For two decades, software’s answer to slipping customer retention was to hire more customer success managers (CSMs). Hiring numbers bear this out: CSM headcount grew by more than 50% over the past six years. Yet median net revenue retention (NRR) among public software companies slid from about 115% in 2023 to about 107% today (see Figure 1). The problem wasn’t the investment; it was the model.

Two years after we argued that customer success is failing and less than a year after we found that 70% of teams were stuck in AI pilots, the tide still hasn’t turned. In fact, the picture is actually worse. Much of that decline is a result of forces outside customer success’s control. SaaS markets are maturing, AI is putting pressure on budgets, and business models are shifting from one-time onboarding to constant consumption driving. Against this backdrop, how to demonstrate the ROI of customer success has become a frequent topic in our regular customer success forums, as chief customer officers work to show their impact on commercial outcomes.

The core mission of customer success remains as relevant today as it was 20 years ago when the function was born: Proactively help customers realize value from their software investments at every stage, from initial sale and onboarding through adoption, usage, and renewal. The premise is simple: Satisfied customers are more likely to renew, expand their relationships, and refer their vendors to others.

The mission hasn't changed, but delivering on it will require a fundamentally different model.

Figure 1

Today’s model has four fundamental flaws:

• Customers want deeper product and technical expertise but complain that vendors “ship their org chart” to them, citing uncoordinated engagement across touchpoints while CSMs get buried in product issues and administrator roles.
• CSMs were asked to be the one owner of the account and, in practice, to do everything for it across the customer journey. This creates an impossible ask: Own the relationship while also delivering specialized expertise across every customer need.
• Customer success leaders are constantly asked to justify their investment and have found it difficult to quantify its return. Companies are not meeting retention and expansion goals, and customers remain dissatisfied with their experience.
• AI is changing how much customers want to buy. Software buyers want to do more with fewer seats. In a recent Bain survey, 65% indicated a propensity to build their own solutions with AI, and AI-native offerings are emerging as strong alternatives to traditional SaaS across many categories.

Meanwhile, the job is getting bigger. As SaaS matures, growth now comes from customer expansion, and AI products priced on consumption pay off only if customers use them. The outcome customer success answers for has shifted from satisfaction and renewal to adoption, usage, and measurable results.

Designing the new model

Post-sale has never been more critical, but the traditional model needs to evolve significantly.

Through our work with software companies at the forefront of customer success, we see four design principles that are defining the path forward (see Figure 2).

• One owner for every customer: Each account needs a single executive-facing owner, a “success architect” (working title), who can lead C-suite conversations, build consensus, and clear the path to renewal and expansion. The skills, experience, and pay look closer to a strong account executive’s than to a traditional CSM’s.
• Accountability for outcomes, not activity: Incentives should reward leading indicators (adoption, usage, advocacy) with outcomes (consumption), with real downside when they fall. Owners should be measured on what they can move, not on how many calls they log.
• Expertise on demand: “Even our best CSMs cannot be effective in every situation,” one client told us. The account owner’s job is to bring the right expert in at the right moment: pooled product specialists, presales, professional services, or forward deployed engineers (FDEs). The best CSMs already do this through personal networks, but it needs to be standardized, system-wide, and on demand.
• Agents for every account: Organizations that apply AI across workflows see at least a 15% to 20% productivity uplift, yet our latest research shows that around 70% of customer success leaders aren’t yet using AI in a meaningful way. With agents now able to handle outreach, renewal administration, support triage, and quarterly business review preparation for every customer, not just the long tail, this needs to change. Coverage no longer depends on account size.

Figure 2

Smaller customers or contracts will require broader success architect coverage and more usage of digital and agent coverage.

Making it work

Moving to this model is a frontline transformation. Start by designing coverage from the customer’s point of view rather than the org chart, deciding where the owner, a specialist, or an agent engages at each critical moment. Enlist the broader organization (product, engineering, sales, services, support, and partners) to a customer-centered and well-orchestrated future model. Define what a success architect must be able to do, and assess the current team against it. Make the call for expertise a system, with clear triggers and an orchestration layer, rather than a favor. Then pilot a few end-to-end motions and manage the team the way strong sales teams are managed, with coaching, observation, and a visible cadence of leading indicators.

Done well, the model fits within the existing cost envelope: fewer but more experienced owners with broader spans, a pool of shared specialists, greater use of FDEs and hybrid service models, and agents doing the work that buries CSMs today. The return is easy to sanity-check. For a company with $500 million in annual recurring revenue, 1 point of NRR is worth about $25 million in revenue over five years, roughly a full year of customer success spending for a company that size. And that return can be measured with confidence that the leading factors will translate to the desired outcomes.

Customer success doesn’t need more people; it needs to rethink how resources are deployed. The companies that rebuild the function around ownership, expertise, and agents will find that post-sale is where the next decade of software growth is won.