LivePerson

LivePerson is the longest-running cautionary tale in conversational AI. Founded by Robert LoCascio in 1995 as one of the earliest web-chat companies, LivePerson reinvented itself in the 2010s as a conversational commerce platform helping brands like T-Mobile, HSBC, Vodafone and Delta handle customer messaging across SMS, WhatsApp, Apple Messages for Business and proprietary chat. The peak years 2020-2021 took the stock from $20 to over $77 on the thesis that AI-enhanced enterprise messaging would replace traditional voice contact centers.

Then ChatGPT happened. From November 2022 onward, every CIO who had budgeted enterprise messaging suddenly asked whether they really needed LivePerson on top of a generic LLM API. Customer churn accelerated through 2023-2024, the company missed multiple guides, founder LoCascio departed the CEO seat, and a series of restructurings shrank the company from 1,800 to about 600 employees.

As of May 2026 the stock trades around $2.22 with a market cap of $27 million — versus over $5 billion at the 2021 peak. Trailing revenue is $236 million, down 12% year-over-year, with continued operating losses but a 71.7% gross margin that still hints at a viable software business underneath. The bet on LPSN today is whether the existing enterprise customer base can stabilize at a smaller scale or whether the conversational-AI commoditization continues to grind toward zero.

Smart-money interest in LPSN has collapsed in line with the share price. Most enterprise-software-focused growth managers (Whale Rock, Lone Pine, Tiger Global derivatives) exited during 2023-2024 when the post-ChatGPT customer-churn data made the secular-decline thesis hard to argue against.

What is left: a meaningful contingent of distressed-credit and special-situations funds who have looked at the convertible-debt overhang and concluded that equity is a residual claim worth a single-digit-percent of recovery scenarios. The 9.2% short interest is high in absolute terms but actually understates conviction — borrow is cheap and many shorts have been forced to cover repeatedly on small bounces driven by activist rumors.

Insider activity has been mixed. Newer management has bought small open-market lots, but no transaction has been large enough to signal genuine high-conviction insider commitment. The 13F season shows index funds and a handful of microcap value boutiques as the only non-trivial holders.

Conventional multiples fail for LPSN: a negative P/E, an EV/EBITDA distorted by debt structure, and a P/S of 0.11 that is either a distress marker or a deep-value flag depending on assumptions about revenue stabilization. The framework that actually matters: enterprise-value-to-recovery analysis across the convertible-debt waterfall. If the equity ultimately recovers $50-150 million in any restructuring (versus the current $27 million market cap), there is meaningful upside. If recovery is zero — the most probable single outcome — the stock goes to zero. The distribution is genuinely bimodal. Wall Street targets sit between $3 and $5, but only two analysts cover the stock and neither has a strong-buy rating; the recommendation field reads none. The honest valuation answer is that LPSN trades at a price consistent with restructuring probability of approximately 60%.

1. Q3 2026: Net revenue retention and customer-count disclosures — first chance to see whether the stabilization narrative has any empirical support
2. Late 2026: First major convertible-debt maturity refinancing window — outcome will determine whether equity holders survive or face significant dilution
3. Q1 2027: Annual cost-base reset disclosures — visibility on whether the cost structure has actually moved to match the smaller revenue base

LivePerson is a textbook deep-value distressed-equity setup where the dominant outcome is zero and the asymmetric outcome is a 5-10x bounce on successful restructuring. The 71.7% gross margin keeps it honest — this is not a fraud or a melting ice cube without optionality, it is a real software business in the wrong cost configuration with the wrong capital structure. I would only consider LPSN as a maximum 0.25% portfolio allocation, and only with the explicit understanding that the most likely outcome is loss of the entire allocation. If you cannot afford to lose this position completely, do not own it. The interesting question is not whether LPSN is cheap — it obviously is on cyclical metrics — but whether the AI commoditization headwind ever stops eroding the underlying franchise.