AI is transforming customer service, promising faster responses, greater efficiency and more personalized support. But automating an inefficient or poorly designed process won’t necessarily improve the customer experience. In fact, it can amplify existing problems, turning everyday frustrations into larger issues that affect customer satisfaction and trust.
To realize AI’s potential, businesses need to look beyond the technology and consider whether their current support processes are ready for automation. Below, members of Forbes Technology Council identify customer service workflows companies should refine before introducing AI and explain how those improvements can lead to better outcomes for customers and businesses alike.
Consolidate And Clean Up Customer Data
Before automating customer service communications and routing, companies should clean up the customer and account data behind each ticket first. When contract terms, prior issues and account details are scattered across systems, an AI tool will give inconsistent or wrong answers. Consolidating and cleaning that information first keeps automated responses accurate once support work is handed to AI. – Micah Remley, Robin
Improve Customer Interaction Classification
Companies should improve customer interaction classification before automating it. AI can identify intent, sentiment, recurring themes and dissatisfaction risk at scale, but executives must define what happens next. High-risk cases should be flagged with context, routed to the right agent and escalated based on clear thresholds. Overall, automation should strengthen accountability, not remove it. – Dr. Vasudeva Akula, VOZIQ AI
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
Simplify Customer Service Processes
Before adding AI to customer service, we should be asking a much more basic question: “Does this process still make sense?” Removing unnecessary steps, handoffs and approvals in the beginning gives AI a better place to start. The value isn’t simply faster service; it’s making the experience easier for customers. – Vamshi Noothi, Cisco
Make AI-To-Human Handoffs Seamless
Before automating support, fix the escalation workflow. The best AI isn’t the one that handles everything; it’s the one that knows when to stop—defining where it should step aside, making the handover seamless and passing the full context to a person. A customer should never have to repeat themselves. Knowing when not to automate is just as important as knowing when to automate. – Pete Hanlon, Moneypenny
Update The Support Knowledge Base
Fix your knowledge base before automating a single conversation. Most teams bolt a chatbot onto outdated, contradictory content written for staff—then act surprised at confident wrong answers. Audit your top 50 topics: Create one clear, current, customer-facing answer for each. Get content right first, then automate. Otherwise, you’re just scaling confusion. – Mayur Khandelwal, EXL
Build A Comprehensive Self-Service Hub
The core reason for adding AI automation and chatbots into support is to take some of the load off human agents, saving time for your team and your customers. But for that to work, your AI agent first needs a clean, detailed self-serve support hub to pull from, including FAQs, tutorials, videos, interactive guides and checklists. You can open these resources up to customers directly too, so they can pick whichever support method fits how they like to learn. – Osman Koc, UserGuiding
Assemble Complete Customer Context
Organizations need support workflows to assemble context. Pulling together a customer’s history, products, installed assets, warranty status, entitlements, prior resolutions and so on helps AI to effectively resolve cases and deliver the best customer service. – Bruno Billy, APGAR
Prepare Support Content For Customers’ AI Agents
Companies over-fixate on the bot-to-human handoff. More and more, the customer contacting you is an agent chasing a refund or a return window, while most workflows assume a person who can be put on hold. Support content is at the top of the funnel now, too. Help center pages and old threads are what models read when someone asks if you’re worth buying. – Andrew Higgins, Parsnipp
Preserve Context Across Support Tiers
Customer service teams should improve handoffs before automating responses. Most support workflows lose the context when a ticket escalates between tiers. Teams manage by rereading the whole thread. Automating on top of that results in a fast way to zero understanding. Fix the context when a ticket escalates. The AI will inherit whatever context you provide. – Sanjay Gidwani, KOSMOS
Track And Escalate Recurring Issues
Before automating customer support, companies should improve how recurring issues are identified and escalated to product and engineering teams. AI may resolve tickets faster, but it can also hide patterns revealing product defects, usability gaps or weak documentation. Automation should help eliminate root causes, not merely process symptoms faster. – Judit Sharon, OnPage Corporation
Better Understand Customers’ Needs
Companies should focus first on understanding their customers. The thing that differentiates amazing customer service from every other interaction you may have is the value and human traits that customer service agents bring to the table: hearing voice and tone inflections, having intellectual curiosity, and showing empathy. There’s a human emotion side to providing good support. If you don’t fully understand your customer, you can’t add any flavor of that using AI. – Erik Jost, Black Box
Identify The Root Causes Of Repeat Requests
Before automating support, identify why customers contact you repeatedly about the same issue. If thousands of people ask the same question, the problem may be the product, not the support team. AI can make answering cheaper while quietly removing the incentive to fix the root cause. – Benedetto Biondi, Folks Finance
Clarify Ownership Of Stalled Support Requests
Before adding AI, fix the moments when a customer gets stuck. Make the path from “I need help” to “someone owns this” clear and fast. AI can accelerate a good workflow, but it can also make a broken one harder to see. – Serhii Zakharov, PayDo
Define What ‘Resolved’ Really Means
Before automating anything, fix what you’re actually optimizing for. Automation doesn’t chase your goals; it chases your metrics, and containment and closed-ticket numbers are the easiest to game. Point AI at those, and it gets good at the wrong thing: clearing tickets fast while the same problems come back—churn climbing behind a green dashboard. Decide what “resolved” really means before you automate the churn, not the fix. – Corey Ercanbrack, Vasion
Offer A Direct Path To Human Support
I would make it easy for customers to reach a person rather than requiring every issue to go through AI. For shopping questions, AI can guide the customer through more steps if it is making progress. But if a payment is not going through or a hotel reservation for an important event cannot be confirmed, the customer needs a clear path to human support. I would always offer a handoff when AI cannot resolve an issue after two attempts—sooner when the issue is time-sensitive. – Jayashree Arunkumar, Wipro
Distinguish Routine Tickets From Escalation Signals
Before automating, separate the tickets you want to resolve from the ones you need to hear. A refund request can be closed by an agent, but a pattern of the same complaint is a decision the business owes an answer to. Automate the resolution and you can lose the pattern, because no one is left counting what the bot quietly handled and filed away. – Anna Drobakha, Groupe SEB
Redesign The Path From Inquiry To Resolution
Before adding AI, companies should reimagine the path from initial inquiry to resolution—especially escalation and ownership. Map where context gets lost, define when a person steps in, and set a clear resolution goal. Automation creates more value when the underlying support workflow is already consistent. – Ashish Agarwal, OMNIA Partners
Establish A Baseline For Support Costs
Measure what the workflow costs today, before AI touches it. Support is one of the few functions where this is easy: You already have handle time, resolution rate and cost per ticket sitting in your ticketing system. Almost nobody captures that data cleanly first. Automate without that snapshot, and you will spend the next year arguing about whether it worked with no way to settle it. – Prashant Jalan, Guickly
Refine Support-To-Sales Handoffs
One workflow to improve before automating is the handoff from support to sales. AI can spot upsell or renewal opportunities quickly, but timing matters. If a customer is still frustrated, a sales pitch can make things worse. Resolve the issue first, restore trust and only then hand the opportunity to sales. – Kostiantyn Gitko, Devox Software
Ensure Customer Feedback Reaches The Right Teams
Improve how customer feedback gets back into the business. Support conversations reveal where products, policies and experiences are creating friction. Before automating more of those interactions, make sure recurring themes reach the teams that can act on them. AI should help companies learn from customers at scale, not simply close more tickets. – Vibhor Kapoor, AdRoll