waited on hold while the system calculated discounts for people who hadn’t even asked for help. When the team finally set up a trigger for this customer moment—a proactive callback for anyone who contacted support twice in a week about the same charge—it outperformed every other offer in keeping customers. The best result came from an action that wasn’t an offer at all.
Decisioning around the customer’s moment
NBX keeps the same decisioning tools: the data, the models, and the orchestration, but changes the main question from “what do we want this customer to do?” to “what does this customer need most right now?” McKinsey gives the example of a global payments processor whose model predicts which merchants might reduce business in the next week, so the right help, like a technical fix, fee forgiveness, or a better product, arrives while the merchant can still be saved (McKinsey, October 2025). The way success is measured also changes. An NBX program tracks satisfaction and cost to serve, along with revenue, because the value of meeting a need often shows up as the call that never happened.
Deloitte’s research on service organizations finds the same thing: AI-first service companies raise satisfaction while lowering costs to serve (Deloitte Digital, 2026). Both firms’ work shows a clear pattern: the best results come where the stakes are highest for the customer, not just where the conversion model predicts the biggest return.
The agent in between
Accenture’s Consumer Pulse 2026, a survey of 25,590 consumers in 16 countries, shows there’s little room left for experiences that only look good in a campaign report. 74% of consumers say they would trust a personal AI agent more than their best friend to make purchases for them, and 37% of loyal shoppers would let an agent switch brands if it finds a better fit (Accenture, 2026). An agent doesn’t read your campaign; it looks at your prices, your availability, your service record, and whether the last three deliveries arrived on time. As AI-assisted shopping grows, the focus shifts from winning the customer’s attention to earning the agent’s recommendation. The agent’s scorecard is the real experience, moment by moment, which NBX manages and NBA does not.
How to get started
The good news is you can start implementing an NBX strategy on a small scale. The first pilot only needs to focus on one moment, not a whole new system.
What is one moment your customers have to handle alone that they shouldn’t? Start by looking at 90 days of repeat contacts and complaint themes, and identify the single most important moment your current decisioning ignores. That moment should be your first NBX.
Can you pilot this in 30 days? Start by setting up just that moment: create a trigger, take one proactive action, and have a holdout group. Measure satisfaction and cost to serve, not clicks. If the telecom’s callback story sounds familiar, the results from this first step can help fund the next one.
The brands that succeed will be those whose experience stands up to AI agents. Rebuilding decision-making around the customer’s moment is how you build that experience, one step at a time.
What is one use case you would pilot in 30 days? Reply and share the moment.
Sources
McKinsey, “Next Best Experience: How AI Can Power Every Customer Interaction,” October 2025 (ranges are McKinsey estimates from client work, labeled in-article; payments-processor example from the same article).
Accenture, “Consumer Pulse 2026 (Talk to My AI Agent),” 2026 (n=25,590 across 16 countries; realized survey figures).
Deloitte Digital, “The Future of Service 2026,” 2026 (n=600 organizations + 3,000 consumers; cited directionally in-article — no specific percentages, per dossier verify flag).