Global – October 12, 2026 -- ChurnZero CEO You Mon Tsang predicts the average customer success manager (CSM) will gain 25-50% more bandwidth per day by the end of 2026 through AI, provided teams avoid the most common adoption error: automating the whole operation at once.
ChurnZero VP of Customer Success Lukas Alexander laid out the argument on the “Customer Success Talks” podcast. He cited two obstacles. The first is AI overreach, where handing too much decision-making to AI could damage long-term customer relationships. The second is balancing automation with a personal touch, because CS teams depend on lasting partnerships rather than the transactional relationships typical of sales and marketing.
Alexander flags three recurring implementation mistakes
Teams treat AI as an all-or-nothing project. Alexander compares this to giving a new CSM the entire customer portfolio on day one. Teams also overlook simple administrative work such as building PowerPoint decks, writing similar emails and preparing for meetings. And many wait for perfect data, even though call records, emails and support tickets already provide enough material for analysis.
Teams should start with one painful administrative task
Alexander recommends automating a single time-consuming task first, such as meeting presentations, account research or summary emails, before moving to complex workflows. He advises writing one AI prompt this week and building a library of repeatable prompts. No new automations should be added until existing ones are proven and used habitually.
Signal-based AI agents read customer data to flag churn and expansion cues
Agents can scan emails, meeting transcripts and support tickets to surface churn risk and buyer intent. In one example, an agent catches a customer saying the product helped grow its revenue by 10% and routes the comment to the CSM as proof for a renewal or expansion.
AI use should also match the service model. Customer success plans may suit customers who value them, while an SMB segment that ignores such plans needs a different approach. CSMs keep relationship-building, which Alexander says AI will never fully replicate.
Alexander advises staff at non-investing companies to propose an AI council
He expects most companies to be asked within three to five years to deliver more revenue with the same or a smaller workforce and more generalist roles. For employees whose companies are not investing in AI, he recommends three steps: assess whether the organization still fits, identify objections such as data privacy concerns, and propose a cross-functional AI council that includes non-leadership staff and meets for as little as one hour a week.